Professional Documents
Culture Documents
UNIVERSITY
UBRARIES
Mathematic$
Library
White
Hall
DATE DUE
The
original of this
book
is in
restrictions in
text.
http://www.archive.org/details/cu31924059156301
Production Note
Cornell University Library produced this volume to replace the
irreparably deteriorated original.
was scanned using Xerox softIt
ware and equipment at 600 dots
comper inch
resolution and
pressed prior to storage using
The
CCITT Group 4 compression.
digital data were used to create
Cornell's replacement volume on
paper that meets the ANSI StandThe production
ard Z39. 48-1984.
of this volume was supported in
part by the Commission on Preservation and Access and the Xerox
Corporation. 1991.
SAGE
ENDOWMENT FUND
THE GIFT OF
Henrg W. Sage
1891
Honbon:
C. J.
lEipjig:
J^ito
gotk:
263,
AEGYLE STEEET.
F. A.
BEOCKHAUS.
THEOEEM
ABEL'S
AND THE
ALLIED THEOBY
INCLUDING THE THEORY OF THE
THETA FUNCTIONS
BY
H.
F.
^AKER,
M.A.
FELLOW
CAMBRIDGE:
AT THE UNIVERSITY
1897
[All Rights TMerved]
PRESS.
dDambdDge
PRINTED BY
J.
AND
C.
F.
CLAY,
PREFACE.
It
may perhaps
analytical developments of
treats.
This book
to
most
artistic
who
use
It is also
for
it
made
first
to be the
ways of formally developing the theory regarded as complete.
The matter
general,
such study.
those
And an endeavour
is
for
has been
is
to point out
illustrative,
portion of the continuous argument of the book, are introduced only so far
as they appeared necessary to explain the general results, mainly in the
scope of ideas to which this has led will be clear from an examination of the
table of Contents.
The methods
of
Riemanh, as
far as
but
all
may
or on the
it
PREFACE.
vi
Dedekind and Weber, GreUe's Journal, xci., xcii. (1882). Aud it is hoped
that what is necessary for the development of the theory from the elementary geometrical point of view will be understood from Chapter VI., in
general formulae given near the end of Chapter XIV., will provide sufficient
indications of how the theta functions can be algebraically defined; the
reader
may
and
XXXVI.
be hoped
is
much
is
to
To the reader who does not desire to follow the development of this
volume consecutively through, the following course may perhaps be suggested
Chapters
I.,
II.,
XV.
(in part)
it
is
IX., X.,
also possible to
begin with the analytical theory of theta functions, reading in order Chapters
The
footnotes
contain
to
the
first
and parts of
its
plan
is
somewhat
different from
it.
it will
provide practice in the ideas involved in the general theory; to suggest the
steps of alternative developments without interrupting the line of reasoning
in the text
at
all, till
For
my
and
much
my first
to place
if
me
acknowledge
VU
PREFACE.
recommended
Biemann
surfaces.
In the final revision of the sheets in
through the press, I have received help from several friends.
Mr A. E. H. Love, Fellow and Lecturer of St John's College, has read
the proofs of the volume
in the removal of obscurities of expression
and in the correction of press, his untiring assistance has been of great
value to me.
Mr J. Harkness, Professor of Mathematics at Bryn Mawr
dealing with
their passage
undetected by
Mr Love
XV. onwards
many
volume.
by the
more
will
be found in
it.
it will
by the
is
equalled
fascination.
1)2
CONTENTS.
CHAPTER
The subject of
I.
investigation.
PAGES
2,
4,
The
1,
the surface
is
deficiency
7,
9,
10
II
is
3p - 3
...
if
of zero
CONTENTS.
PAGES
29
28,
30
33
31
34_36
37
The {p-\)p{p +
\)
34,
35
36,
37
CHAPTER
38
41
44
40
44
46
IV.
39
Definition of dimension
40
41
42
Fundamental
43
44
48
is
p + n-\
set for the expression of integral functions
infinities are
ordinary places
52
53,
54
54,
55
55,
56
57
60
61
64
45
48
47,
first
6568
kind
46
47
48
49
The discriminant
50
The
deduced
68
70
7173
74
.......
....
75^77
CHAPTER
79
78,
V.
52
When p>l,
involves a
5355
80
correspondence
hypereUiptic equation
58
59
Number
63
57
integral functions
86
85
87
....
60
84
81
and
Fundamental
Examples
56,
81
any deficiency
89
88
90
92
CONTENTS.
XI
PAGES
6466
68
67,
6971
7279
9398
99101
102104
105112
....
CHAPTER
VI.
Geometrical investigations.
81
80
83
of intersections of curves
Introductory indications of elementary form of theory
84
The method
85
86
87
Adjoint polynomials
....
113
113116
117
.
88
Pliicker's equations
90
89,
a singular place
91
92
Number
first
95
Definitions of set,
96,
97
Theorem
98
lot,
123,
124
125,
126
127
;
determination of
128
reciprocal theorem
of coresidual sets
^4-1
is
expressible
by
100
136
<^-
137
138141
Cayley's theorem
109
105
114
110
134
135
104
108
101
132
133,
polynomials
99,
121
122
kind
94,
120,
93
119
zeros
;
118,
142
147
146
151
152
153159
these
115
119
Elementary considerations
CHAPTER
160
167
VII.
Transcendental uniform
functions.
120
121
122,
123
The
function
i^(:r,
a;
z, c,,
first
...,Cp)
168
169
kind
expressed by
Riemann
....
integrals
124
125
170,
171
172
173
CONTENTS.
Xll
PAGES
XX
126128
The function
ylr{x,a;
z, c)
its
utility
for
expression
the
of
.......
129
functions
131
130,
Algebraic
the
of
expression
);
functions
\jf{x,a;
z, c,,
^^'
..., Cp),
179
134
.........
relation with
its
integral
135
136
parameter
Elementary canonical integral of the second kind
Applications. Canonical integral of the third kind deduced from
.
the function
i//-(a;,
a;
182
185
185
187
186,
z,c^, ...,Cp).
188192
Associated integrals of
138
182
137
178
177,
yjf(x,a;z,c)
132
133,
176
174
rational functions
first
Further canonical
in
140
139,
141
....
142
143
yjr
145
144,
146
147
(x,
z,
c)
.......
156
The disappearance
157
154
158,
159
theorem.
160,
161
162
200
201
202204
205
206
210
212
215
......
of the
207210
212214
The number
199,
211,
equation
197
198
155
153,
195
194
VIII.
152
....
...
Abel's Theorem.
151
theorem
Expression of uniform transcendental function in prime factors
General form of interchange of argument and parameter, after
Abel
CHAPTER
150
Mittag-Leffler's
148
193,
216
own proof
....
....
222224
231234
217222
by the
225
231
CONTENTS.
CHAPTER
IX.
163
164
165
166,
168,
The
167
169
170172
235
236
237
....
238,
239241
242245
CHAPTER
X.
General theory.
176
Number
177
of zeros
is
179
The
180
181
183
places
ij, ...,
Fundamental
mi,
m^
properties.
178
182,
246
248
249 251
252
247,
255
...
Solution of
188
Theory of the
191
special inversion
....
problem
curves
by quotients
268-273
theta
of
Ex-
193
196
198
283
282
286
287
288
289 292
292295
Introduction of the
f>
functions.
functions.
.
296
The
XI.
276
CHAPTER
199
275
274,
identical
197,
259
contact
functions
194
257
258,
259267
..., jp
187
192
256,
Geometrical developments
189
254
253,
186
184
239
297,
299
298
301
CONTENTS.
XIV
205
204,
203
206213
215
214,
Rational expression of
216
algebraical integrals
function.
jl>
Rational expression of
218220
by
Algebraic deduction
f>
of
and rational
by
p function
addition
of the equation
for
the case
p=2.
222
226,
227
232
233235
228
236,
237
337
342
XII.
343
fundamental proi)erties
Convergence of a series; functions associated with the group
The fundamental functions. Comparison with foregoing theory
of this volume
Definition and periodicity of the Schottky prime function
;
...
.
Its connection
238
239
The
240
327
330337
.
225
223
323
functions
CHAPTER
A
317
318 323
309
tr(u+v}a-(u-v)
= (Ai.a^v.{((>v-flhi)
Examples
221
302
303309
327330
function
when p = 2; generaUsation
PAGES
functions.
217
....
....
343348
349352
353359
359364
364366
367372
372,
373
CONTENTS.
XV
PAGES
255
256
257
When
258
259
260
infinities.
o-
+ l>0,
values of a + l and
262
265
is xs'
o-'
269
270
402
403
404
401,
finite integrals
When
functions.
o-'
+ l=0,
primary
+l
+l
arbitrary poles
....
<r
...
272
The theory
The radical
Factorial
tested
(a;,
a;
by examination of a very
particular case
407,
410
413
411
411
413
419
419,
420
422
a particular case of
277
407
276
406,
273
275,
406
essential singularities
271,
405,
Least
+2
268
400
+l
267
399
number
266
........
n-'
264,
+l
398
397,
system
263
07, <r
261,
The formula
The formula
elliptic
272
.
278
279
examples
Connection of theory of factorial functions with theory of automorphic forms
280
281
433
437
437,
439
438
442
introductory.
....
variables,
series of exponentials
431433
XV.
is
426
430
430
CHAPTER
427
443
which
by a
443
445
CONTENTS.
xyi
PAGES
rr
283
282,
Proof that the 2^p theta functions with half-integer characteristics are linearly
285
284,
independent
by
sion
same
theta functions.
rP
order, periods,
Any p+2
its linear
447
447
455
456
461
theta functions of
286
general case
286, 288
Number
are
when jd<4
Examples.
289
same parity
The Gopel biquadratic
of the
relation
CHAPTER
446
expres-
which
....
461464
465470
XVI.
292
293
290
291
theta functions
...
CHAPTER
474
471
474
477
485
477
XVII.
Abbreviations.
Definition of syzygetic
and
azygetic.
References
...
487
488
486,
295
296
297
299
298,
300
303
304,
305
preliminary
lemma
489,
490
490,
491
492
494
502
501
306
308
309
311
317
Other applications of the principles of the chapter. The expression of a function d {nv) as an integral polynomial of order
ant identity
theta functions
312
7)2
in
2''
504
505510
.........
functions ${v)
494
...
510
516
517
527
XVU
CONTENTS.
CHAPTER
XVIII.
318
319 323
324
Riemann
surface
.....
of a theta function
Convergence of the transformed fimction
325
326
327
of syzygetic characteristics
....
....
329
331
332
333,
334
.....
534
534
538
539,
540
541,
542
328
330,
529
538
of even characteristics
529
528,
543
545
544,
Linear transformation of two azygetic systems of theta characteristics into one another
Composition of two transformations of difierent orders supplementary transformations
Formation of p+2 elementary linear transformations by the
composition of which every linear transformation can be
formed determination of the constant factors for each of
.
.......
546
550
552
551,
553557
these
335
336
The constant
Any
factor for
any
linear transformation
linear transformation
may
339
340
341
337,
338
to,, ...,
case.
Weierstrass's
is
m^ upon a Riemann
356
357
362
561
number notation
564
563,
565
567
568
569,
570
XIX.
Columns
of periods.
sion of
351
560,
surface
assigned
CHAPTER
343
344 350
559
562
teristics
On
558,
571
aU period columns by a
finite
number
Expres-
of columns,
579
571
theta function
Expression of Jacobian function by means of theta functions.
Any + 2 Jacobian functions of same periods and parameter
connected by a homogeneous polynomial relation
579588
...
588
598
CONTENTS.
xviii
CHAPTER
XX.
363
365
364,
366,
367
368,
369
of odd order
370
371
372
373
374
375
...
CHAPTER
600
606
607610
611616
617
600
599,
619
619
619 622
622 624
624 628
628
XXI.
377,
....
379
382
383
384
385
386
387
391
393
..........
And
When p>l
spondences
394
388
389
390
number
...
of
^= 1
CHAPTER
632
629
632
636
636
639
639,
640
642
640
645
645 648
642
648,
649,
649
650
651
1) corre-
...
(1,
629
652
654
654
656
XXII.
Example
Weierstrass's theorem.
The property
657
involves a transformation
657
658
CONTENTS.
XIX
PAGES
397
The property
theorem.
involves
= l/r
658,
398
399, 400
401
Determination of
References
403
necessary
form
fundamental
of
659
659
660, 661
equation.
661663
APPENDIX
I.
On algebraic curves
404
in space.
is
an interpreta-
a Biemanu surface
(cf.
162)
APPENDIX
On
405
410
411415
416
418
419, 420
417,
664,
II.
matrices.
Introductory explanations
Decomposition of an Abelian matrix into simpler ones
665
....
particular result
Lemmas
Proof of results assumed in 396, 397
..
666
669
669
674
674
675
675,
678
679
680 684
677,
.
676
ADDENDA. CORRIGENDA.
FAOG
LINE
6,
2,
8,
22,
12,
11,
16, 16,
18,
4,
for
25,
dx
X
read
dx
37,
38,
31,
43,
20,
56,
61,
66,
70,
22,
24,
22,
14,
73,
81.
28,
for x"'''"'''"^
3,
Sj, j,
104, 72.
114,
137.
157.
166.
189,
21,
for
196,
23,
197,
198,
206,
247,
282,
284,
316,
320,
327,
340.
Si, i-
read
xii,
xi.
24,
24,
11,
18,
3
23,
842.
379,
420,
16,
18,
441,
533,
557,
575,
15,
13.
587,
read x"^''"^
of 52 supposes p>l.
See also Hensel, Crelle, cxv. (1895).
from the bottom, add here.
To the references, add, Macaulay, Proc. Lon. Math. Soc, xxvi. p. 495.
See also Kraus, Math. Annal. xvi. (1879).
See also Zeuthen, Ann. d. Mat. 2* Ser., t. in. (1869).
The argument
15,
20,
fc'
= 0.
for Q, read a.
the equation is *Oy=iriP + i>P'.
from the bottom, for u, read Uq.
heading, destroy full stop.
for m(Xp), read fij(Xp).
Further references are given in the report of Brill and Noether (see
Preface), p. 473.
For various notations for characteristics see the references in the report of
Brill and Noether, p. 619.
/or Tj,i, T,,p, read ,''', Vp'".
read ...characteristic, other than the zero characteristic, as the sum of two
different odd half- integer oharaoteristica in
for one, read in turn every combination.
The relation had been given by Frobenius.
for id', read w^.
for from, read for.
8 and 11
the quantity
is
AiA.
In this volume no account is given of the differential equations satisfied by the tbeta
functions, or of their expansion in integral powers of the arguments.
The following referWiltheiss, Crelle, xcix., Math. Annal. xxix., xxxi., xxxm., Gotting
ences may be useful
Pascal,
Gotting.
Nachr.,
1889, pp. 416, 547, Ann. di Mat., Ser. 2, t.
Nachr., 1889, p. 381;
The case p = 2 is considered in Krause,
XVII.; Burkhardt (and Klein), Math. Annal. xxxii.
Transf. Hyperellip. Functionen.
:
(1)
(3)
The following books of recent appearance, not referred to in the text, may be named here.
The completion of Picard, Traite d^ Analyse, (2) Jordan, Cour d' Analyse, t. ii. (1894)
Appell and Goursat, Tlieorie des Fonctions alg4briques et de leurs integrales (1895), (4)
CHAPTER
The Subject of
This book
1.
is
when a quantity y
x by means of an equation
Coy"
wherein
Investigation.
ao, a,
I.
+ Oiy"-' +
. . .
+ a_iy + a = 0,
supposed to be irreducible
that
is,
defined in terms
is
of the form
The equation
x.
is
Of the
2.
various
regarded as fundamental.
Of
an account
may behave
in one of
two ways.
Either
may just
Or the sheets may wind into one another in which case we shall
sheet.
regard this winding point (or branch point) as constituting one place this
place belongs then indifferently to either sheet the sheets here merge into
:
one another.
In the
a be the value of x
convenience of statement to be
first
case, if
for
finite,
B.
Gesam-
[2
a,
exponents.'
curve
say that
hence, adopting a definition given by Rieraann*. we shall
In the
places.
is an infinitesimal of the first order at each of the
a;
second case the attempt to enclose the place by a curve leads to a curve
will
be
-a
that
lying partly in one sheet and partly in the other; in fact, in order
may be
the curve
closed it
id log [(x
- a)i]
place.
In this
the plane of
t,
surrounding
= 0.
3.
In general
points^
for
the
denoted by Wj
any
finite
number
make the
may
k, branch
wind at these branch points may be
respectively, where
be several, say
of sheets that
1, Wo-t- 1, ...,
wic+
corresponding w.
no branch point
is
two of the n values of y are the same, each of Wj, Wa, ..., w^ is zero and k is
equal to n and in the second case above, the values are A; = n. 1 Wi = 1 Wj = 0,
Wj = 0,
Wt = 0. In the general case ea<:h of these k branch points is called a
:
. .
place,
*
The
0)"'+^,
is
...,
o)'+i
fnnction of x, y.
may
p. 171.
Prym,
Crelle,
Bd. 70.
4]
similarly have
or a less
0"^^'
'
t)"'^'-
^^^^^
number
For the
...
+(w, +
a?
x we
infinite value of
of places and as
K + l)+
l)
many
shall
infinitesimals, say
= n. And
as in the par-
two characteristics that for the immediate neighbourand y are uniquely expressible thereby (in series of
and conversely
is
place of the
And
Riemann
a place
letter.
in
fact
When we
We
4.
We may
algebraic irrationality.
a certain
is
is
practi-
a reversible character.
of correspondence,
it is
surfaces
class of
make it
Riemann
and are not the exclusive outcome of that one which we adopt
initially.
Let f be any one of those uniform functions of position on the fundamental (undissected) Riemann surface whose infinities are all of finite order.
Such functions can be expressed rationally by x and y*- For that reason we
speak of them shortly as the rational functions of the surface. The
order of infinity of such a function at any place of the surface where the
function becomes infinite is the same as that of a certain integral power of
shall
The sum
of these orders of
infinity for all the infinities of the function is called the order of the function.
The number
equal to
a^f.
^.
t is
Let
is
is
rj
And
zero of f - o
the
sum
infinity of
log(|-o),
|,
is
equal to the
as the case
may
of the integrals for all such places is equal to the value round the boundary of
Cf.
12
CONDITION OF REVERSIBILITY.
4
order
fi.
[4
surface.
The values
of
at X^,
t]
...,
X^ may
or
may
value of H, namely,
Let
not be different.
17.
unique
clearly the
same
function of
^.
7]
sX Xi,
...,
power of f
as that of a finite
Hence,
X,r
if
Hr
together,
denote the
77
satisfies
sum
o.
Hence
ia
a.
rational
an equation
It is conceivable that the left side of this equation can be written as the
product of several factors each rational in
f and 17. If possible let this be
done.
Construct over the f plane the Riemann surfaces corresponding to
77
definite values,
imposed
surfaces.
17)
Let %,
(and
its
be the values of
77 belonging, on
one of
these surfaces, to a value of
f and 77/, ...,77/ the values belonging to the
same value of ^ on another of these surfaces. Since for each
of these surfaces
there are only a finite number of values of
^ at which the values of 77 are
not all different, we may suppose that all
these r values on the one
surface are different from one another, and
likewise the s values on the other
surface.
Since each of the pairs of values
(f, ,7.), ..., (f, 77,) must arise on
both these surfaces, it follows that the values
77,
,7, are included among
771', ...,
V- Similarly the values 77/
,7; are included among 77,, ..., ,,^
Hence these two sets are the same and r = s.
Since this is true for an
infinite number of values of
f, it follows that these two surfaces are merely
repetitions of one another.
The same is true
...,77,
:
for every
Hence ris a
divisor of v
V"
when
reducible,
- H.r,"--^ +
...
-^{-y H,
= 0,
It
(f, ,,)
5]
Let now X^
Z be
another except
these by
Pi,P
Let
pairs of values {x^, y^, (x^, y^), ..., which the fundamental variables x, y of the
original surface assume at Pj, P^,
Then to any pair of values (f, t)) will
correspond only one value of
namely,
is a one-valued function on the
(^, 17)
a rational function of
In particular
17.
f,
There
surfaces
Hence
surface.
f,
17
of the
kind which we
call
(1, -
is
as also are
y^,
j/i,
between the
(^, 17)
and
{x,
y)
(f, 17)
surface; to every
The
case
In that
surface.
that
Many
advantages
may be
We may compare
fact.
new
surface
is
The
5.
of
I,
following
method*
the expression of x,
in terms
r,.
ix,
y)
^,
7 in terms of x,
be given by
^ {x, y) -r)x{x,y)= 0,
and let the rational result of eliminating x, y between these equations and the initial
equation connecting ar, y be denoted by F{^, ij) = 0, each of <^, ..., x, i^ denoting integral
Let two terms of the expression <^{x, y)-k^{x, y)=0 be ax^y'-(:hx'-y.
This expression and therefore all others involved will be unaltered if a, h be replaced by
such quantities o+A, 6+i, that hx^y'=ikx''y'. In a formal sense this changes F{^, 17)
polynomials.
into
a and b of order
Hence the term within the square brackets in this expression must be
that k=hxji.
possible, choose now r=/ + l and s=s\ so
* Salmon's Higher Algebra (1885), p. 97, 103.
zero.
If
it is
ALGEBRAICAL FORMULATION.
[5
This
I).
is
And y
is
to,
by which x
is
similarly determinate.
x,
y by
when
ij,
|,
possible
it is
^(^,3')-w(^'.y)=. /(^.y)=o.
it is only the places x, y
first two eqiiations
which are rationally obtainable from the places |,
satisfying the equation f{x, y) =
satisfying the equation F{^, i) = 0. There do exist transformations, rationally reversible,
They are those known as Cremona-transformations*.
subject to no such restriction.
^
They can be compounded by reapplication of the transformation x y l =
*(-i>2/)-^'K-^.^')=o.
i;
AVe
may
give
t]
an example
(.?;
of order
is
a)~, and
2,
the function
Ti
= xjy
is
of order
From
we immediately
find, as
x^+x+l=0,
4,
at the place.
t/,
2,,-f + 5^-5=0,
and
infer, since
order 2 in
i;,
Thence
the
new
in accordance
and
(^,
ij)
on
,,^=xy{x''+x+l), y=x/,,.
If
however we take
i=yV{x^+x+l), ,=y/(^-6,2),
where
ij
is a fimction of order
equations are reversible independently of the original equation, giving in fact
a>
is
3,
these
having a
(1, I)
jHjssible to
obtain reversible
When a surface (x, y) is (1, 1) related to a (f, t)) surface, the defi6.
ciencies of the surfaces, as defined by Riemann by means of the connectivity,
must
clearly
be the same.
*
RELATION OF DEFICIENCES.
6]
how many
-^
ax
Consider at
It is infinite
is zero.
at the places where f is infinite: suppose for simplicity that these are
separated places on the original surface or in other words are infinities of
the first order, and are not at the branch points of the original surface.
At
a pole of
where
a;
?>
is infinite
twice.
- a = '*+': namely it is
places where j-
f,
and f be there
2p-2
/Vt
2n =
2v+2p -2
ax
times:
of
Sw = 2m + 2p - 2
like t^ at a
is infinite
infinite
It
namely
,
.
"^
2+w
ax
before.
Now
for
same
value.
case are
When
v/r
is
new
1,
-f-
is
same
of the
new
2v+2p 2
surface.
is
different.
value,
among
they can
Since -
there are
the case
accidental coincidences
there are
= 0,
or
branch places of
of deficiency p.
The consecutive
places
Conversely, to
a branch
place of the
new
surface,
new
surface,
w'-
+ C=2v + 2p-2,
(1879), p. 326,
364,
-a
is
first.
leads to the
be the deficiency of the new surface (of r sheets), this
if p'
and
[6
equation
(2r+2p'-2)-^+G =
+ 2p-2,
2v
from which
V
C = 2p-2-(2p-2)-.
r
If
Corollary*(7
0,
=p', then
We
is
C = {2p- 2)
(l
- ^)
Thus ^
>> 1, so
reversible.
first
order.
The number
at infinity.
that
In that
is
correspondingly
less.
are of
places
The
unaffected.
is
In the example previously given ( 5) shew that the function | takes any given
Ex.
value at two points of the original surface (other than the branch places where it is
infinite), i; having the same value for these two points, and that there are six places at
which these two places coincide. (These are the place {x=Q, y = 0) and the five places
where
x= 2.)
There
is
tj)
surface is -
where v
is
the
order of f and r is not greater than v, being the number of sheets of the (f r))
surface hence, if there were a function f of order 1 the correspondence would
,
be reversible
7.
and
would
be of deficiency 1.
The
the equation.
upon a
finite
But conversely
it is
the equation be reversibly transformable into another given one, that the
many
surface.
number
BY TRANSFORMATION.
'7]
It will
is
limit,
and that
number of
Further
it
will
appear,
that
characterised by 3p 3 constants
one, and for ^ =
none.
when p >
the number
tor
statement
is
being
shall content
When
we
p=
a surface
is
new
surface
there will be dp
the
new
of
them whose
surface
position
is
is
reversible
settled
by the
we may regard
We
depends on 3^ 3 parameters
or on less, for the 3p 3 undetermined branch points of the new surface may
have mutually dependent positions.
infer therefore that the original surface
contains^
are prescribed:
Qp+l
may
it
more
contain
for
values of
when
its
Q<2p1,
poles
but we
Ex.
1.
The
of deficiency
Ex.
2.
3.
The
It
depends on 5 = 2p-
parameters, k% X^
/j,',
i^,
p\
wherein the coefficients are integral polynomials of the orders specified by the suffixes, is
Shew that it can be transformed to a form containing only 6 = 2/> of deficiency 3.
parametric constants.
* See the Chapters on the geometrical theoiy and on the inversion of Abelian Integrals.
The
reason for the exception in case p=0 or 1 will appear most clearly in the Chapter on the selfcorrespondence of a Biemann surface. But it is a familiar fact that the elliptic functions which
1 depend upon one parameter, commonly
and the trigonometrical functions involve no such parameter.
+ Forsyth, p. 459. The theorems here quoted are considered in detail in Chapter III.
called
of the
present book.
p.
113.
Klein,
Ueber Riemann's
Theorie
(Leipzig,
[8
SELF-COERESPONDENCE.
JO
possible to
a case in which this argument fails. If it be
transformareversible
rational
a
by
itself
transform the original surface into
on the surface are efifectively
tion involving parameters, any r places
Then the
other r places.
equivalent with, as being transformable into, any
disposeonly
but
supply Q
Q poles of the function | do not effectively
are
there
3/) - 3 + r
So that
able constants with which to fix the new surface.
In this
control.
our
branch points of the new surface which remain beyond
But there
8.
is
Q-r
disposeable
the surfaces of the class contain 3p -3
and serve to
parameters bm.de r parameters which remain indeterminate
It will be
the surface.
represent the possibility of the self-transformation of
case
all
possibility only
shewn in the chapter on self-transformation that the
cases, respectively
these
in
are,
of
r
or i) = 1, and that the values
for p =
the fundamental
when
=
that
We remark as to the case p
3 and 1.
into itself by
transformed
surface has only one sheet it can clearly be
arises
'
^^^
^ regard top =
l,
the
the case of elliptic functions, that effectively a point represented by
by an
elliptic argument h is equivalent to any other point represented
argument u +
and
7.
poles (a,
for the
9.
/3)
same value
at
u as has
is
^a+y.ff+r at
i6
-(-
so that the
Riemann
and
it is
= 0,
surface,
may
also
be regarded
on the
make
therefore proper to
frequent use of
The branch points of the surface correspond to those points of the curve
where a line x = constant meets the curve in two or more consecutive points
On the
as for instance when it touches the curve, or passes through a cusp.
other hand a double point of the curve corresponds to a point on the surface
where two sheets just touch without further connexion. Thus the branch
place of the surface which corresponds to a cusp
larity to that
' Cf.
p.
273
etc.
is
is
touched by a
GEOMETRICAL VIEW.
9]
line
11
Riemann
of
an ordinary branch
Riemann
where n
is
= (i-l)n-28-3,
that the
y, it follows
w = < 4- K = (n - 1) n _ 2 (8 + k)
= 2?i- 2 + 2 {^71 - 1) (ft- 2) - SThus
since
w = 2n 2 -^p,
number
of
is
/cj.
is
\{n-V,{n-2)-h-K,
namely the number which
is
touches f(x, y)
riant, is
= 0. The
known
to
where n
is
the order of
<^ (x,
y)
f, to
<f)
so,
<^ {x,
y)
y)
a degree* n (n
(x,
0.
n (w 3)
2S
2/f + 2nn' =
number is therefore
the number of intersections of the curves /"(a;, y) =
namely
it is
rj
is
total
<f>
arising for
denoted by v
The
Now nn' is
2/) 2 + Inn'.
and (x, y)
(x, y) = 0,
as before.
Ex. Prove that if from any point of a curve, ordinary or multiple, or from a point not
on the curve, t be the number of tangents which can be drawn other than those touching
* See Salmon,
(1879), p. 81.
GENERALITY
12
[9
at the point,
and
isg*.
The geometrical
10.
may however be
x^, ...
.Vk+i-
Let
u, v
u-^
same
coordinates of the
The values
order.
of
at the points
where
^v
to
order
finite
it
And
Of these
in space are
we
deal
upon a Riemann
we can
surface.
and
is
their
is
order.
Riemann
If the
it
For instance, we
*
may
start with
desires
vii.
to
H.
Brill u. Noether,
Math. Annal.
7.
vi.;
that of an
may consult:
Noether, Math. Annul.
J. S.
is
OF THE THEORY.
11]
13
we may
a plane which
is
elliptic functions.
Or
drawn upon the plane, and construct for this surface a set of elliptic functions.
Much light is thrown upon the functions occurring in the theory by thus
considering them in terms of what are in fact diiferent independent variables.
And further gain arises by going a step further. The infinite plane upon
which uniform functions of a single variable are represented may be regarded
as an infinite sphere and such surfaces as that of which the anchor ring
above is an example may be regarded as generalizations of that simple case.
;
Now we
Riemann
Riemann
surface,
we may
not uniform upon that surface, the branches being separated by drawing
barriers upon the surface.
The properties obtained will obviously generalize
the properties of the functions which are uniform upon the surface.
Forsyth,
p.
318
Eiemann,
Ges.
Werke
[12
CHAPTER
II.
Chapter VI.
it
be seen that
will
descriptive theory
it offers
exclude
it
It is therefore
It is
is
Nevertheless
all
Let
finite place,
all
if it is
finite,
a;
t'"+^
if
an
infinite
place,
(""'"+''.
There exists a function which save for certain additive moduli is one-valued
on the whole surface and everywhere finite and continuous, save at the
place in question, in the neighbourhood of which it can be expressed in the
form
* See for instaDce Forsyth, Theory of Functions of a Complex Variable, 1893 Harkness and
Morley, Treatise on the Theory of Functions, 1893 Schwarz, Gesam. math. Abhandlungen, 1890.
The best of the early systematic expositions of many of the ideas involved is fonnd in
:
C.
Neumann, Vorlesungen
study.
iiber
d' Analyse,
I.
2, 3.
is
recommended
to
14]
Herein, as throughout,
when
vanishing
P (t) denotes
0, C,
A,
...,
and r
15
is
assigned beforehand.
-l + G + P(t).
This function
We
is
by 2p cuts+, which we
rendering the function one-valued over the whole of the dissected surface.
We
across the
loops of the
first
The
arbitrary place.
of
is
then denoted by
vanishes and
a;
The
X.
kind
will
or
a;
Fo^' \ c
oo
that
is
When
the infinity
the function
is
is infinite
an ordinary place,
either like
(Xiyi), (x^y^)
made between
11'''
by
ti, t^,
of these places
denoted by
unique
cut be
is
be denoted by
Let
14.
=a
at which either x
or
special
It
for transit
"
the places
which
{x^y^), {x-^y^.
and
let
the
x Xi = t.^''-^^ Let a
There exists a function, here
= <i"''+',
p periods arising for transit of the period loops of the second kind
has
no periods at the period loop of the first kind, (7) is everywhere
and
continuous and finite save near {x^yi) and {x^y^), where it is infinite respectively like log ti and - log t^, and, (S), vanishes when the current place
denoted by x is the place denoted by c. This function is unique. If the
(/8)
has
cut between (x^y^), (x^y^) be not made, the function is only definite apart
from an additive integral multiple of 27n, whose value depends on the
*
is
also called an
cuts.
the
the function has no periods at the period loops of
kind.
second
of
the
integral
normal
a
function
the
calling
rally denoted by
The
fact that
first
kind
is
gene-
16
is
is
c.
[14
It will be called* the
(ti/ti).
Beside these functions there exist also certain integrals of the first
kind in number p. They are everywhere continuous and finite and oneFor transit of the period loops of the
valued on the dissected surface.
15.
first
Qi.
This period
is
Vi,
here taken to be
1.
The
scheme of periods
i""
loop,
...,
np.
the
We may
16]
A more general
17
same property
is
i.J
wherein the arbitrary coefficients satisfy the equations Afj = Aji. The property is usually referred to as the theorem of the interchange of argument
(x) and parameter (xi).
The property
as
a function of
should be
Xi
KL is
But
finite
if
x-i
c,
x.
made
ig
x,
the function
(< - ^) =
n:;'r^
w+
x^
- log (x,' - x)
in the
is
neighbourhood of
sheets wind,
and
a;,',
x.
x be two
KU -
^"&
(*''
- *)'
n^'x,
x approaches
- ^i^i
and
x^'
ig (*
Xi,
- ^)
so that
n;-':''-iog(^.'-^)
is
In this case,
let
a continuous function of
This
is
a;/
a;/,
Then the
x.
or f/
up
to
to
is
that
is
based*.
^V
regarded as a function of
jt,',
is
such that
r'-'
Xi.
is
and
f be the represent-
correct statement
of its existence
up
is
ar/
x^.
x.
p. 220.
18
may be desirable
to give
of these integrals.
17.
It
(a)
[17
y2=a;(;r_ai)...(a:-a2p + i),
wherein
Oj,
a^p+j are
...,
and
all finite
different
integral
^-U'
y \^-k
being places of the surface other than the branch places, which are
(0,
0),(ai, 0),
...,
(ajp + i, 0).
-,),
-);
(|,,
for (y+7)/(-^-^)>
(y+liM^-fi)
At a
and the
place
a;
= oo
being a constant.
At a
place y = 0,
;;:
being
and Pj
(J)
a series of
form
It is therefore finite.
for instance
where
x=a^ + t\ y = Bt\\+P^{t)l
B being
C jdt\\+P,{t)\
C being a
constant,
Thus
It
it is
may be
and
is finite.
J
is infinite
Near
at (fj,
(0, 0),
ij,)
like -log(a:-^,)
infinities at (f,
;),
(f j,
i;j).
integral
\x
and
is
x^^j
not elsewhere
tdt
/
It is therefore
branch place
(0,
0)
=is't
infinity at the
where
ij'
= ^.
fdx d
/y+,\_
is
(f,
jj).
17]
19
POINT.
f^ v+{^-$)'i'+H^-$f')"+I.(^-|)=''7+(^-f)V+i(^-f)S"+..."
which
is
equal to
/<^.[-*f'---]Thus the
j, and
is
(0, 0), is
a multiple of
il-.
x=0, writing x=t\ y = Di[l+P{fi)],
In fact near
this integral
becomes
or
which
is
equal to
as desired.
The
integral
Example
is
1.
i_
as the place
{$,
Example
ij)
the limit of
[^riji_f\
Shew
2.
is
(0, 0).
first
kind
(Ai+A^+...+Ap^jXp-^).
O) We have in the first chapter 2, 3 spoken of a circumstance that can arise, that
two sheets of the surface just touch at a point and have no further connexion, and we
have said that we regard the points of the sheets as distinct places. Accordingly w^e may
have an integral of the third kind which has its infinities at these two places, or an integral
For example, on the
of the third kind having one of its infinities at one of these places.
siu-face
where
(x,
suffixes,
y\,
{x,
y\
two such
places, at both of
coefficients,
and
OTj,
m^ are
by the
^=0
which y=0.
In this case
[
dx
JfiyV
where f{y)
=/
is
(0,
infinities at
and
fy
I'-
- m.yX+Ax^+Bxy -'rCy^ dx
Lx+Sy
/'(y)
22
EXAMPLES.
20
is
[17
C be
so chosen that
y-m^x+AaP'+Bxy + Cy'^ vanishes at one of the two places other than (0, 0) at which
Lx-^My is zero. Its infinities are at (i) the uncompensated zero of Lx-k-My which is not
at (0, 0), (ii) the place (0, 0) at which the expression of y in terms of x is of the form
y = m-fS + P.'t' + a,-3 +
In
fact,
. .
dx
I^tt.
is finite
At each
of the places
(0, 0),
easy to verify
is
they lead immediately to the proof that the integrals have the character
enunciated.
The
C will
not determine
them uniquely
the
""*'/^'
/'
y<'={x-a)(x-by{x-cy,
= {x-a){x-b) {x-cf,
y^={x-a){x-b) {x-cf,
y'' = {x-a){xb)
{x-cf,
y*
first
kind are
'^{x-b){x-cf.
^{^-cf{Ay + B{x-c)\
C are
arbitrary constants.
y''
= R(x)
"
Eueen
-^
the
sum
Riemann
is zero.
('Fdx, for
all
18]
It is
supposed that the number of places where negative powers of t occur in the
is finite, but it is not necessary that the number of negative powers be
expansion of
finite.
21
integration of
18.
integral* IT"''
dv".'"
formed by the two sides of the pairs of period loops (a,, 6,),
a direction that the interior of the surface is always on the
.
to the value
Round
clockwise direction.
sole infinity,
(a^, bp), in
j dv''"
hand,
left
a, in
closed curves
is
such
equal
a counter-
is
taken once positively in the direction of the arrow head round what in the
This value is fl, ( &),>), where toir denotes the
figure is the outer side of br.
period of
Vi for
is
the inside of
The
relations indicated
is
will
by the figure
adopted
for
zero except
when r =
i,
the
sum
of these
p contour
t'-''
= -- + A+Bt + Gf + ...,
t
-y,
integrals
Taken
D denotes at
and
where
where
cair, t,>
Hence, as
Wj,
tD<-'' ^
..
= 1,
X,r\
V.
p. 439.
dt,
22
This
if
is
In this case
times at
iCi,
it is
Wj
From
19.
^ x,c\
0,2-,ri/"'\
times at
x^,
may be
+1
closed.
we can shew
these results
derivable by
may prove
0,
is
[18
differentiation
is
its infinities
is
of the
this is not
We can in
namely,
in its
if,
fact
most general
to take the
fx=0
Xi
case, x^
= Xi + t
a;/
a place
the equation,
-n"'"^^l = r-'
,-rn"v
^*'
'
tx, I
^"^J
is
to Xi than
^d
a;
a;,
n^;,; ^
x^'
log (x
= x,
- xo + ^' = log (f -
Xi
= (| - f, )"'+',
^-^/ = (?-?.')[c+P{f-f')].
Then
where
we may regard
so situate that
is.
K^,x,
^1
log {>-X,)
?/)
of the places
4,'.
and reviains
+ ^ = log (? - fO +
4>,
finite
19]
where
is also finite.
<f>
X,', X,
Therefore
^_^^y^T'
^y
xx,
and thus
lira.
where
Now
yjr
as
J-/, X,
Xj
-^
,
,.
is finite.
fi'
moves up
to fi, for
a fixed position of
^,
we have
23
24
[19
_1 _1 _1
fl,
We
is
"X,
I'll
mean t
will therefore
iii.
By means
instead of A^,,
...
17
it
can
When
X)?^,
a;,
etc.
now be shewn
[Fdx,
I'
which F
having only
in
20.
One
parametric place.
iii.
be stated.
the deficiency
is
is
zero),
and of
finite
we
can,
its
series of
- r^.r^ + A,D^r^-''+
-^
'
...
2)^-> r^ "1
obtain a function nowhere infinite on the surface and having no periods at the
first
p period loops. Such a function is a constantf Hence can also be
.
functions
certain
necessary relations
among the
coefficients
t Forsyth,
p. 439.
p. 262.
Ai,...,Am.
^IJ
Of
21.
special
dx
first place,
justifies
T^ ^
dvp
dVi
dx
'
these cannot be
For they
the surface.
integral
importance
In the
p whose
all
25
'
all
" dx
zero for any ordinary finite place a of
first
order
it
would be a rational
sentable conformally upon another surface of one sheet*, f = ra"^-" being the
new independent variable; and the transformation would be reversible
I. 6).
Hence the original surface would be of deficiency zero
which case the only integral of the first kind is a constant. The functions
are all infinite at a branch place a.
But it can be shewn as here that the
quantities to which they are there proportional, namely DaV^, ..., DaVp, cannot
be all zero. The functions are all zero at infinity, but similarly it can be
(Chap.
in
quantities, Dvi,
...
Dvp, cannot be
all
zero there.
Thus
vanish in an infinite
number
We
and
vanish.
we
special,
is
x.
We
no place at which
all
may
...
dvp has| 2p
zeros,
...,
dvp
For instance,
some of which
occur at infinity.
first
kind
\^Vi'+... -\-\pVp'
must necessarily be
surface
is
finite all
* I
owe
this
argument
X See Forsyth,
p. 461.
it
to Prof. Klein.
p. 450.
26
is
[21
M {\,^^+...+\
where Fj
new
= Vi
is
an integral of the
first
^^
new independent
variable,
first
kind
they
and
first
M = dP
-^
may be
said to be invariant
may be made
This point
of the
first
kind
for
clearer
by an example.
The general
integral
the surface
r=
(a;,
1)8
can be shewn to be
/^{A+Bx
+
y
A, B,
G being
Ga?),
arbitrary constants.
If then <^i ^2 4>, denote the ratios of any three linearly independent
integrands of the first kind for this surface, we have
:
1
for
x'
= ai<^i +
bi<f>,
Ci<f)s
a.<f>i
b,(j)2
c^<f>,
a3<f>i
b,<f>,
+ c^,
Cj,
and hence
(a,<^
6,</)2
c,<^3) (a3(^,
63(^2
It
An
integral of the
way
first
integi-als
in
period loops will be a linear function of the integrals of the first kind which
are normal for another set and an integral of the second or third kind,
which
is normal for one set of period loops, will for another set differ
from a normal
;
integral
by an additive
24]
27
CHAPTER
The
Uniform Functions.
Infinities of Rational
23.
III.
order,
number being
aj
tt,,
...
a^,
finite.
the
We
deal
the
all
of
first
all
first order.
of
is
the form
where the
coefficients
being
left
undetermined.
( 20),
fi,,, fJ'i
ft^k
must
all
be
zero.
may
as
its
poles
must be of
this form.
...,
f2p(a),
+ fi,ni
(fl.)
i.
. . .
+ /iifii (at) =
values of
i,
0,
when
for all
the equations
n.- (C)
=f,ni (C,)
. . .
+frili (Cr)
hold for finite values of the coefficients /i,...,/r, these coefficients being
independent ofi. Hence we may also say
In order
its poles,
that
each simple,
me
exist
. .
. .
28
upon
tti,
[24
Oj, ..., a
we have
= n,+,,
^i (a*)
= i,
Hi (a,)
. . .
+ n,+i,
In particular,
be as great as p and
if s
aj
. .
same
enough disposeable
values of
be independent, equations
For then we have
...,*.
a,
,
.
When
^i (a.)
^t (oi)
equations.
it
Rk
wherein
ai ...
Xj+i
and
ttj
+^i
=<rk
[r^i
... \jt
-"*,>
r^,
r^,]
-*-*,.
order at
a-^,
is
infinite to the
...,aic,
Mo
+ ^1
r^,
+ M*
rn^
R,+,
+
r 1
n.+i,. r^,
- T^M
+ ,+,,
+ n,. r^^^*"|,
It.
M*
Rk +n,,,
r:,
'
\-k
XiJ
+ ^iT^, +
"o
If this function is to
when
hold.
Vi
+ y,+iii,+i+
+i/,ni(a,)
+VkRk.
= 0,
(i=
1,
^),
and involves k
places
Ta,
i',n.(aO
must
+v,
among
s+l
Oi,
...
be called arbitrary
..., aj;,
"s+iRf+i
is
of the form
+ VkRi,
ajt
25]
which has
all
its
among
poles
only infinite at
tion,
Oj,
...,
given below in
The
a^.
Oj, ...,ai
29
37, is called
enunciated by Riemann*,
first
25.
order,
simple infinities at
By
the
first
Oj, ...,
Oj
...
tttOi+i
...
upon the
...
surface in a definite
on
at+i, at+j,
oa-i, that
possible
be that
depends on
is,
Oi
. . .
for
as great as p,
a, at+i
. . .
Ot+i
... ai,_i
at+i
. . .
depend upon the preceding places a, ... a,, it will not arise before:
case there will be no rational function having for poles the places
a* Ot+i
for
any
a/,_i
in that
*+j
p s.
But in order
than or equal to
value
which a^ depends
This will
a^-i.
a^+t, ,*
it
will
which beside
ajc+i,
p s,
which
at-Lj
ajc+j,
ffll
Then there
at+i
ttj
t+i
it in
no one of
the series
O'k+j-
at+j,
a.
>
^k-\-j+i
~ "t+j+i.i+j I ai + ,J
""
' o,
Ta^
^.
from
ak+j, aic^j+i
Ug, aic+i
Oj
to
h k -j.
p.
101
( 5)
and
+ CreUe, 64.
has been much extended by Brill and Noether.
p. 118 ( 14)
{Math. Ann.
vii.)
of the
theorem
30
!
is
[25
*+>+
ajt+j^t+j+i
ajcttk+i
in fact
+ ^t+j+i Hk+j+i
+vtRi + vjc+j+i Bk+j+i +
+ 1/,+, Rs+i +
number
involves k-s + i+1 arbitrary constants, namely the same
j/
and
as
at+iflifc+j+i
*ti+i
asOn+i
Oi
+/islt,
+ /it
+ Mmlt,+i+
+ f^i+i 1^1+1+
r^i
namely,
k-s
+ 2
/ig+r
hr
+ "+r,l
a,
"^
r=l
which
I'o
is
of the form
+ J'ir^, +
+
+Vs^a,+
vk+i Iti^,
and the
"t+jlt^^,.
VtRk
vk+i+iR^^i^i
+ Vk+j+iR^^j^r,
+ v,D, (a^) +
i/,n(a,)+
Vs+iJis+l+
vk+iil (ak+i)
v^ ...
. . .
Vk+j
VgVk+i
+'t+jQ(ai+_,),
be
ak+j
places.
result
number
of places
pendent
i...a_
of
.
those
a,,
a,
preceding
it,
but each of a
it
o-q
in the series aj
a depends" upon
a,
"f
&-,,+! ", "-r
Oi'
...
"e,-?,
+\^q
o
.
a e.+i
^^
+i,---
^o
a l
Vl +
a ,-,,
Vl-9l
...
Q.-9. + [Q.-9.-Q.]>p.
,.
. . .
...
^^
is
inde-
depends on
independent of
V9~V2>
but each of
26]
a.Q^.,^ is
ai...OQ_aQa
ao,,
-q
but each of a
it
^,...a
series.
31
We
shall eventually
come
to places
'
the total number of independent places included, that is, of places which
do not depend upon those of our series which precede them, is
p so that
the equation
+ (Q, - q,-Q,) +
P = {Qh-qh- Qk-i}+
= Qh-qi-q2-qh
Then every
will hold.
in order ffom a^
whole
...
will
series.
using
it follows,
among
a,
. . .
a/,
R/
exist.
For
To
- q,)
^^,a^
do not
(Q,
is p.
a rational function
Rf
eadsts.
numbers
1, 2, ...
places arise by
its
then, denoting F*
which has
by
poles at
its
poles at
aQ,%,+i
- V-9+'-
i be one of the
upon the preceding
32
is
[26
of the form
and involves
The
5,
result
1, 2, ...
g'j
+ gm-i + ^ +
may be summarised by
(Qa -?.),
(Q.,-9,
(Q>-?.
+ l),
arbitrary coefficients.
putting
-!. Qi
down
+ 1.
+ l),...,QQ.+ l,...,Qft-i +
(Q.-?2),
l,
..(QA-?^).(Qft-?^+l).
with a bar drawn above the indices corresponding to the places which depend
upon those preceding them in the series. The bar beginning over Qa - g/i + 1
The total number of indices over which
is then continuous to any length.
no bar
drawn
is
above, for
The
is
p.
is
pose that for our initial algebraic equation or our initial surface,
Sup-
we were able
Kl, discontinuous
at
order, this function being one valued save for additive multiples of k periods,
and these periods finite and uniquely dependent upon a, then, taking arbitrary
upon the
...
that
is,
we
for
characteristic
number k
We
shall
which
of the
is
p based
sometimes
refer
to
the
proposition
as
Weierstrass's
gap
theorem*.
When a place a
27.
61, 62.
6r, it is
"Luckensatz."
is,
I believe primarily
under
referred to
is
published by Schottky
27]
33
A,a,(b,)+...+Apnp(b,)=o
+ Apilj, (K) =
+... + Apnp(a)=0
A,a, (6,) +
A^iliia)
the last
last
. . .
is
linear aggregate
il (x)
wherein
i^ qm,
{Q.-qd + (Q2-q^-Qi)+---+{Qm-qm-Qm-,)
or
Qm-qilinearly
-qm
independent equations.
If then T
the form il
+1
{x),
Denoting Qm
qm+
T+
l=p-{Qm-qi- -qm)-
by
Q,
+ 1,
merely additive, by 5
q
We
+ l=qi + qi+
...+ g^^i
+ i+l.
therefore have
Q-g=;)-(T + l).
Recalling the values of n,(a;)...
that every linear aggregate of
that
when Q
is
greater than
them
2p - 2,
and the
fij,(a;)
fact
vanishes in just 2p
I is
(Chapter
II.
21)
we
see
places,
necessarily zero.
which vanish
Oi'*!!
is
given,
article,
by taking
7n
,%+!
and
",.!'
= qh-i
by the equation
T
B.
= h-l
in the places
[27
34,
and therefore
QA-gA-i>>2p-2
or, the
is
given below
a^^^_^^
of our
exist, is
series,
corresponding
2p-l.
to
is
Ex.
37,
2.
the
limiting case of the problem just discussed is that in which
surface.
the
of
place
series of points a a^, ... are all consecutive at one
28.
rational function
surface,
and there
which becomes
infinite
"^
r'
f"^
is
only at a place,
a,
of the
like
O^,
. . .
'*'
zero,
If X;
being the
= Ci/(i 1)
!,
\+
X, rs
+ x^DaFs +
. .
+ vDj-'
where, in order that the function be one valued on the undissected surface,
the
equations
Xifii (a)
must be
satisfied
...
sufficient conditions
X^.
must be expressible
fi,-
(a),
is infinite
each of the r
to
1th
= l,
for
the
D^~^D,i(a), for
column
and
2, ...,p,
i.
We
imagine the
moment we
we
quantities
call
the rth
dition for the existence of a rational function, infinite of the rth order at a,
infinite,
is
that the
rlih
column be a
linear function
Then
themselves into two categories, those which are linear functions of the pre-
is p,
the
number
columns
30]
35
will be just p.
Let them be in succession the Ajth, Jb^th, ...,/tj,th. Then
there exists no rational function infinite only at a, and there to these
orders A;,, k^, ..., kp, though there are integrals of the second kind infinite
But
to these orders.
if
Qth
order, its
A,, ...,kp,
there does
>-Q^?-^rs
+ \q-xi>r^vi +
. . .
+ x^rs + X,
where p (t
+ 1)
is
the
number
form
Xifti (a)
\J)li{a)
...
+ \Qi)-ini (a) = 0,
{i
1, 2, ...,p),
satisfy the
= 0,
first
...
+ Apiynj, (a) =
...,Q-1.
29.
ki
1, 2,
the
conditions
A^D'D., (a)
forr
1 is
+ ...+Aj,D.p{x)
AiD.^{a;)
which
ki
...
kp
we remark
firstly that,
unless
p = 0,
functions of all other orders with their infinity at this one place, and there
would be no gaps (compare the argument Chapter II. 21) and further
that in general they are the numbers 1, 2, 3 ... p, that is to say, there is only
a finite number of places on the surface for which a rational function can be
formed infinite there to an order less than p + \ and not otherwise infinite.
We shall prove this immediately by finding an upper and a lower limit to
;
the
number
Some
30.
Chapter V.
in considering the geometrical theory, that the zeros of the linear aggregate
It
is
may
find
it
more convenient
vi.
3-2
36
[30
form
(fi
wherein ^i
<^p
. . .
equation
of
^j
...
it
in
consecutive points.
likely also to
it
in
a, is
</>,
Q1
fix, y)
wherein
{x,
y\
is
K+
(x,
y\ +
{x,
y\ +
{x,
y\ + {x, y\ =
0,
.fiT is a constant.
Interpreted as a curve, this
equation represents a general curve of the fourth degree
it will appear
subsequently that the general integral of the first kind is
;
/.
where
(jr)
= 9//a2/,
and A, B,
we
recall the fact that n^{x), ...,n,p{x) are differential coefficients of integrals
of the
first
A,n^(x)+...+Apnp{x)
may be
line.
Take now a point of inflexion of the quartic as the place a. Not every
straight line there intereecting the curve in one point will intersect it in any
other consecutive point
in
tive points.
Hence it is possible to form a rational function of the third
order whose only infinities are at the place of inflexion in fact, if
;
Ax + B,y+1
'^{A^ + B,y +
1)
+ ,i(Ax + By +
1)
=0
hand
the left
namely
"^
A^ + By + 1
'
GEOMETRICAL THEORY.
30]
is
37
is
immediately obvious
it
infinite.
Ax + By +1 =
first
Q-q=p-{T + i).
Q = 3,
wherein
requires t
+1 =1
now by
drawn
to
In
of the fourth.
fact, if
A^ + Boy + 1=0
and A^x+B^y + 1
is any other
'\x + fiy-'t v =
general
forms,
their
most
in
osculation
A^x
+ Bjy +
\x + /Jiy +
\,
/*,
at the orders
1,
2,
(Cf the
5.
subsequently become
where
S=
\b
any
\S
form
and
X,
fi
-)-
satisfied
fiT,
where
is
any
by
Q = 4,
The present
given
5 = 1, p = 3,
article is
appear to find
-f 1
Q-q=p-{T + \)
\s
clearly
= 0.
The
the examples
FUNCTION OF ORDER
38
find
them
desirable to read
it
[30
31.
now what
we can form
For such a
determinant
A=
n,{x)
Dn,(x)
flp(a;)
n^ia;)
,Dl,{x)
,Dnp(x)
must
vanish.
Assume
at that place.
for
It is clear
by
28 that
vanish
we
are considering.
Let
Hi (x)
and
We
kind.
first
can write
the form
-J- in
at
-=-
/dvV
d'v dvi
r^\
d%
/dvy
dPVi
dvP
is
dv"-'
dVi
XT
di
4)th order,
its
Hence
dif
r-
'
is
d%
d'-Vi
is
-2
2p
it
and
is
infinite only at
is
and a similar
state-
D is a rational
(2^ - 2) (1
and
'
so write
where
(4p
similarly put
nn
and
-^
at dv
+2+
number
...
of zeros of D.
(p
1),
number
^2]
Now A
39
(^
" ""^^
.
The
or by the
the poles of D.
We
shall see
many
as
(/)
infinity
3^
common
have
Moreover
+ X,DD.i (a)+
+ iJ^Dli{a)+
Xifli (a)
ti^D,i{a)
namely,
and
^ would both be
The
0,
condition that
this
surface
Therefore
when
^+
1.
of the
1, 2, ...,
The reader
function of order
1,
p -1
exists,
not exist.
32.
When
first
kindj,
is
the curve has a double point, there are only two integrals of the
and p is equal to two. Thus in accordance with the theory above,
there should be (p
- 1) ^
(p
+ 1) = 6
we can form
functions
may compare
Forsyth, p. 395.
[32
EXAMPLES.
40
any
line
In fact
if
six
tangents
A^ + By =
be the
be the equation
fi (A^ + Boy) =
Ax + By
Ax + Boy
gaps occur
for the
The quartic with a double point can be birationally related to a surface expressed by
an equation of the form
')^
The
on the hyperelliptic
section
1)6 >
case,
I.
and the
Chapter V. below.
Ex.
33.
= (l>
f(x, y)
where the brackets indicate general integral polynomials of the order of the
equal to 4, and the general integral of the first kind is
suffixes,
is
/
where f{y)
= J-
of the 4th order exist, infinite only at these places, the following equations are satisfied
2y7y-3(y'/y)^=o,
3^
'''''
^XZx^J'
where
y = ^,
Explain
how
Enumerate
Ex.
2.
etc.,/,
= ^,
"^
h^y
Zy-^^^
f'""
^^ "^ ^
^x-^J"
8^2 f"^''' ~ If J
-^"^
'^dx?iyJ''J''^'dx''^''
'
etc.
all
* Here the
number
+ My) If {y) (cf. Chapter III. 28), which vanish in two consecutive points at the point of
contact of AgX + Boy^i), is clearly 1, or t + 1 = 1: hence the formula Q-qp - {t + 1) is verified
by Q=2, q = l, p=:2, bo that the form of function of the second order given in the text is the
(L.x
most general
possible.
34]
We
34.
now
pass
to consider
whether the
(p-l)p{p +
41
l) zeros of
V=
o.ii^ix)
,a^{x)
xni(*.-i)(f),
,ii/'-i)(f)
wherein fiW
let its
value be denoted by
X,o)i(a;)+
so that
Then
Ur=
Wr {x) dtx
For Mr (a;)
is
to the (A:,
n,{x)
,a^{x)
,
/*'--(?)
fii'*'-^'(f)
l)th
n/'-(f)
n/'-'-(?)
/*'- (a
(A;,
kind.
l)th order.
^/*'-"(?)
now the
first
the determinant
= (-)
V,
+\u3j,{x),
an integral of the
is
<,() vanishes at
...
n^<*'-i-i)(f)
,fi/'-^>(f)
numbers k^, k^, ..., kp. But all differential coeffiV, of lower than the (kr l)th order do vanish at ^: some, because
^ they are determinants having the first row identical with one of
x=
the following rows, this being the case for the differential coefficients of
orders A,
Ajj
1, ...
others, because
is
a linear
when
fi
function
is
of
),
...
for
i)*'~ini(f),
In particular
eui(a;)
those of
i)*2""' ft,-
linear
1,
k^,
ki,
The
all
II. 21),
V,
may
also vanish at
that there
first
is
no place at which
results in 34, 35, 36 are given by Hurwitz, Math. Annal. 41, p. 409.
be useful subsequently.
f.
They
will
42
An
important corollary
function
only 2p
a>p
2.
can now prove that if k^ > 2, the sum of the orders k^, k^, ... kp
than p'. For if there be a rational function of order m, infinite only
is less
^,
For
than 2p.
f, is less
kpl<.2p
times, namely,
We
35.
at
is
exists, infinite
[34
otherwise
function of order
function of order
Let
Vi
The powers
k^
?)i)
1,
2, ...
also one of
is
would be an existent
kp are divisible
the modulus
for
m,
ri, Vi
occurs
are existent
by m.
k^ ... kp which
by the remark just made,
gruent to {(<
Ti
...
kp, r
...
r.
none of
functions, hence
ki
among
and
kj...kp,
all
m
Vi
all
then,
2m,
...
m+
congruent to
con-
i,
i for
the modulus m.
Take
is
Since
in turn equal to
m-1.
so that
Now
r-i
the
sum
+ r^+
...
-1-
r,_i
= mp \ m(m 1)
= I i (2^ m+1).
is
m-l
2 [n + in - m) + (Vi - 2m) +
...+i],
1=1
which
is
equal to
1 m-l
^^(ri + m-i)(n +
i)
= ^^n[ri-(2p-l)] + J-'S.ra2p+m-l]
+ im(7K and, since
Sn = ^ m (2^ -m + 1),
1)
-^ (m - 1) {2m -
1),
this is equal to
[4p
- (m -
1)=]
-(-
J^
(m -
1)
(m
-|-
1),
P'-^'^ri{2p-l-ri)-i(m-l){m-2).
oi-
what
follows.
and similarlv
36]
> 2.
2 then also k^
it is
m is
than p^ unless
present Article
Now
1 or 2.
??i
2p-l
r;,
and
if
is justified.
An
p\
is
cannot be equal to
is
43
example
is
furnished
by
it
is
1, 3, 5, ...
2^
- 1,
whose
32 above.
Ex.
which p = 3, there
whose sum is ^^ - 1.
for
We
36.
is,
have in
at
34 defined
(Oi(x)dtx,
place,
1, 2,
by means of a place
o)j,{x) dtx
of
kind.
Wj (x)
...
fli (x)
...
terms of
i()
Dx(Oi{x)
-Dr^i(^)
(>>p{x)
Bxa)p(x)
I>r'o>p{x)
where
of
(k,-l) + (k,-2)
Thus, the determinant
to the
sum
it
...,p-l, p
first order.
its
zeros to
an order equal
For example,
1, 2,
We
Ip (p + 1).
lp(p
below, on the hyperBurkhardt, Math. Annal. 32, p. 388, and the section in Chapter V.,
elliptic case.
KIEMANN-ROCH THEOREM.
44
Hence, since
...
<
fcp
p=,
an order
Further,
i
ifci
[36
-nh,
if
from
and
that r
> 2p +
2,
1)>
or
2p
is
equal to 3
We
37.
there are 12
tained in the most general rational function having simple poles in distinct
places ( 27) and to the
number
stants
may be
in the
28)
in this enumeration
is,
becoming
infinite in the
whose order of
infinity
Let Oj, ttj, ... be any finite number of places on the surface, the infinitesimals at these places being denoted by ty, t.^,
The most general rational
function whose expansion at the place a; involves the terms
whose number
i_
Jl
J_
ti''-'
if-'
ti'-'
is finite, = Qi say,
and no other negative powers, involves
q-\-\ linearly entering arbitrary constants, of which one is additive, q being
Q-g-=p-(T-|-l),
where
is
the
sum
of the
numbers
Qi,
and t
-I-
1 is
the
number
of linearly
EXAMPLE.
'^'^l
which
45
whose
A,D^-^ SI,
(a.)
A,I>.-^ n,
(di)
+ A.D'^-^ n, (ai) +
number
total
. . .
. . .
of
from
clear
regard
in
fully
0,
infinite
an aggregate of functions
as
allowed,
is
and
it
is
to the
functions
as
given by
+ Apl>-^ n, (at) = 0,
As
is
is Q,
places a
a^, ... it
may
of further information in
quite well happen that not one of these
Q-q=p-(T+l)
The equation
Riemann-Roch
Theorem.
Ex.
For a rational function having only simple poles or, more geneXi, jjn, Vi, ... for any pole are the numbers
1.
rally,
1.
2, 3,
2p
...
Q>
if
Qi,
2p
of zeros
2,
+1
is
zero, since fi
(a;)
Qp + \
number
constants,
dV
T^
where V,
if
or
Q = 2p
3,
the aggregates
negative
if
Qp +
T+
and therefore
Q = 2p 4, and
vanishing aggregates fl
we have already
is
(a.)
^3 2
constants,
two of
or
in which case 2p 3
is
1 constants,
called hyperelliptic
on 2p 1 independent moduli.
involve p Z constants.
2.
1,
be equal to unity
of a special character
fact only
Ex.
would be
1 (x)
Q p +
the surface
or
constants,
fl (x)
namely p
first
and
depends in
definite place
2^-2
zeros of
dFand
to
the second order at a; the general form of such a function would contain
p+l=j9 + l arbitrary constants. But there exists no rational
2p 2 +
function infinite
to the first
to the first
order at
IMPORTANT EXAMPLE.
46
the place a.
of
[37
2 + 1 ^4 l=p
arbitrary constants
the places
a^,
...
a^^^ are
but the
coeffi-
For when
of dV.
all
A,n,(x)+ ...+ApD.p(x),
the conditions that the periods of an expression
\ + \i
be
all zero,
rs,
...
+ \^, Tl^^_^ + /. rs
X, n,-
(tti)
. . .
0,
(i=l,2,
..., p),
lead to
/j,[A,n,{a)
and therefore to
...
/i
+Apnj,(a)]=:0,
= 0.
\V+X,V,+
wherein V2,
number
....
...+\pVp,
of constants
and
is
dW
first
a-i,
...
aap.,
23,
a^,
...
a^p^^,
a, ... is
1,2,3,
may
It
...,p-l,p,p +
2p-2, 2p-l,
simple poles in
term -
l, ...,
a,,
a, ...,
is
2p,
2p+l,
....
a.
Ex.
each of
3.
these places,
these places.
is
rational function
R-c being
zero of the
c at
order at each of
first
- c)
is infinite
at
38]
CHAPTER
Specification of a General
38.
integrals
IV.
Form of Riemann's
Integrals.
The
Riemann
natural to the
theory.
surface.
II.
or,
the
more pre-
present the
anterior to
Riemann
surface
may be
regarded as
Conversely,
when the
surface
is
72
ffi).
may
is
It ia convenient to explain beforehand the nature of the difficulty from which the
Let the equation associated
44 of this chapter has arisen.
theory contained in 38
with a given Riemann surface be written
wherein A, A^,...,
A^
considered
are all
48
[38
Thus when
in x.
value of x, and will, therefore, be integral polynomials
divides all the other polynomials A^,
A
y
form of the
Conversely, when A divides these other polynomials, the
A^.
j^^
of
x, and is therefore
value
finite
any
for
infinite
become
cannot
equation 'shews that y
for
any
is
finite
an integral function.
an integral polynomial
is not an integral function, we can always find
finite poles of y, that ^y is an
of
the
each
at
order
an
to
such
vanishing
X,
|3,
Then also, of course, ^y', ^Y,...a,Te integral functions: though it
integral function.
such that ^^f is
often happen.s that there is a polynomial /Sj of less order than p%
order than ^,
an integral function, and similarly an integral polynomial 183 of less
of y.
powers
higher
similarly
for
and
such that 03^3 is an integral function
in
When y
say
In particular,
and
J)
is
if
an integral function.
is
clearly,
also
equation
y3 + ByH-+Cyx''+D:>^-E iy^-x^) = 0,
y=x+\x'^+...,
may
y=-x+iJix''+...,
y=E+vx+....
Thus, the rational function {y^-Ey)jx is not infinite when x=0. Since y is an
integral function, the function cannot be infinite for any other finite value of x.
Hence (y^ - Ey)lx is an integral function. And it is not possible, with the help of the
equation of the surface, to write the function as an integral polynomial in
x and
y.
For such a polynomial could, by the equation of the surface, be reduced to the form
of an integral polynomial in x and y of the second order in y and, in order that such
a polynomial should be equal to (j/'^-Ey)lx, the original equation wotild need to be
;
reducible.
Ex.
r]
= l^^ Ey)lx;
and
We
39.
concern ourselves
rational functions
x-a =
The
i"''+i, ...
x-a = <i"t+\
such a function.
39]
Since this
of X.
is
We
49
X.
is
prove that there are rational functions \, ..., hn-i infinite only at
a, such that every rational function whose infinities occur
the n places x =
+ CTi +
\, Xi
is
greater than
o-
+ 1.
way
1,
. .
We may refer
+ cr_i +
denoted by a symbol
(i?i,
R^,
...
certain polynomial in
is
call
...
Ri, R^,
unbranched at x
is
...
= a,
may
be
possible to find a
it is
(i2,,
We may
function.
iJ),
=a
is infinite
i2...,i2)-f
\X d
l)^
",
= (-SfS,,
...,S'^0)say
(i),
/-"n
Consider then
all
a;
a,
of which the
sum than p +
1,
+ 1,0,
...
0),
3rd,
first index can be reduced by increasing the 2nd,
smaller
with
In constructing the successive functions
consider
...
first
(n
how
far the
l)th indices.
index,
it will
be
which this
the
sum
be done.
shall
But
if
The proof
is
we
( 24, 28).
[39
SPECIFICATION
50
possible for which the
index has a
first
^ven
There
value.
will therefore
the
only be a finite number of functions of the kind considered*, for which
that
such
possible
least
is
the
index
further condition is satisfied that the first
it is
functions of the
is
classand in general
first
less
than
its
second class
than the
first class
let
when
its first
index
is
greater or not
all
first
the functions whose second index has the least value consistently with this
condition be called the reduced functions of the second class; let their
number be
class {i
<
h^
r^.
greater than the preceding indices and not less than the succeeding indices.
Let there be ki reduced functions of this class, with ith index equal to
Clearly none of the integers r^, ... r_i are zero.
r,.
Let now
n > Si,
where
...
r-j
s_i 0),
...
n > s_i,
Ri> Si,
Then by choice
{Si... Si-i
...
...
5i_i
Ri
class other
Ri> Si-i,
Si+j
Sn-i 0)
...
Ri >
Si+i,
Ri Si+i
. . .
S_i
Make a
0)-\{x-a)
...
Ri> Sn-i.
\ we can
write
Sn-i 0)
in the form
{T,...T^,RiTi+,...T^Ri-ri)
where
R^ <
Ri] 2\
may be
(ii),
and similarly
T^,
...
Ri
(n -
s^),
but
is
than Ri;
while 2\+i may be as great as the greater oi Si+i, Ri (vi Si+i), and is therefore not greater than Ri; and similarly Ti^^, ... T^^j are certainly not greater
,
than Ri.
* Functiona
is
way.
may
involve a certain
number
identical.
Of coarse the
of arbitrary constants.
By
the
here meant any one such, chosen at pleasure, which really becomes
OF A FUNDAMENTAL SYSTEM.
39]
Further,
(x
{-^, i)d
if
a)~^, we
51
can write
= {S\
. . .
\x
S'i^i
R"i
/Bi - I'i
S'i+i
. .
S'n-i 0)
(iii),
where R"i may be as great as the greater of R'i, Ri - n, but is certainly less
than Rii S\ may be as great as the greater of Ti.Rt n, but is certainly less
than Ri; and similarly (S'a, ..., /S"i_i are certainly less than ii,-; while /S'i+i
may be
than Ri
and similarly
Hence there
8'^+^
)S'_i
iJj-n, and
(1)
namely,
is
> S
R"i
...
R'U
>
S'i_ R'U
is
> S^i,
of the
still
-R"i
class,
ith.
'S'_
and in this case the greatest value occurring among its indices (R"i) is less
than the greatest value occurring in the indices of (S^... (S,_i Ri Si+i ... Sn-i 0).
occurring
the
first i indices.
among
the
function,
{S'l
R"i S'i+i
S'i-i
. . .
. . .
S'n-i 0) is
(S'l ...
or
of the
it is
that
is,
to
an
if (ti
if_i Ti ti+i
where r'iKu,
...
to a
more advanced
class,
The function
. . .
number
fj+i
tn-\ 0)
. . .
s_i 0),
. . .
may
we
can,
by choice of a
. .
s_, 0)
(iv),
ifn-i is
{t\
r,-
write
(i,
less
than
Vi,
while similarly
greater than n.
Sn-i 0),
Sf_i rt Si+i
tn-i 0) - ^ (Si
= (t\...t'i-.,r'it'i+,...t'n^^O)
t\ ... t'i-^
<'i+i, ...
U-^
. . .
no one of
. . .
pairs
. . .
{i
(<i
possibilities
among
. . .
its
t'n-i
its indices
may be
as great as
r,-
(and not
42
EXAMPLE
52
than
(s,... Si_i
for
certainly of
n Si+,
Denote now by
2nd,
...
(n -
(iii),
Namely
s-, 0).
more advanced
(ii),
. . .
class
[39
than
(si ...
(t\
. . .
. . .
t'^-,
0)
is
and not of
this.
h,,
l)th classes.
(iv),
Any function {Si ... Si-i RiSi+j... Sn-iO) can be expressed as a sum of (I)
an integral polynomial in {x - a)-^, (2) one of hi, ... h^-^ multiplied by such
a polynomial, (3) a function F which is either of lower dimension than the
function to be expressed or is of more advanced class.
,
the
to be expressed.
/=
(/Si
. . .
as far as
Si-i
Ri
Si+i
may be
. . .
Sn-i 0)
+(-^,l)
(-^..l)
A, +
\x a
\x a
/Xj
/\
where F,
is
+ f-i-,
\x a
...
l)
h_,+F,
(v)
/x_,
Applying this statement and recalling that there are lower limits to the
dimensions of existent functions of the various classes, namely, those of the
ki + ... + kn-i reduced functions, and noticing that the reduction formula (v)
can be applied to these reduced functions,
f={Si
tion
...
Now
Si-i
Ri Si+i
\xa
J\
we
. . .
/x,
\x
/x_,
to
is
applicable to equation
(i).
way
...
iJ)
infinity is less
functions
is
smaller than
wij
or
m,
TOi
+ ma + TOj = p +
1.
40]
class require
TTii
tjIj
53
+ mj + ma = p +
1,
mi>mi, m^^m^:
those of the third class require mj greater than mj and m^ but otherwise as
small as possible subject to mi + vi2 + m3=p + 1.
therefore immediately
We
obtain the reduced functions given in the 2nd, 3rd and 4th cohimns of the
following table. The dimension of any function of the ith class being denoted
by
o'i
o-j + 1,
+ o'2 +
o"3
the values of
in the sixth.
o-j
The reason
sum
54
{M,M+1,M- 1,
0)
When p = BP we
(Jo)
{P + l,P
Ex.
I.
Shew
for
obtain
+ l,P-l,0) = \}H+AiP,F+l,P,0) + B
=-\h, + A {fih, + GJh + D}+B.
[40
42]
55
0-,
1, ...
\n-i
+ O'n-i +
1,
of the function.
r,
form
...,
\_i attain the upper limits prescribed by these conditions, is a function of the
desired kind.
(r
is
or
. . .
arbitrary constants.
p +
is
proved.
The result is of considerable interest when the forms of the functions hi...h_i are
determined algebraically, we obtain the deficiency of the surface by finding the sum of the
dimensions of Aj...A_j. It is clear that a proof of the value of this sum can be obtained by
That theorem and
42.
fact, here,
specified
x=
<x>
let
fact,
by k equations
rational function
g which
is
infinite only
/>
+1
such that
a;~'''+''
it is
is finite
+1
=
x cc
clear that p
at every place
the
is
We
the dimension of g.
0)"'+^
( a)^-!"^' hn-i,
hi
,
a;
oo
o-_i
are
oo
Denote (x
it is
made
all finite at
the places
a)''i+^
hi
by
gt.
x=a,
If hi
40 (a), nnless /x, \, A' be specially chosen, the right hand represents a function with
index equal to ilf + 1.
its
third
56
[42
cr;+
1,
less
than
less
dimension than
o-^
if
hi
>
o-j
same
result
pi.
Hence, in
hi.
fact,
<Ti
= pi.
exist
The reader
is
will
is
paragraph.
Let now
a;-'p+i)
^is
is
a; = a
if
a)~
F
<t
of
(a;
'''"'"''
By
39 we can write
+f^-,i)
\x-a J^^ K+
(^-a)-<p+')if=(-!--,i)
\x-a
Jx
where
and
a- -{
therefore,
S\ +
o-i
+ f-^.i)
\x-a'
J^^_^
^711
+ l,
a fortiori,
p
Hence we can
also write
(x
- af-K-r"^-,
(x
a)"-^'-"-. g,-\-.
g_^.
or say
F^{l,x\ + {\,x\^g,+
where
/^i
namely, there
of
+ Pi +
=/>-o-i
+ /3i+
+ {\,x\,_^g^
1 =p + 1 -(o-;- pi)5p +
is
(B)
1,
a.
The set (1, g
will
, g_^)
for the expression of rational integral functions*.
It can
+ Pn-i=p.
of
be called a fundamental
Verand. Crelle,
e.
set
t. 92.
Kronecker, V. die Discrim. alg. Fctnen. Crelle,
arith. Theor. d. algebr. Grdssen, Crelle,
t. 92 (1882).
t.
91.
Kronecker, GrundzUg'e
43]
which
r
is
+ 1). And
It
ai = Pi.
the equations
(wjc
that
x=<x>
taken with
{x 6)~'''i+^'
1,
g^,
same dimensions
as 1,
^fi, ...
>
gn-i
g^
+ l),
suffice to
p,-,
...,
shew
we can
gn-i*-
43.
o-,-
r(wi
respectively to orders
%p = Xa = p,
57
is
{x,l),
+ (x,l\G,+
+ {x.
(C),
!),., Gn-^
For
Gi
in X,
(ic,
1X,G,+
+ (x,
1)^.^ (?_!
0.
if
...
IV +
which are integral functions when /3i, ...,/3_j; are proper polynomials in X, can be expressed linearly in terms of the nl k linearly
independent functions occurring among Gi...Gn-i, with coefficients which
By elimination of these n lk functions we therefore
are rational in x.
y3_i y"~*,
obtain an equation
A+A^y+
+ ^_iy"-* = 0,
whose
is
coefficients
A, A^
irreducible.
(b) Consider two places of the Riemann surface at which the independent variable, x, has the same value suppose, first of all, that there
Let X, Xj,
\_i be constants.
are no branch places for this value of x.
:
Then
cannot
\:
-I-
X_i u^Ti
all
values
of \,
\l-l-
For
this
fh(Gi
+ Mn-l(<?n-i - an-i).
Math.
of
- tti) +
Acta
For surfaces
109, 111
[43
GENERAL PROPERTIES OF
53
the function
is
not infinite
at
It
is
an integral
therefore
rational function.
Now
this is impossible.
the form
{x, l)>.
and
it
is
contrary to
+ {x,
what
1)^ G,
+ {x,
1). G,^,
is
\ + \G,+
+\,i_, (?_,
can have the same value in each of two places at which x has the same
value, is disproved.
x=c,
and no
+ \i~i (t_]
\iGi+
cannot vanish to the second order at the branch place, for all values of
Gn-i can vanish to the second
X_, namely, not all of (?],
\,
,
find
an integral
{H^G,+
More
Riemann
the
lias
at
the
A
Ex.
to
1.
all
of Gi,
may be
present for x =
c,
same place
it
is
of
for which x
be a branch place, they cannot all vanish
,
same value
and
if
by K, and its values in the n sheets for the same value of x by ^O, ^P), ..., ^("), we
have shewn that, for a particular value of x, we can always choose X, X,,..., X_j, so
that the equation ^(i) = A'(2) is not verified.
Prove, similarly, that we can always
choose X, Xi,..., X_i so that an equation of the form
TOifi'U)
where
m,,...,
mi_i,
rrit
+ TOj^P) +
, . .
+ mj^KV')
whose sum
0,
is zero, is
not
verified.
FUNDAMENTAL SYSTEMS.
43]
Ex.
59
2.
possible to
is
it
finite in
fi,,...,
/*,
at the places
x=y^,
places
be
shall
and
all
all different.
Hi, H^,
-Hji-i be another fundamental set of integral
functions, with the same property as 1, Gi,
(?_i, we shall have
linear equations of the form
If
(c)
1,
1=1
-ffi
where
Oi, j is
= ai + ai,i(?,+
an integral polynomial in
+ aj,_iG'-i
(D),
x.
1,
H<^\
we
clearly
ir_,<Mr_,^
2, ...,
a general value
1,
J
ir_, (")
for
0,
,ir,<)
ifi'^'
(i=l,
fi,
,1
1,
"^n 1
>
"-
"n
n 1
we form the square of this equation, the general term of the square of
hand determinant, being of the form Hi^^'Hf^ +
+ Hi^''^Hf, will
be a rational function of x which is infinite only for infinite values of a; it
If
the
left
is
x.
We
shall therefore
have a result
A
where
is
(1,
If
/3
the determinant
function,
^-0 = V= A
Oij
\.
(1,
(1,
(?
H,
Hn-i)
may be
called the
say, is
an integral
Hn-i-
written instead of
1, iT,,
values of x
all
all
But because
values of
1,
x.
H-i.
it
Since then
follows that
1,
(1,
r),
(1, G,,
A
where Vj
is
(1,
values of
77""'
17',
rf,
G^
are
rf-^) does
G_i) does
H^, H^,
all
rj,
ij,
set in terms of
(?_0,
x,
(?-0
A (1, IT,
and that
= V,^ A
(1,
H,
Hn-,),
only.
which
,
all
if_i)
FUNDAMENTAL SYSTEMS.
60
Hence V^
Hence
Vj^
thus each of V
and V,
an absolute
is
d,
(1,
[43
of
(r-i)
constant.
in terms of
all sets
which integral functions are thus integrally expressible, are identical, save
for a constant factor.
Let
denote their
functions whatever
common
then
if
value and
(17,, 172,
wherein
is
(s,
an integral polynomial in
The
x.
follows,
if,
for shortness,
we speak
n integral
of
1,
Since this
i/i,...,
Vn, as a set of
to the
The
sets
(1, G,,
functions, tliat
function
'?2,---,
(d)
r^-^,...,7)n
is
r}^,...,
Vn)
77,,
common
highest divisor
-^n-i)
is
by A,
divisible
is the
(771,
not supposed
are
terms of them, no term shall occur of higher dimension than the function to
be expressed.
condition
is
gn-i
g^,
If (1, ^1,
9n-i)
be a fundamental sj'stem
for
which
this
satisfied,
will
{x,
this is contrary to
\\ + {x, \\
what
is
sum
sum
is that
which
is
of
gjc
is
gu
g^,
Gi
= 0;
(a).
G^-i
Hence,
is
We
-\-{x, 1),^
proved under
of the dimensions of
(?,+
p + n \.
last
Chapter a definition
sum
which
(1, Gj,
of
^-i).
is
G_i)
namely,
p founded on
of the dimensions
of^'i,..., fifn-i
Ex. 1. Prove that if (1, g^, ..., ^_,), (1, h^^ ..., a_i) be two fiindamental sets
both
having the property that, in the expression of integral functions in terms of them, no terms
44]
61
occur of higher dimension than the function to be expressed, the dimensions of the
same as those
other
set,
Ex.
2.
satisfies
= (3f2- 6y + a, c)/ai
the equation
1)^
and that
A
^ (1>
y.
<A
/) =
In general
1,
y,
tj
(1,
y,
t;)
^ (1.
(1. y, V)
V, 'f)
= "-i A (1,
y,
A (y, f,
r,)
ri)
= ai^C^ A (1, y,
r,).
44.
Let now (1, g^, g^,
gn-\) be any set of integral functions in
terms of which any integral function can be expressed in the form
,
{x,
and
the
let
sum
\\ +
{x, 1)^,
+ {x,
g,+
of the dimensions of ^i ,
l)^_j 5r_,,
^_i he
p + n l.
^ly^
9i
where
_ H-i,n-i y"~'+
Mi.n Mi.
fH,n-\,
expression by gi
+ Mx,i
+ Mi
-^_
and
let xt (y> ^)
oo)
= Qo y" +
Let
)(^
(y,
^o',
x)
Denote
Q, 2/"-'
to
be an integral
+ Q.-.y + Qn = o,
f + Q, y'-' +
+ Qi-.y +
Qi,
is Qo-
<^i'.
>
^'n-i
in the identity
4>o'
'i>i
this
so that
x.
(y, x).
-91 {y,
* Since
p,,
...,
xn-i
(y', a;)
ALTERNATIVE DEFINITIONS OF
62
[44
,2/""'
^) be
5'-i (y-
in terms of
= 1,
y = Ol + a^^l
1
2/"-!
where the
= a_, +
atj
coefficient
</)'_,
So that
^r,
ffn-l,
a_,, -i 5r-i
Ol,
Xn-2
-l
(2/'.
^)
+ n-i. n-l
Xo-
write
(1, y,
a_,,
we
if
is
+ !, n-i
ffl
r-) = ^
y\
.9n-r),
(1. 9>
where Xi = X'
columns of fl,
But
if
(^'> *)>
its
^n-2>
>
Vli
Vo
Vn2>
Vn3>
>
Vo)
Qi,
Qo,
Qo.
0,
0,
we have
(X"-i. Xn-2
Hence, changing
3/'
to
(^o, <^i
Either
this,
4>o
Xi. Xo)
y in
<^i'
-/"n-O
= (Q)
')
/(y'.
+ <^i
iV' ^)
^)-/(y
=n
i'. (2/.
2/"~0-
2/' S'"'
(1, g,.
W'
(1.
^)
is
</><
for
(^j',
we may
^_0
write
(E).
equivalent to
^n-i
iy',
^)
s^"-! iy<
^)
y'-2/
= xo
= Xo
may be used
The
y'""'
y"~'
+ y'""' xi (y. ^) +
+ y""' Xi (y'. ^) +
latter
form
will
immediate application:
now be
let y^,
t^^t <^i.
(F),
^ti-i-
44]
to
Then putting
Hence
for
gi^''>,
in (F) in turn
y = y' =
+ <^.
5'i"'
^o'"
^,''1
if,
distinct.
Denote
etc.
<^o'^'
<^i
63
y^
and
= y^,y = y^, we
y'
Ci 9Z, = (|)^_
<^l'li 'l'!i
= 0,
obtain
=/'(2/.) ay.
(*=2,
Co, c,
3,
c,^,,
71).
we
write
we have
<^i'
/'
c,
C_i
sr,"-'
^-.'^'
^.""
^n-a'"'
^i"
S'n-i"
Co
Ci
(2/0
= 0,
(^'"
Co
9n-/-
^l'^'
.(G);
1
and we
shall
find
9n-l'
9^
this
we put
all
of
5-1"
(71)
Cj,
c_i,
S'n-i
it
clearly takes
an inde-
x.
except
Cr,
left
hand,
where,
if Sij
+gi^''' g/"-\
St^j
in the determinant
0;9i'9i'
9n-i)
= n
s.
^1, 711
^711
Since this
^,
fW)"
is
Sr + Sr,igi+
A(l,sri,
we
+ ^r,7-i
^711,
^?i 1, 711
^711, z
therefore have
ff7i-i
,f_i)
1
aA
A 9s,
9n_i
(H),
[44
INVERSE DETERMINATION
64
and
therefore, also
/' {y)9r
= Sr
+ Sr,.
<f>o
<^i
+ Sr. n^i
also express
4^
^1
fiyVfiy)'
For
let
determines the
(y)
But we can
_^
it
-.^ from
(H')-
^n-i
'f
'
from
gn-i so that they are determined
^ knowledge of these
^j-^^
only.
(y)
by
70, 71.
the definition of
s^
let
i,
a-r,i
=t
7^'*' 7t""-
gM =1
^^.
Sr,
Si, 1
Sr, n-l
Sf,
_i
'^ L
s=l
or 1 according as
=)=
r or
= r.
rr
A/
rv-l*'
=1
=
^
may be
7r
If then 2r,
(Tr^i
which
written
= Cr, + 0"r,
determinant we
5^1+
=1
4=1
S-1
may
<Tr, i
call
0"r,
in the
n-i ^n-i-
(7,,,
71,
>
7n-i)
^^
have, in
Of course V
Ex.
wherein
1.
= -^
Verify that
i),, ...,
and
if
2^_
<
= -r Sr,
i,
g-^, ...,
is
(K.)*
Sr, -i77._i)
...,
arise
by
immediately from
45]
Ex.
2.
2
=1
We
45.
^^^
places
(dv/(h:),
first
in
(i= 1
2,
...,-
1),
= 0,
kind.
a position
bo
Riemann
express the
^I'.x,
x^.
a;,,
now
are
= 0,
\.^ilf'{y)\
V being
65
a;,,
integrals.
be infinite like
w,
thus
(x
^
X,)
'
Similarly (x
+1
dP
dx
x^)
^+
t^'^'
^i""'-'
first
of
Wi
a;
-.
x^
let
= a + P+S dP/dx
is infinite
like
if
to
there equal
^
is
is
for the
all,
dP
a)-r-
and
x,^
at the place
f
finite
and therefore (x
Hence,
~y~ is finite at x^
are there
At any one
x^
-r~ is
Assume now,
nor x = X2
= {x Oi) (x a^).
be the integral polynomial which vanishes at all the finite branch places of
the surface, and g be any integral function whatever, the function
K = a.g.(x-x,)(xis
x.^)
^
values of x and vanishes at
Therefore the
sum
of the values of
K in
x.
Since
it
vanishes for
all
x,
x,
is
the values of
5
EXPRESSION OF INTEGRAL OF
66
by
zero, it is divisible
[45
a,
This
is an integral polynomial in x.
wherein Xj and \^ are constants and
is uniquely possible.
Let
be of degree ^1 in a; denote it by {x, ly-^.
'"
'
X X^
dxJn
X X2
namely,
Similarly X,
it is
a;
a;,.
g{x^,
is
dP _ _ J^ dP
dx
w + \ dt
so that x^dPjdx
is finite
at
.dP\
places
all
infer, since
if
be the dimen-
,dP\
_ x-^)
f?U
\dxJi
Hence
xi^^ {x
'
and we write
a;/>-i (a;
we can
a;
= <"<"'+',
0, t,
+ 1,
...
+f?U
\dxJn X
af~^
'
at most equal to p.
/x is
\, g^, ...
x-i)
t_i-I- 1,
'
Xj
Xi'
*.(S),+--.(S).-=^^-=^V(',"--"".
where
r\,
...
greater than
for f-3-j
Tj
. .
t_i respectively.
suffix,
45]
67
dP
dx
_^
<t><,
+ 4>igi{a;i,yi)+
_ ^0 + 'f>iffi(Xi,
where
This,
can have
is
</>{
a;
stands for
</){
2/2)
<l>n -i
/>i-ig'-i(3, ya)
+ T2+
{x,
the
\i
. .
.
is
any
and
'"
dx
undetermined.
is
(^P\
\dx)
dvp
^ dx
dvj
dPjdx
Now
iy~'
+T_,,
...
y,)
<
wherein
(a;, 3/).
T1
or
(a;,
ajj
gn-i
(i)
wherein
t'j
at
Xf
and the
kind
is
(:,
y)l
the two
Xi)/(x
dx
J
first
A special
(ii)
infinite
log [{x
of integral of the
rx
<f>o
(a^.
y)
<f>i
{',
y) gi
(a^i,
yi)
+ ^7.-1 {qq,
y) g-i
(a^i
yO
f
fky)
_ ^0 (^,
y)
+ ^1 (^.
y)
(^8. yg)
+ ^-
x,
y) g_^ {x^ ,
a?2
3/2)1
_j
or
p
I
dx
['',
d [ <f>o (x,
f{y)Jj^d^l
y)
fj),
(x,
y) g, (I
,,)
^^
y) g_,
']
(g, t))
J-
In the actual way in which we have arranged the algebraic proof of this
result we have only considered values of the current variable x for which the
n sheets of the surface are distinct: the reader may verify that the result
valid for all values of x, and can be deduced by means of the definitions
is
of the forms
<f>o,
..., <f>n-i,
(G).
Ex.
to obtain the form of the general integral of the first kind only.
52
gg
We
[45
for the
expression
<f>,
(a;,
y)
+
by
it
^, {x, y)g,
is
+ <^-i
i'^'
V) 9n-i
(f.
Then the
{x, ?).
{^,n)+---
v)
an elementary
given by
=rdx[{x,x,)-{x,x,)\.
P^'"
periods at
This integral can be rendered normal, that is, chosen so that its
suitable
of
a
addition
the
zero,
by
are
the p period loops of the first kind
linear aggregate of the
Chapter
in
II. 19,
that
if
(x,
^^'
denote an elemen-
"
y) given
by the
differ-
ence
is
not infinite at
(f,
It follows from
i?).
depend upon
{x^, y^).
a function of (x, y). Therefore, if not indean aggregate of integrals of the first kind. Thus we
infer that one form of an elementary integral of the second kind, which is
Hence
it is
pendent of
nowhere
infinite, as
{x, y), it is
(f,
like
tj),
(x
f )~',
is
given by
"^
dx
>o (x, y) +
(f>,
{x,
(g, t;)
y) ffi
The
is
-I-
</>,._i
{x,
y)
when the
at a branch place,
which
is
given below,
infinity
another proof of
this result.
We
46.
one or both of
before, let o
branch places.
places
when
places;
all
and
the finite
The function
ga (x
Xy) {x x^ dPjdx
=
and though it may
a;
order,
first
a;i
it
vanish at some of
high
order as
(a;
a;,).
46]
Hence
we can
as before ( 45)
ax
\^
Ji
69
write
X Xi
ax Jn
x^
Multiply this equation by x x-^ and consider the limiting form of the
resulting equation as {x, y) approaches to {x^, y,) let ; + 1 be the number of
sheets which wind at this place.
Recalling that the limiting value of
:
(x
Xi)dP/dx
l/(w+
is
w+l
sponding to the
1),
we
w+l
see that
terms of the
left
hand, corre-
form
^j^
where
is
g{xi, yi)l(w
[1
+ l)th
(ly
+ l);
The
root of unity.
2/0
The
+ Dt"+..
Ct
of this
limit
g(xi,yj) as limit.
for
will
left
.],
when
is
therefore have
limit
/i
The expression already given for an integral of the third kind holds ivhether
(^1)
'
If
P^'
(^2> y-i) ^s
2/i).
it follows,
by
is
infinite at
a branch
place
lim-ix-*.
lim.
dx
=0
[(x,
/)
- {x,
xi)}
=iim.i^';.
.t-'
I, X,
,
Now
if
we write
for
and
tx^
a;,'
+ <f>i.(g, + tgi
+ P+S
a;i
P^\''
+ ... + ct>n-i
x-x,- '+i
+...)
(g,i-i
tg'n-,
...)
X Xi
wherein
<j>o,
D denoting
4>n-i
are functions of
y,), ...
x, y,
Dg,(x
gn-ii^i, yO,
a differentiation in regard to
<^i ffi'
t.
and
g^,
...
respectively,
+ </>-! g'n-1
*
are
is
EXAMPLES.
70
That
introducing
is,
f,
[46
tj,
dx
{"
(x, y)
<f>,
gr;
(g, 1?)
given by
is
<^-i {x, y)
ff',.-i
{^. rj)
^-?
f\y)
rf),
where g[ (f ;), ... are the differential coefficients in regard to the infinitesimal at the place.
It has been shewn in (6) 43 that these differential
,
cannot be
coefficients
all zero.
by a
when the
infinities are at
(1, 2, 3, ...); in
Shew
1.
[_^ rXi
'
{x.^,
y^)
can
form
00 [x, y) + SX^^r ^^
y) g^ {x^ , yj)
(^x,
_ X.,
'"r
'
(^p [x,
is
y)
+ iX^r ^^ {x, y) g^
the dimension of
^r,.,
{x.^
"!
,
and a
y;)
is
any
can in fact be immediately verified that the difference between this form and that
is an integral of the first kind.
Or the result may be obtained by considering the surface with an independent variable ^ = {x-a)-'^ and using the forms of
39
of this chapter for the fundamental set for functions infinite only at places x = a.
The
It
previously given
Ex.
2.
Ex.
3.
<^
by equations (H)
44.
Obtain, as in the previous and present Articles, corresponding forms for integrals of the second kind.
infinity at
Obtain the forms for integrals of the third and second kinds which have an
a place .r = oo
Ex.
4.
wherein
a^,
...,
03^ +
are finite
and
IS
5.
taken
{x-a.'2P+1A
different
Ex.
and
^ = 1,
0, is
1
idx
[-1
\_X
Putting
(a- yj)
a:,
= oo
y+2^-2^j ^ ^ y+^^^^-2^n
X-Xi
X
x-x^
i^ndy, = ,a:,''+,ix,
2/
1/
+ A + Bx,-i + ...,
{^-^2)^^2{X-X^)J
x~x^^x^
*
Chap. V.
56.
form
I,
when a
47]
Ex.
2/2),
namely
-log(^-X2),
like
if (:C2>
{j-\, 1/1)
(00
00
log(-| and
is
otherwise
where
1-
2i
otherwise
finite.
The
-x-^
it is
i/
y + wM-q
is
fdx
and that
I/2)
like
6.
fdxrj\ i/+.A\-'-yi
3s
x = a:
at one place
is infinite
71
is
It is therefore
this
place as
its infinity.
We give
47.
whose
process
now a
77)
we suppose
that {w
the surface wind at this place, and find the integral which
like
is
1) sheets of
there infinite
F be an
Firstly, let
so that in the
integral which
neighbourhood of the
is
F= -L--+A+Bt + Cf+...
Forming
as before the
sum
g being any
f )~\
form
term {x
integral function,
^y dF/dx
in
we obtain an
expression
s' (;
- ^?
1=1
Putting
x= ^ y/e
the place
(f, r))
5]
= X + M (a; - r) + (^ - ^)^
terms on the
left
(^',
y-'-
vanish, that
\ = -(w + l)g{l
Differentiating, and then putting x
belonging to the
f we
,
v)-
infinity,
-l+(<-^r-~(A + Bt+...)
the summation extending to (w
1) terms.
DIRECT INVESTIGATION
72
[4)7
Now
\a=-&^^{A+Bt + ...)
dx
vanishes
when
zero
t is
{w
^[r+H-B +
2Ci
lyt^ dt
+ ...)]
hence
^y
first
given by
..
dx
<!),
1)"'+'
dx
[^,
4>igi i^,
+ <^iffi (f
true whether
is
1?)
(f, tj)
of the expression a
is
^n-ign-ii^, v )
than
less
(x
If,
^)
is infinite
at (^,
i?)
w + l.
as before,
dEjdx, wherein g
is
of the
n values
like t~,
TYt
=-
is infinite like
...
^-^
Jf{y)
This result
v)
all
the finite
shall obtain
To
find \, let
w+1
g+
where
the surface at
expansions
(f,
t)),
and g
is
written for g
(f,
tj).
The sum
of these
w+
is
+ (w+l)Ag+{Ag' + Bg)'Zt+....
Now
in fact every
summation
"Zv,
being a
r-i-eT-i-...
wherein
e is
multiple of
6""+'"'r,
a primitive (w
w + 1. Thus
sum
48]
sum will
when < =
vanish
;
and
A log {x - f
terra
in fact
)
when
is zero,
(x,
y)
otherwise
near to
is
17).
(f,
E would
73
r*
^^;7in[ j
da;
^rr|
fjy)
t'^ C. 2/)
4>n-x (, y)
gn-i
{^,
v)\
+2
(a;, 2/)
<^r {x,
y)
ffr (?,
v)
by ^, an
integi-al
which
is infinite like
-"1
t
is
-"-tc
'"
an expression
-"If+ 1
x-^'
P"
given by
Of course the
7),
D(f)
(f
Tj)
in regard to
JSx. 1.
Hx.
2.
t,
ij
+ P(t),
that
in
<f>
must be understood in
rj) be any function of
for x, ^ + *"'+', and for y,
i;)
If ^(^,
(x, y),
we then
as evanescent.
P^'
Obtain by the formula the integral which is infinite like A/t + B/t'^ in the
Verify that the integral obtained
(0, 0), the surface being y2=j;(.rj i)^.
neighbourhood of
48.
gi, ...
gn-\)> of
is
x-'^i-^s,
X-'n-r'^Sn-i,
In
this
x-'i-^Si
x-'i"'^-^ s,2
A(l,
x-'i-'n-i-^ S,n~i
X-\-i-'i-^Sn-j,i,
^Tj,
...,
a;-^-,-is_i
x"'-'+9.
|
also a;~'^P~=+'"'
is finite
^'-i
'
^ Sn-i
n 1,71
at every place
x=
cc:
hence
Thus
74
A (1, ^i,
than
...
^_i),
2n 2 +
2p
which
an integral polynomial in
shew that
in fact easy to
it is
x=ac
if /o?'
A (1,
x, is
in x.
is
[48
are separate,
any value of
1,
W2+
1.
it is
not of
x = a,
x,
less
there he
....
43)
'!
i/i
1
!>
yi
>
>
li-l
iln-
<
1
I
n'"l
n*"'
yn-i
>
which A(l, gi,.-., g,i-i) is the square, can, for values of x very near to
be written in a form in which one row divides by t^, another row by
t,-, ..., another row by ii"', in which also another row divides by t,, another
row by t^^, ..., and another row by <"'', and so on.
of
a;
= a,
Thus
this
when
and hence
a)""' (x a)"'^
^^iw^ftc+i) __^
A (1,
g^, ...,gn-i)
=2n+2p2 ( 6).
In that case then A{1, g,, ... g,)r-\) is exactly of order 2n + 2p 2: and,
all the branch places occur for different values of x, its zei'os are the
,
when
When
all
the surface
is
branched at
and
x= a where
Then by putting
choose a value
infinity,
let gi
appropriate order.
to its
= (x
ayi*^
hi.
is
an integral
is
immediately
obvious that
A(l,5fi, ...,g.^,)
= {x-ar^+-'P-'A{l,h
...,/i_0-
A_i) be
f",
A (1, gi,
...
gf_i)
number
Ex. 1. For the surface y^=x'^(x~l){x-a) there are two branch places at .v = 0, and
a branch place at each of the places x = l, x=a, where all the sheets wind. Thxis
2re+2p-2 = w=2.
*
Chap.
+ 3 + 3 = 8.
II. 21.
49]
For
this surface
+l
g^)=2b6x'
75
POLES.
ffi
i/,
g2,=y'^lx, 93=y'^lx-
x^ corresponding to the superimposed branch places at j?=0, while the other factors are of
Ex. 2. The surface y' = .2;2(j; 1) is similar to that in the last example, but there is a
branch place at infinity at which the four sheets wind, so that, in the notation of this
As in the last example 2ra + 2/) 2 = 8, and 1, y, y-!x,i^lx are a fundamental
Article, s=3.
g.^,,
- 256.r' {x\f,
g^) is equal to
^+
independent places,
will
therefore
and
become quite
+l
of the form
AF-\-B,
definite
is
.F is
The function
there
g^,
49.
where
A (1,
poles
if
we
prescribe
the
we
^+
1 poles
a,
now
{z, s),
instead of
the function
dz
<f>o
+ ^1
{Z, S)
(g, S) ffi
+
wherein
at
(z, s) (z,
1)-'-'+
+ ^_i
(z, s) (z,
l)\-r\
Let now
ffn-i (z, s)
<}>,
+
a
Ci,
Cp
denote
{z,
iy^~^,
finite,
{z,
l)\-r^.
the determinant
A=
<f>,''\
</./"
0i"",/>i""cp
^a'" C,^-'
<^'Vi, </.'Vi C
,^i'^'c/.-S
.Z.""^,, ./.'^UCp,
</.<"_>
cA-r'
<|.<^'_iC/-i-'
wherein ^t^ is the value of (jti (z, s) at the place Cr, does not vanish. That it
Vp denote a
is always possible to choose such p places is clear: for if d,,
,
Chap.
III. 37.
76
set of
[49
first
expresses
/'()
involving only
the places
Ci
vanishes at
cients,
to the
[Z,
by
X,
Cj,.
coefificients in
Cpi
c^,
^ {x,a; z,Ci,
,Cp), so
Af'(s)
that
yjr(x,a; z,Ci
coeffi-
Cp) is
equal
determinant
X]
- [Z,
a],
[C, x]
[Cp. a;]
- [cp, a],
where
disposeable ratios \,
\
,
Choose the
dz
[z,
[c a],
Z<t>,
</.,"',
Ci
Cp
(f>,^"\
(/>!
..., cr--'
</.,<",
</>,"",
. .
c/.-^
</>,'",
<f>:"\
Z^-r'-
<f>n-i (z,
....
c,^n-r' 0SIL.
. .
c/n-r^
.,
s)
(/)</l,
{z, s)
ffi ( x,
y)
Suppose now that {z, s) is a finite place, not a branch place, such that
none of the minors of the elements of the first row of this determinant
vanish.
Consider i^{x,a; z, c^,
Cp) as a function of (x, y).
It is
,
and
is
It is infinite at
the limit of (z
z,
{z, s)
x)
-^fr
z, Ci, C2
(x,
of
,
z, Ci,
all
c
c^) is
s) gr {x, y)
f{s)
namely, unity
at
Ci, ...,
( 44,
F)
so that
Cp it is similarly
To obtain
its
&tx = z,^\ris
infinite like
behaviour when x
at infinity,
is
{x
z)-\
And
first order.
we
notice that,
by the
which
is
of the form
- ^-<^+') gi
{x.
r
y) yz^
^T.+i
-T.+2
+ f^ + f^ +.
"1
.
gi{x,y)_gAaJi) g^jx.y)
g^{a,h)
columns
,
,_,
>
-,
c^
FUNCTION WITH p
49]
+1 ARBITRARY
POLES.
77
the determinant will contain only quantities which remain finite for infinite
values of x.
a;
{x,
z, Cj,
c^
is
and
it
vanishes at
a rational function of
z, (\,
if
is so
chosen that
d, be denoted by
Let now
that of
is
(x),
a>i
zero at all of
z,
let its
Then the
a rational function of
c,-
Ci...Cp) is infinite at
(x,
manner of
and
c,-
is
like
unity at
'
X Ci
infinity at Zi
function
{x,
except
c,, ..., Cp
a rational function of
(x,y)-\yfr(jr, a:
however these
pdes,
{x z)~^
^^S-
then -^(x, a;
Xi (a: ^,)~^.
it is
R (x, y) be
ordinary places
p+1
^S+
which
having only
x,
It is infinite at z like
c^.
x = a.
It is
we can
see,
Cj,)-...-
y) which
is
\,ji//-(j",
a; Zg,
only infinite at
Since
no such function
nor
exists
does there exist a rational function infinite only in places felling among
Hence the function just formed is a constant thus
Ci, ..., Cp.
:
R{x,
2/)
= Xi i/r (x,
a; z^,Ci,...,Cp)+...+
\Q^{x, a;
z^, d, ...,Cf,)
+ \.
Conversely an expression such as that on the right hand here will represent
a rational function having z^, ..., Zq for poles, for all values of the coefficients
\j, ...,Xq,\ which satisfy the conditions necessary that this expression be
each of
finite at
Cj, ...,
Xj
where
= l,
When
2, ...
Cj,
tOj {Zi)
+ Xj <Bi
(^j)
+. .+ Xq
.
&),
(Zq)
equations
= 0.
p.
Q-P + i
arbitrary constants
(Chapter TTT 37). The excess arising when these conditions are not independent is immediately seen to be also expressible in the same way a-s before.
We
thus obtain the Riemann-Roch Theorem for the case imder con-
sideration.
The
function
function.
The
-^jr
(x,
a;
will
employed Chapter
m.
which
23.
is
necessary
when
78
some of
its
means
present chapter by
place
x=
<x>
becomes a
is to
39 of the
of the transformation
^ = m.
finite place
.50.
order,
[49
in this
in
a different
on an algebraical basis.
wherein a^
function of
By
...,
+ y.-i
a,+.. .+ yL/_,
+ a = 0,
is
an integral
x.
methods only
algebraical
it
can be
she\VTi
functions g^
can be expressed by
them
in a form
l\ + {x,
{x,
1);^,
g^+...+
{x, l);,_i5^,^i,
other set
which
in terms of
all
\\ + {x.
[{x,
If the
sum
p + n-\, pis
}h+...+
of the dimensions of
{x, !),_,
K.,]l{x, 1).
^Ti, ...,
The expressions
1),,
</>_!
(f>, (/>,
form
f
wherein
of terms
shewn
dx
< t etc.,
which enter
t'.
r.-
is
+1
most
at
t,
+ r_,
g^.
or p.
is
an integral of the
first
(^
fr-(y^'^'<l>i
is
everywhere
*
finite:^
provided
:f>
:^ t,
1, 2,
,-l)
1.
44)
OF THIS CHAPTER.
50]
79
as in this Chapter:
method
fundamental equation
is
explained in Chapter V.
given form of
( 73).
CHAPTEE
V.
in
many
We
59), is
is
;
the
first,
characterised
the
the hyperelliptic
by the existence of a
second, which
we
shall often
It may be stated for the guidance of the reader that the results obtained for the
second and third cases ( 60 71) are not a necessary preliminary to the theory of the
remainder of the book but they will be found to furnish useful examples of the actual
52.
We
We
is
consider
of the hypothesis
Let ^ denote
be any constant and a, /S denote the two places where
f = c, so that (^ c)" ' is a rational function of the second order with poles
The places a, fi cannot coincide for all values of c, because the
at a, fi.
rational function d^jdx has only a finite number of zeros.
We may therefore
such a function
let c
regard
. .
. ,
being arbitrary constants, each of the functions /i, /,, ... must be of the
second order at most and therefore actually of the second order by choosinc
the constants so that the sum of the residues at a is zero, we can therefore
:
53]
81
obtain a function infinite only at /9, which is impcssible*. Thus the most
general rational function having simple poles at a, /8 is of the form
Therefore, from the Riemann-Roch Theorem (Chapter III.,
(f -c)-> + B.
Q-q=p-(r
= l;
which vanish in the two places a, yS is p -1. Since a may be taken arbitrarily
and c determined from it, and ^p 1 is the number of these linear aggi-egates
which vanish in an arbitrary place, we have therefore the result When there
a function of
exists
another place
every
that
and
/9:
linearly
may
the detei'minatiun
independent
linear aggregate
ft (x)
which vanishes in
53.
independent
To
see this
l=p
^1
p1
let
H,
(a)
be other than
a,
zero.
Since
and, by hypothesis,
fl (x)
)3,
it
in
/9.
ai-e
at
linearly independent
vanishes in
n, (x)
/9,
aggregates
- Hi (a) Oj (x),
fii(a)n,(/3)-nx(a)fii(/9)
fl, (a)
n, (a;)
equations
= 0,
(i
= 2,
S,...,p).
fi,(/3)r:-n.(a)i1
has each of
a
and
the
/3
its
and
periods zero.
flj (/8)
Thus
it is
first order.
n (x) aggregates
are
number
linearly independent
is an
For any pair of places the relation is reciprocal, namely,
is similarly
if the place o determine the place yS, a is the place which
determined by /9: in other words, the surface has a reciprocal (1, 1) correspondence with itael It can be shewn by such reasoning as is employed in
infinite
vanish, there
By
the equation
Q-q=p-{T + l),it q
is
were
2,
+ 1 would be
p, or
aU
THE HYPERELLIPTIC
82
CASE.
[53
I. (p. 5), that if {x^, y^, (x^, y^ be the values of the fundamental
variables of the surface at such a pair of places, each of x^, y^ is a rational
Chap.
of x^ and y^,
function
functions of a^ and
x^,
y^ are
y^.
54.
accordance with the explanation given, for which the poles are so situated.
Denote
this function
by
Then* the
z.
is finite.
2:
Ci
is
fall
at
Xi would mean that z a is there zero to higher than the second order,
which is impossible because z is only of the second order. By the expla-
it
follows that
greater
Hence
number
there
is
r-
a linear aggregate ft
x^^.^,
il {x) vanishing at
= 0.
{x),
which
2 zeros.
p or any
The general
x^,...,
Xp+r
p+r
q = p,
oi q + 1 of constants given hy
1 constants.
Such a function will therefore not
number
when
no linear aggregate
which has
presciibed
...
rational function
exist
x^,
way does
^
X, \i,
...,
Let
= X + X,Si+...+X,.2r,
h, h'
value of h
infinities
function
* Chap.
t Chap.
+ Chap.
Chap.
III.
o{
I.
6.
III. 37.
m.
27.
37 of
55]
Since h and
z
\b
83
h'
Cj, ...,
is
+ h' = (Z,
Ci, ...,
hand
left
and
Cp+r
is
l)p+rl{z
-C)...(Z-
Cp+r).
at most, at any
first order,
zero to the
is
right
hand must
by the
divisible
From the
say.
equations
=X + \i2i
+...+v2r,
+ \.2'i+...+\r2V,
also a constant, = 201 say
/l'=\
we
then that 2,
infer
+ 2'i
is
for
h was chosen to
C = \ + \Ci+
We can
+KCr.
= 0:
that s +
in
s'
...
therefore put
its
....
...,
h,
such however
made
the
for
functions
..., 2r-
Consider
55.
ratios Xi
X2
...
places
Xr so that s
is
a, a'
at which z
zero to the (r
is infinite.
l)th order at
a.
Choose the
This can
always be done, and will define s precisely save for a constant multiplier,
unless it is the case that when s is made to vanish to the (r l)th order
then be other
itself,
p + r -2(r-l
manner of formation
this
+k), or
number
p- r+ 2- k,
zeros of
is
s.
From
Consider
the
now
the function
/=(2-Ci)...(^-Cp+r)s=.
At the
OT
places where z
finite;
is infinite
At
/ is
the places,
z- c
...,
infinite of order
coi,
...,
z -Cp+r
is
= - as', /
p + r-2{r-l + k),
Xp+r where s
is infinite, it is
a symmetrical function of
any prescribed value.
has
the values which s takes at the places where z
follows that the funcHence by such reasoning as is previously employed, it
it vanishes.
Since
s"
is
62
CANONICAL EQUATION.
84
[55
/ is
tion
at a or
fail
But
a'.
sum
since the
is zero, it
follows that s
is
2k must
p-r
2
in
z,
+
r
2
2k
+
Thus, as if is a polynomial of order p
and
zero,
k
is
Hence
2.
p-r
+
be at least equal to 2p + 2-{p + r) or to
Supposing
multiplier.
is determinate save for a constant
the value of
places Xp+r+u
. a^ap+a-
we may
{Z-Ci)...{z-
Cp+r)
S''
therefore write
(A)
and
s is
started,
a-,, ...,
is
and
it is
1,
x-^, ...,
a;p+,_i, it is
zero to the
first
order at
a, a'
a>p+r,
where z
is
infinite.
{Z
+ B,
wherein B, A, A^,
and
it
from
it
each of
where z
...,
Xp+r-i
and vanishing
(r
- 2)
Ex.
1.
Determine
x.
2.
, + ,_, in
a, a'
Ex.Z.
Ex.
4.
is infinite.
is
of the form
root
+ AiZ+ ...+Ar-iZ''-^)+B.
-Hr(^+^ig + ...+^,.n-i2'-^'-')
'l^^ziShAt^hE^
can be interpreted as an
56.
The
Chap.
I.
4.
(A) connecting
'z
and
36]
at z
= Ci,...,
(A) gives
Csp+j,
85
rise to
rationally
is
and
reversibly
represented.
It is therefore of interest to obtain the
functions
for this
new
surface.
(Z-C,)...(2where k
A - 1, is
to order
is
l)i_,
a, a'
where z
p + r-(r-l) + k-l, = p + k:
(Z-C,)...{Z-
is
infinite,
l\_^
+ {z,
fact
by suitable choice
we can
l)p+i,
and in
find
a rational
1)^+1,
is
V
which becomes
Cp+r) ^p+r
= (Z- Cp+r+r)
oo
...(z-
on the surface
Cq,+0,
infinite to the (p
+ l)th
l)p+,
order.
and
of
l)x+(2,
=x
39,
,
can
1)^1?,
the dimension of the function, namely, the number which is the order
higher infinity at these places, be p + 1, \ and fi are such that
if
its
+ l^/t+p+l.
Therefore also,
equation (A)
if
may
i)
('^>
l)p+i. ^^
which case
we have the
+ l>X,
...
{z
- dp+i),
result that all such functions can be also expressed in the form
l)v
(z,
+ (^,
1)m'0-,
with
+ l>\',
*
+ l%fi'+p+l.
Chap.
m.
28.
[56
EXAMPLES.
86
By means
of this
given Chapter
integrals
result, hitherto
II.,
17,
Let
57.
be such an
integral.
where z
Where z
or ;? - 1.
where dz = 0, that is at
immediately seen that the
is infinite (2)
It is
the branch
thus-
is infinite
the function
is infinite
an integral polynomial in z of
Hence
2,
to the order j) + 1
first kind* is
the
order f \. Namely, the general integral of
it is
{z,
l)p-i
dz
/ (2:-C,)...(^:-Cp+r)Sp+r
Ex.
58.
1.
contains j5-/t-jo+T+l
coefficients in a general
that
{z,
+ l=r+2-/ir
polynomial
c,,
{z,
...,
infinite only at
Cp_^.
exist.
Ex.
It
2.
can be shewn as in
p + l,jo + 2,
...:
by
At a
2, ...,p.
which a rational function, infinite only there, can have, are expressible in
one of the forms 2(p-k), 2p+2r+l, 2p + 2r, where i is a positive integer less than p, or
zero, and r is a positive integer: and we can immediately put down rational functions
infinite to these orders at the branch place z=c and nowhere else infinite.
Prove in fact
that the following functions have the respective characters
possible orders,
(^.
l)p-t
{z,
l)rO-
(2-0)"-*'
+ (2-e)(z,
l)p tr
(3-c)P + + l
''
'
(z. l)p + r
(2-(3)J' + '-'
wherein (z, l)p_i, (z, l) (z, l)p+r are polynomials of the orders indicated by their suffixes
with arbitrary coefficients.
Shew
2p-2k
times at the
3.
Ax''
coefficients;
z=^i+\xly,
* Cf.
59]
wherein X and
fi
is of
87
And that
same
Prove
fxirther that
is infinite
given by
is
This function
is,
therefore,
sum
We may
is
is zero,
(0,
0) at
which
has
2.
then proceed as in the text and obtain the transformed surface in the simple
But a simpler process in practice is to form the equation connecting
hyperelHptic form.
z
and
Writing
h.
k=h\
]c^{[PZ? + QZ^+RZ+Sf-i{AZ'^
Hence,
if
left
we
write
T=\_{P'-P)Z^+{Q!-Q)Z''+{R'-R)Z+^S'-S)]lk
1)6,
which is the equation of the transformed surface. And, as remarked in the text, the
transformation is reversible verify in fact that x, y are given by
;
referred to one
59.
The equation
represented,
by which, as we have shewn, any hyperelliptic surface can be
Cy,+,. If we write 2 = {ax + b)/{x + c)
by suitable choice of these
constants;
we introduce three new disposable
in which there are only
form
the equation of the surface can be reduced to a
2p - 1 parametric constants. For instance if we put
contains 2p
+ 2 constants, namely c
(z
c,,
. .
s=Aa{z-c,)-^\
where the constant
is
given by
(C3
- CjpH-^)*,
[59
EXAMPLES.
88
the equation becomes
- ai)(x - a,)
Oap+j),
s'
= X {x -
a,
1) (x
...
(x
wherein
It
is
as rational function of a,
s,
Thus
z.
positions of the 3p
(Chapter
I.
Thus a
surface for
which p
(pl)p(p+l)=Q
fact
is
places*
Sp-3 (2p-l)=p-2
relations.
There are in
which we can construct a rational
for
(Chapter
I. 8),
p=
1 is also h3?perelliptic
but
it is
more than
this
itself in
On
1.
(6,
0) at each of
K)
six
= 6-l + 6-l-|-2(3-l)=14.
w=2n + 2p-2,
putting
Ji
= 6, we
Thus there
obtain
^ = 2.
can be
^_x b
is infinite to
the
first
and
is
not elsewhere
infinite.
To
y
obtain the values of $ at the branch places of the new surface, we may express either
in terms of ^.
Since there are two places at which | takes any value, each of x and
y will be determined from f by a quadratic equation which may reduce to a simple
equation in particular cases. When | has a value such that the corresponding two places
or
Now we
^_
*
have a repeated
root.
have
Chap.
III. 31.
{x-bf _
y2gi!
T{x-a)~ {b+y^){b-a+y^y
+ Forsyth, Theory of Functione, p. 349,
59]
89
Hence
2''(l-l)-y|'(a-26)-6(a-6)|4=0.
is
The
factor
a2(f6-l) + (a-26)2,
equal to
is
[a^{{x-b)^-a{x-a)} + {a-2bYx(x-a)]lx{x-a),
which
is
or
{[x (a
Thas
which
x=
- abl{a - 26).
And
if
we put
we obtain
,2
which
is
= a2(|6_l) + (_25)2,
,f-3=[x(o-26) + a6]/[:r-6],
we obtain
:r=[6,+a6|3]/[,_|3(_26)],
3^=[26(a-6)f]/[,-|3(_26)],
which give the reverse transformation.
Ex.
2.
f=x{x-a){x-bf{x-cf
that
^=2
^={x-b){x-c)ly
of the second order.
is
to
= [a^3_J_(,p + 4jc(|3_l)
and that
x=[a'^^^-\-ari-\-'ibc-a:b-ac\l[a^^+i)
+ b + c-2a],
3.
In the following
order, that in
of
X and y
each case
i;*
is
five cases
as rational functions of |
and
y^o=x{x-aY{,x-b)\
i = {x-a){x-b)/y%
r,
(;8)
y=x(x-af(x-b)*,
i = {x-a){x-b)ly\
r,
(y)
j/-
()
(f
17
(a)
= x{x-a){x-bf,
is
i={^-b)/y,
r,
= -J^.{x-a)^(x-bf
= sla.{x-af{x-bfl]^
= [x(a-2b) + ab][x-bf/y3
90
Ex.
Shew
4.
[59
y=(^--a,)"'...(^-a,)'
can always be transformed to such form that Mj, ..., k^ are positive integers whose sum is
by n and in that form determine the deficiency of the surface. Shew also that,
in that form, the only cases in which the deficiency is 2 are those given in Exs. 1, 2, 3.
divisible
i^=x{x-aY{x-hf, y^=x{x-a){xh),
y*=x(x-a){x-bf, y^=x{x-a)[x-h){x- e).
The
De Transformatimie Aequaiionisy^=R{x)
E. Netto,
The equation
y=(;r-Oi)'...(a;-a,)'is
to put
I.,
in a position, immediately
60.
now
of a
c,
Chapter
Riemann
III. 28,
surface
g^f,
Examples
31
1, 2, 3.
considera-
36.
<
a.
be the
all
We
a.
suffixes of all
N= fM +
i,
where
the values
= 0,
modulus
The
. .
.
(a 1) will arise.
whose order
is
Let
f/^a
congruent to
+i
i for
..., 5'M_iO+a-i-
and a constant \ can be chosen so that gra+ia rational function infinite only at
c,
is
m cannot be
^g"^''''
g,.,^+i,
less
which
than
is
/tii,
clearly
/tfa
i.
wherein
/ui
+j
clearly reach
is less
rm +
i.
'
g^^+i
+ g^+j
we
;rein
than
= f^g^
= A + Bg^,a+i + Cg^^^+
the coefficients
A,B,
...,K, whose
...+Kg^^_^a+a-i
number
is a,
(i)
Briot
62]
In particular,
if
gr,
91
g^, functions,
we have equations
gf =Ai
+ B^^^a+i +
+ K^^,^ _ ^a+a-i
(ii).
61.
jr^^jO+o-i in terms of ga
and
...,
gr,
g^n+i
wherein
= Qm i9r - A,) +
Q,- ,
-A,)+... +
{g,^
^r^,
Q..
_, (^""^
- A^,),
of integral form.
If
of the form
P,{gr-A,) + P,{g,^-A,)
+ ...+Pa-,{gl-'-A^,) = Q
or say
Pa-, gl-^+Pa-,g"-'+
...,
Po^i,
...
+P,gr +
P=
(iii),
Pi-o-i
Then
Pj,
is
+ PtiA^ +
+ rk
of order aX^
Denote
P denotes
may
(iii)
+ PaiAai.
In order
equation as
= aXif + rk",
a\k +rk
and therefore
r/a
Now
k and
r and a have a
62.
k' are
both
common
= (\^-\i)/ik-k').
than a
less
divisor.
be linearly independent.
to
And
then
it
gr,
(ii)
must
^r^^o+i, ...,g^^_^a-\-a^i
cau
Also by equation
(i) it
Lgl
+ Ly-^ +
gr'.
and
in
...-\-La-,gr
particular that
+ La==0
there
is
is
an
(iv).
92
wherein Z,
...,
however, since gr
only infinite
is
rational
integral
io are
[62
when ga
is
It
follows
g~^
is
a branch place at
is
it is
infinity at
clear that
which all
To every value
highest power of ga in equation (iv) is the rth, and this term does actually
While, because ga only becomes infinite when gr is infinite, the
enter.
term
coefficient of the
The equation
(iv) is
g^^ is
is
gr).
This equation
poles of
snu.
63.
(i)
(1. giL^a+i,
for
fl'M_ja+o-i)
of the
is,
dimension
an integral function
of such
is finite
an integral function by
no terms of higher dimension than the
this fundamental set is therefore entirely such
function to be expressed
an one as that used in Chapter IV. If k be the order of infinity of an
it is
integral function F, at the single infinite place of the surface (iv), it is obvious
F is
That
is,
an
infinite
number
of values of g^.
POSSIBLE POSITION OF
64]
93
We
method
that the
is
rational functions
not exist of
exist
number
may
is
We
otherwise.
fall
limited
infinite
Such functions do
4.
(ii)
(iii)
their
by the considerations
In fact F^'Fi'...
of deficiency
orders there are four orders for which such functions do not
may be 1,2, 3, 4, and this is the commonest case*,
a fimction of order
2/j
is
or they
(i)
all
is
niA;i
+712^2+
...
where
any positive
integers.
such a function.
The number
of non-existent functions
must be
4.
bef
gi-eater
than
1 or 7.
It follows that
oi-der 2 exists
a function of order
Consider then
down all
numbers
first
up
when a
to
8 to indicate that
2^ or
all
a function of
if
for
the case
positive integers
2, 4, 6,
sum
every positive
3.
Write
Draw J
8.
1234.5 678
(0).
a gap beyond
8,
which
is
1, 3, 5, 7.
We
(iii)
above.
Hence
Consider next the possibility that a function of order 3 exists, there being
no function of order 2. If then a function of order 4 exists, the symbol
will
be
_ _
1234567
8,
8.
III. 27.
exists
94
and 5
[64
symbol would be
12345678
(/S),
functions of orders
the function of order 8 being formed by the product of the
So far then as our conditions are concerned this symbol represents
3 and 5.
a possibility.
Another
is
12
3 4 5 6 7 8
(7)-
In this case however the existent integral function of order 8 is not expressible
as an integral polynomial in the existent functions of orders 3 and 7.
When
is
other-
Consider next the possibility that the lowest order of existent function
Then possibilities are expressed by
4.
12345678
(8).
TS
(e),
3 4 5 6 7 8
(?),
2 3 4~5 6
12
as
arise.
is
to
5 exists, g^ven
when no
by
12
3 4 5 6 7 8
(v).
For these various cases let a denote the lowest order of existent function
and r the lowest next existent order prime to a. Then the results can be
summarised in the table
p=i
65]
3, JJ
9c
4.
That the seventh and eighth columns of this table should agree is
accordance with Chapter IV., 41. The significance of the last column
explained in 68 of this Chapter.
Similar tables can easily be constructed in the same
/)=],
Ex.
2.
1.
3.
p=s
^=3
the
I'etiults aj-e
given by
way
for
in
is
the cases
96
wherein
{g^, 1),
upper limit
for
[65
left.
Similarly
Instead oi g,, g^ we may clearly use any functions g^, {gz, ^\, gs~(g3, 1)2Choosing these polynomials to be those occurring on the right in the value of
g^gs,
= <^.gs + ^^7 +
= 72^8 +
04 , 9i
Os^f,
atj
g^g^
= 0^
(A),
their suffixes.
Multiplying the
subtracting,
first
and
last
we obtain
9705
+ a, = 0,
0,0,-/35
0,
a,a,
+ 0:^, = O,
from which, as a, is not identically zero, for then g., would satisfy a quadratic
equation with rational functions of g-j as coefficients we infer
05
+ A3 =
(B).
we have
oii)
+ a>g,,
and thence
iSs-
0203
= 0,
03/32
+ 05 = 0,
72/85
+ a3a4 = 0,
we have
7,02
+ 04 =
(C).
The equations (B) and (C) have been formed by the condition that the
g-rg^ and g^-gT, however these
latter products be formed from equations (A).
We desire to shew that, conequations (A) should lead to the same values for
polynomial in
g^
and
g^ should
* Chap. IV.
43.
66]
g.,ga
gTgiK=gi.gj K.
97
Let
condition is that g, {Lg^^ + Mg^g^ + Ng^ shall be equal to g^, (Lg^g^ + Mgj' + Ng,).
This requires only gr ga' = g^ g7ge and gT g,ga = g,. g,': and it is by these
conditions that
Hence
(C).
also
g^g
Thus every
and g^ will, when the conan unique value however it be formed from
equations (A).
are
equivalent to
<
=-
0272,
ySs
012
= tta/Sj,
atj
5'?"
gs
= 725'9 +
- <h^i
"sfir?
gige,
ct,
03.
Thence
St'
or
~+
= "2
g?''- ^2g7''
which
is
(iv)
~ *272.
^^97
+ Oiyo.gr -ai'a,= 0,
which belongs to the possibility under
consideration.
1, ffj,
g^ in terms of g^
and
g^ is therefore
'
= 4,
'7.
66.
"t
r=
5,
possibility
We
64
6.
where
For a
Oj
we may
above,
2,
3,
1, 5^5, 5'5^
5^7-
1,
where
e,
c is
= 97(94,
a constant
1)1
let this
gig7
+ 09,"+ 95(94,
h for g^ - a^
and
+ (5'i.
1)3
= <^i97 +
S'n'
+ ^i9= +
Write
l)i
hj for
3,
g^".
g,-h-^i- 2a,.
Then
AA = ai'+i/3. +
Replacing
now
h^, h,
by the notation g
g,
03-
and
a,
-|-
a,^,
-|- a,''
by
write
g,g7
= a
g,^
g.^
EXAMPLES.
98
g^
gf =
/Safifs
(hg^
gr,g.,
[66
g^ requires
oi [73
^las,
from which
a3=ai72,
y33
+ ai/32 =
a2+Oi7i=0,
0,
a^'i3
= - ^i^h,
and thence
ai73= -/3iai72, or
not zero,
if Oj is
5''
i" *^^
^^y^/ff^'
or
which
is
(iv)
may be
(e).
taken to be
terms (at
prime to
a,
infinity.
Ex.
1.
a = 3, r=4
^=3
is
9i+9i^ia3, 1)1+^4(5^3.
y3 +yj;
X being
of the form
Thus the
Ex.
2.
Ag^+B, y
(a;
is
form
of the form
surface depends on 3p
l)3+(5'3. 1)4=0,
-4
Cfft
+ Dg^ + E,
A, B,
C,
D,
being constants.
or 5 constants, at most.
may
prove that
the homogeneous equation of a quartic curve which has a point of osculation, can be put
into the form
By
putting *'=i)/|, y =<>/$, this takes the form of the final equation of Example
1.
Com-
Ex.
3.
a = 3, r=5
p=3,
is
9i+9&{9i,
Ex.
4.
operator
Denoting the
left
'^)i+ 91.93
(ff3>
hand of equation
by
by
/'(cfr)
and the
67]
by D, prove that
infinite,
where
if
y be any
rational function
which
is infinite
99
A'o,
We
67.
function
fact that
an integral
is fi
should be expressible
m n
9=ga9r +
it is
necessary that there should be a term on the right hand whose order of
infinity is the
same
we must
therefore have an
= ma + nr
is less
than
a,
we may suppose
??
less
than
a.
And
is impossible for any value of
of the form nr ka.
k may be taken equal to any positive integer less than nr/a, the
number of integers of this form, with any value of n, is E{nrla), or the
Hence on the whole there
greatest integer contained in the fraction nrja.
This equation
/j,
since herein
are
"S E{nr/a)
orders of integral functions which are not expressible integrally by ga
and
gr.
Ha-1)(^-1)-
Thongh
it
72
100
[67
is
h{a-l)ir-l)-p.
In the table which we have given for p = 4 ( 64) the existing integral
functions which are not expressible integrally are, for the case (7), of orders 8
and 11 for the case (8) of orders 6 and 11 for the case (e) of orders 7 aud
;
11
14, 19.
numbers
which
pLs.
Ex.
1.
5'6'+^5'(S'3-c)+'6.9'3(i'3> l)2
+ 9'3%3.
1)3
= 0,
the function
is
an
Ex.
2.
where
02
the
/i being of the first order in g,, and c, 61, ^1, t^ being constants,
are integral functions not integrally expressible
9i=97
integral
Gi,
+ ft)/2>
(i'7
is
is
functions
(x2, ...,
.9ii=5'7 {9i
-\)-p
68.
tation which
1,
g^.
Weierstrass canonical
the
for
From
(?_!.
equations whereby
the
1,
Or' ^^
deduce an
g^, gr>--->
43) to
equation
A(l,^...,^r')=V^A(l, GG...,(?_a
wherein A(l, g^,
is
the
sum
of the values oi
g"'^)
is
is
g'^'^^'^
therefore
an
at the
same value of
integral polynomial in
(i,
j)th element
which also
ga,
is
is
^fa,
A(l, G,
sum
the
, . .
Go-i)
of the values of
an integral polynomial in
containing the right angle, and at unit distances from these sides and each other, so describing
squares interior to the triangle, the number of angular points interior to the triangle is easily
a-1
seen to be
(nrja). On
=l
a rectangle whose diagonal
is
the other
hand
if
it is
is (a
1) (r
1).
number
lines be
^^]
and
lOl
is
The determinant A
of the
values
a-
from the
it
when
only infinite
is
g^,
and gr are
infinite.
It
infinity
surface
A(l,
at
^r, ...,gr^"^)
Since there
is
of degree
Hence we
r(a-l)
infer that
the polynomial
in g^.
= a-l
Thus
is
of order
i[ria-l)-ia-l+2p)l
that
is,
of order
^(r-l)(a-l)-p,
^gaThis interpretation of the degree of v is of interest when taken in connexion with the
theorem Every integral function can be written in the form
polynomial v.
See
48 and
ffa
88 Exx.
ii,
^^^
(5'a,
ffr>
1) thiis
iii*.
When the factors of v are all simple we may therefore expect to be able to associate
each of them, as denominator, with an integral function which is not integrally expressible.
In this connexion some indications are given in a paper, Camh. Phil. Trans, xv. pp. 430, 436.
For Weierstrass's canonical surface see also a dissertation, De aequatione algebraica...in
quandam formam canonicam transformata. G. Valentin. Berlin, 1879. (A. Haack.)
Also Schottky, Crelle, 83. Conforme Abbildung...ebener Flachen.
69.
exemplified in
65,
of the general form of the equation of a surface when the fundamental set
is given, is not limited to Weierstrass's canonical surface.
of integral functions
Take
for instance
* Cf.
sheets,
and
let 1,
^r,, jrj
p. 268, 186.
be any set
INDICATION OF GENERALIZATION OF
102
[69
assigned in
of fundamental integral functions with the properties
exist equations of the form
Then there
42.
Chapter IV.
^192
ri'
= 7 + /3 + a f2
= 7i + /3i5'i + iS'2
sr,
left.
polynomial in
g^
and
g^,
of
is
namely
gi (7i
gx (72
+ a^S'i +
ft^rO
= g^'^.
7=
and thence
ia2.
7i
to
g^'
(v)
and
,92
= Oias/^Ti
it
rationally
by x and
g^.
two
equations
is
final
reversibly represented.
be represented
these equations.
The fundamental
equation, being
Ex.
1.
are given
where
tj
Ex.
-3i^i''
+ ai/32^i
-0/02 =
by
+ I,
2.
equation
^)=^/
first
r,
+l
jr^
and/'(gr,)=3//9^j.
when jd
is
odd = 27i-
1,
say,
i,
40,
'^0]
103
where
o, 0, y,
Ex.
fi
are polynomials in
3.
first
a;
of the equations
( v)
becomes
013^-^^^/323^-02 = 0.
If the
dimensions of
suffixes,
(A)
^1.
o-i,
number
jr,
-^d
^2-
70.
is
p.
we have spoken
portant
lie
+ i/3ii5'.-.'2 =
(E)
It is ivi-
Suppose our given surface to be any surface for which a rational function
of the third order, f exists. Take c so that the poles of the function
- c)-',
(f
which is also a function of the third order, are distinct ordinary places of the
,
surface.
poles.
g^
which
rational function
and
g^,
is infinite
form
7+
a^'i
+ ^g-,,
And
as before
other than
Oi,
a polynomial
a,
K of the form (x
-4i)">
(x
...
(E).
if
be any
Ai)'^...
such that
Further,
KF only becomes
Thus as X, ffi are rational functions on the original surface, (E) represents
a new surface upon which our canonical surface is rationally and reversibly
And it is as much the proper normal form for surfaces itpon
represented.
which a rational function of the third order exists as is the equation
J 04
o..
= (2_
l)^p^^_
[70
upon which a
exists.
Ejc.
upon
In the same way we can obtain a canonical form for surfaces
exist
there
that
shew
can
We
which a function of the fourth order exists.
as
three functions g^ 9i, 9z satisfying such equations
71.
^"-
'
- 9z
(a^
These
Ex.
the conditions
r^b,
a^h^ a^a-Jji-a^b^
is
(a^b^ - a^b^) - 63
(h^b,
+ bjc^),
hi
= aj>.^ki.
if
= OSS'! + hg^ +
CsfiTs
+ *o.
9i
''4'
"6> *>
9i
are respectively
"-'OI
while
"'4
*^2*'5
*'l''4
"'5
'^
"^e
^ ^l*-6 + ^3*'5
72.
When the order of the independent function, denoted in 69 71 by x, is known,
and the dimensions of the fundamental integral functions in regard thereto, the general
forms of the polynomial coefficients in the equations, whereby the products of pairs of
these integral functions are expressed as linear functions of themselves, can be written
And thence, if the necessary algebra (such as that indicated in the example of
which serves to limit the forms of these polynomial coefficients, can be carried out, a
canonical form of the equation of the surface can be deduced.
down.
71),
generally
n-1
is
an integral
74]
lu either case
Ex. iii).
( 88.
necessary to be able to calculate the fundamental integral functions.
73.
105
of the fundamental integral functions for any given form of the fundamental equation
associated with the
Riemann
may*
This equation
surface.
thus y
;;,
is
which
values
(A)
an integral function oi x
arise for the
their
sum
same value
will be
(x, y),
<f>
( 38).
of x, will be
denoted hj
and
Srj.
If
being integral
^ + ^y+.. +A-iy"-i
.
(^)
ri=
A,
...,
j4_i,
J)
common
divisor.
In the
first
integral polynomialt
forms.
For
let
A i\,y, ...,y~')
so that
is
a factor of the
is
number
of
(t=0, l,...,{n-\))
Ai={x-ayBi-\-Ci,
is a polynomial of order
which divides D, not all of C, Cj
wherein Cj
is
may
i;
. .
is
less
,
than
(7 _ i
since
.4, ...,
can be divisible by
common
j4_i have no
a;
- a.
divisor
its first
form of
expression.
From
the equation t
:^{\,y,...,t-\i,t^\...,r~^)=v\i^{i,gi,..-,9n-i\
recalling the
is
left
hand
side,
we
infer
VIVi = {JC-ayiCi,
so that {x aY divides V-
Thus the
first
u^ ...Uv\
wherein
it,, ...
is
to put A(l,y,
+ Chap. IV.
43.
106
factors.
common
[74
This can always be done by the rational process of finding the highest divisor
It will include
...,^' ') and its differential coefficients in regard to x.
A
to
{\,y,
we
if
of
Suppose then that x-a is a factor which occurs to at least the second order in
Denote x-a by u. By the solution of a system of linear equations,
A(l,2/,...,y""').
we can (below, 78) find all the existing linearly independent expressions of the form
75.
If the highest
a, aj, ..., a_i are constants, which represent integral functions.
power of y actually entering be the same in two of these integral functions, say in f and f
we can use instead of f a function of the form f - ^f, where ^ is a certain constant. By
continued application of this method of reduction we obtain, suppose, k integral functions,
wherein
'
of the form
f,
1, 2,
. .
f, for
1/*.
(m I)
all
let a
denote in turn
the
all
n\k
than
n,
among
the sequence
Put
As
is less
These values of
different.
(C)
before
we can determine by
In-l-
form
(/3+Pifi + -+/3-if-i)/,
/3_i are constants, which are integral functions ; and, as before, we can
choose thera so that the f 's of highest suffix which occur shall not be the same in any two
wherein
(3, /3i,
...,
Then in place of 1, f ,
_ i we obtain a set
, f
unless there be an integral function of the form
wherein ^^
is {r
{^ + ^1
wherein the f of highest
(,
there are
any
1, |i
+ ... +ffr(r)/u,
. .
. . . ,
$n - 1
>
(D)
form
(r+yili+-"+y-il-i)/.
being constants. If there are, the process is to be continued +. If there are
none, let v denote any other hnear factor occurring in A (1, 3^, ..., y-") to at least the
second order. Then, as for the set 1, j/, ..., y>- ', we investigate what linearly independent
Ti lyn-i
form
(a+aj|i + ...+an_i|_i)/y,
76.
When
for
as for
jj
and afterwards
we
shall obtain
work below,
be polynomials of order
+
The nnniber of
if
+ cl,i;, + ...+o_i.,_lV(.r-o),
u be a polynomial of order
r 1.
steps
is finite,
by
74.
r, it
is
necessary to suppose
a, a,,
...,
o. to
77]
wherein
a, ...
107
successive
,_,), that
wherein
i),
!),,...,
= + Bji,, + ...+J'-i>;-i,
l,
<)_i
Hence every
wherein E, ...,En-i,
are integral polynomials in x without
and we write
:i; c be a factor of
common
divisor.
now
If
Ei={x-e)Gi+ai,
aj
is
an
+ 6'ii;, +
..
of the set
1,
iji,
...,T)n-i there is
each of
side, unless
By
the property
of the right-hand
u-i be zero.
a, oi, ...,
wherein H,
77.
H-^, ...,
But
H^^
it
(E)
side
which are of higher dimension than 7. We prove now that this requires a further
reduction in the forms of 1, iji, ...,ij_i, which is of a kind precisely analogous to the
here,
Let
(T-t-l
be the dimension of >;, pi the order, and therefore also the dimension of the
( 76) and tn+l the dimension of 7^; we suppose o-i :|> o-j :j> ... > <r_i
polynomial Hi
then
$)p.,
an integral polynomial
in f,
this equation is
h^...+(i,$),.hje*-^'i-''+....
If
now
in equation (E) a
tj,
p (<j--|-l), ...,pi+<ri-<r,...
is
Then we can
write
th={...+ii,i)n>i/ii+...m
wheiin the symbols
wherein
aj is
(1,
a constant,
in
$)^ denote integral polynomials
Putting
we have
left
hand
is
a function which
is
not infinite
when x
is infinite.
Hence,
108
when the
condition of
set 1,, ...,,-, are such that the
[77
dimensions*
is
not
satisfied,
+ aiAi + ...+a_iA-i)/f,
of the form
a, ...,
when
infinite
zero or
is
is infinite.
when x
is infinite,
are not infinite
In virtue of their definition the functions A ..., /i-i
them
regard
therefore
may
We
infinite.
and are therefore infinite only when x is zero or ^
the form riJx, the
since there exists no integral function of
as integral functions of
As
And
...,/i-i as functions of
dimensions of Aj,
| are
o-i
+ l,
...,
<r-i
+ l-
a, ...,
Then
when x=0, as appears from the form of the right-hand
and since k^ is not infinite when x is infinite
Denote it by Gr- Then ij, can
dimension
is only a-r
x
whose
function of
a function which
is
side
it is
is
not
infinite
an integral
be expressed in the form
it is
r,,=
-^[j^'''+^+^,rx'"-" + ...+^r-,Vr-i^'"~''-'-Grl
(F)
and
in the right
less
,
. .
symbol G,
l)^Gi + ...+{x,
...,
G_i
is less
form
l)^_G_
by
of the dimensions
of >;i,...,i;_i.
If now in this expression of integral functions by (?,, ..., G_i any terms can arise
which are of higher dimension than the functions to be expressed, we can similarly replace
the set Gi, ..., G^i by another set whose dimensions have a still less sum.
1, the sum of the dimensions
whereby integral functions are expressed, cannot be diminished below n 1.
We shall therefore arrive at length at a set gr^, ...,g_^ of integral functions, in terms of
which all integral functions can be expressed so that the condition of dimensions is
of the functions
satisfied.
It is this
system which
it
Chap. IV.
39.
78]
109
For the surface associated with the equation y^ = {x, l)2p + 2 all integral functions
{x, \)j^ + {x, l)^ ij,, where iji=y+a;'.
If m>jo + l
Ex.
iji
m.
is
wherein a, uj
can be an integral function of
Shew
it is
exist
^.
{x, \)j^
is
satisfied
+ {x,
y.
l)j^
= 01^.
Thus
when a + ai = 0.
all integral
condition of dimensions
is
now
satisfied.
There is one part of the process given here which has not been explained. Let
be integral functions, and let u denote a linear function of the form x c. It
required to find all possible fmictions of the form
78.
i)i,
is
...,
i7_i
(a
+ nii;i + ...+n_ii)_i)/M,
We
suppose
u_i are constants, which are not infinite when m = 0.
ij,, ..., i)_i to be such that the product of every two of them is expressible in the form
i'n-i'?n-i) ".. "n-i being integral polynomials in x; this condition is
+'i'i'ji + ...
wherein
a, ...,
The
wherein
an
A', is
polynomial in
divisible
by
x.
u*,
integral polynomial in
a, ...,
wiU
satisfy
is
A'; is
also
it is sufficient
an integral
that
A'; be
always be
will
And
satisfied.
of ff
which
it is
arise for
But in fact these conditions can be reduced to a set of linear congruences, and eventshall not give here the proof of this
ually to a set of linear equations for a, ..., i_i.
For
in
many practical cases we can obtain
equations.
resulting
the
give
but
reduction*,
We
Let
much
11J,...,
-,
shorter way.
.,.,...,
a, oi, ..., u,
(mod.
?)
values of the quantities c, -.'i, ..., c,.. Substituting in the left hand the
c^ we
and equating separately to zero the coefficients of c, c,
equations
we
can
By
these
a^.
o,,
constants
n,
...,
the
for
linear equations
value of
for
obtain r + l
is
Which
is
which u =
. ,
given by Hansel, Acta Math. 18, pp. 284292. His use of homogeneoas variables
tlie theory of the reduction referred to.
85, But it is unessential to
110
number*
express a certain
/3i, ..., /3,
of
a, oj, ..., Ur
.-.
+ ar^r takes
the form
fJ with the
first
term - of the
+ ...+0,f
/3if,
with constant
[78
coefficients,
wherein
We associate [
..., j9,.
fj, ..., f,
We
proceed thence
to deduce a set of integral functions associated with the next term of the series,
in
the functions
t'.
gruence
2(yif,+
.-.+y.f.)(i^, + ...+e,f,)'-'=0
(mod.
u'^^^).
Herein y,, ..., y, denote constants, {denotes in turn all positive integers not greater
than n which are exact multiples of the denominator of the fraction f, so that ie is an
integer, |if'| denotes the least integer which is not less than ie', and, for any proper value
of i, the congruence is to be satisfied for all values of the quantities ,, ..., e^.
It will be
found in practice that the left-hand side divides by !' -i for all values of
...,y,,
yj,
we
If
ii . 7n-
carry out the division, then, in the result, substitute the value of
coefficients of the
."*
(
products of
which enter on the left, we shall have this number of linear equations for
Solving these, and thereby expressing as many as possible of
yj, ..., y, in
terms of the remaining, which we may denote by y,', ..., y'^,, y,|i-f...-l-y^ will take a
form y^'^l'-^...-^-y'm'f', wherein y/, ..., y'' are arbitrary constants, and
i^', ..., f. are
definite linear functions of fj, ..., |.
We say that f/, ..., ^'' are associated with the
!, ..., ti
Til ,?.
fraction
t'.
This process
case
when
e'
= l.
The functions
is
...,
Gt, are
all
may
therefore be denoted
1.
The
successive
by
(mod.
u^),
(52)
(mod.
u^),
(53)
(mod.
u'),
(i^i-l-...-l-efJ-.
In fact 3
And
is
3> 2> 3i J)
J Divisible by
x', in
a sense.
r.
f,,
..., f,
79]
We
therefore have
('
i
1
111
EXAMPLES.
112
Hence
1,
y,
Ex.
is 1
[79
1, gi, gr^,
g^ in Chap. IV.
and
+ 1 + 2 (4 1) = 1.
ii.
equation
y3 _
for
3^2 + 3y.J!
(:;;
- 1) +
.x:2
(a^
- 1 )2 (9.r3 + 7a-2 +
5.?;
+ 3) = 0,
= 3x,
2=3^H6:i7,
^4=
by
it
The determinant A
By
- 108^ + 132^' +
Denote
it
by
y^) is divisible
(a
wherein
+ 36.j;3+ ISx^.
(1, y,
We
ij.
3;j:<
a, ui, 02
The congruence
the coefficients of
(<S'2)
c, Cj
Cj
=2
(a
and
(mod. .r
1)
leads,
considering
aj + o,| + a2^*, = 0(
,.r-l),
which give a=0, a, = 802, and shew that the only function of the kind required
for a constant multiplier,
is,
save
{r,-Zy)l{x-\).
The
the congruence
But
find
(.S'3)
in fact, if
= 0(
we
,x^)
write
for example,
immediately
becomes
so that
is
+ A'ix{x-\)-^9Ax = 0,
an integral function.
y'^jx
is
+ a,y + a2y2)/;r.
Prove that the dimensions of the functions
1, y>
are respectively
Putting x=\l^,
the form
Lv'^-'^^y)lx{x-\)
0, 3, 3.
y/.x^
= h,
+ tt,A + 3a2
(A2-3|2A)/^(1 -i)]/i,
The
is
3 + 3 - (3 -
1 )
= 4.
81]
CHAPTER
VI.
Geometrical Invhstigations.
It has already been pointed out
80.
associated with a
plane curve
is
( 9)
may be
we may
call this
Riemann
the variables
is
surface,
mental curve.
surface
Riemann
expressed.
(x,
may be
regarded
Thus we may
on the Riemann
surface with the theory of the intersection of a fundamental curve with other
In the present chapter we shall make full use of the results obtainable
from Riemann's theory and seek to deduce the geometrical results as consequences of that theory.
when
it
greater than n;
in
x and
y, is
not
Suppose, further, that the curve has no multiple points other than ordinary
double points and cusps, S being the number of double points and of cusps.
Consider
now another
at our disposal.
^fr
to pass
curve,
By proper
i/^
{x, y)
= 0,
-v/r
we can determine
through any given points of/, whose number is not greater than
Let k be the number of the
of disposeable coefficients in -v/r.
the number
prescribed points, and interpret the infinite intersections of /and <fr, in the
Then there
usual way, so that their total number of intersections is mn.
INTRODUCTORY SKETCH.
114
remain
- it
toji
all
of/ and
intersections
We
already prescribed.
the coefficients of
-|r
[81
yfr
to prescribe as
many
of the intersections of
and/
yjr
and introduce further the condition that -^ shall pass once through
each cusp and double point of/ then the number of remaining intersections
which are determined by the others will be ^ = |- (n - 1) (n - 2) S - *, for
For, if m ^ w, the intersections of yfr with /are the same as
all values of m.
as possible,
those of a curve
^+
U,-rJ'=
0,
is
Jlf
Of
therefore the
is
these intersections,
disposal of 8
Um-nf
coefficients in ^/r+
form,
71
suitably.
new
+ 1 = 1 + ^m (m + 3) - i (m - + 1) (m - n + 2)
arbitrary coefficients.
with
{m-n + l)(m -n + 2)
to zerof.
and cusps
will,
number
of the intersections of
by choosing the
by hypothesis, 2(8
of the curve
coefficients in
+ k)
-v^
yfr
are to be taken
mw - 2 (8 + )
is
number
the
disposal
and
of intersections determined
this
number
^(n-l)(n-2)-{S +
In case
m < n,
of the
mn 2(B + k)
of/ we
K).
intersections of
can,
ijr
by means of the
at our
is
-^jr
is
^m(m+3) 8
/c
prescribed to vanish at
mft-2(S + )-[im(m+3)-(5+/r)];
when m = w
1 or
it is easily
is
Let us assume now that the polynomials which occur, as the numeand denominator, in the expression of a rational function, have the
82.
rator
*
Beasons are given, Forsyth, Theory of Functiom, p. 356, 182, for the conclnBion that this
is the deficiency of the Biemann surface having f (y, x) =
as an associated equation.
shall assume this result.
number
We
?/,
in
^+ U^_^,
j/*",
j*"-', y'^'^x,
that remains,
is !/""'.
..., j/",
y'x,
...,
j"x">-, in
which case
INTRODUCTORY SKETCH.
83]
115
property here assigned to yfr, of vanishing once at each double point and
cusp of/! Without attempting to justify this assumption completely, we
remark that if it is not verified at any particular double point, the rational
function will clearly take the
/, -4],
...
wn -
2 (S
+ a:) - Q
jB,,, ....
Through these other points B^, B^, ... of/ and
through the double points, draw another curve, ^, of the same degree as i/r.
The curve ^ will in general not be entirely determined by the prescription
of the
mn
+ k) Q
2 (S
mn-2(S + K)-Q + q
^
points of ^, to determine
completely.
all
but
mn 2(B + K) Q + q = mn
Hence Q
where
q = p,
\, Xj, ...,
...,
all
k)
p.
passing
through
i/r,
the
mn
2 (B
q+1
are
k)
linearly
points
as well as through the double points and cusps; and the general
X+
where Ri
2 (8
independent curves,
Bi, B2,
and
= ^i/-^
and
Xi 2?i
. . .
+ XqRq,
q+1
arbitrary coefficients.
83.
In this investigation, which is given only for purposes of illustration, we have
assumed that the prescription of a point of a curve determines one of its coefficients in
terms of the remaining coefficients, and that the prescription of this one point does not of
and we have obtained not
itself necessitate that the curve pass through other points
the exact form of the Eiemann-Roch Theorem (Chap. III. 37), but the first approximation to that theorem which is expressed by Q-q=p; this result is true for all cases only
;
We may illustrate the need of the hypothesis that the curves ijr and ^ pass through the
double points and cusps, by considering the more particular case when the fundamental
curve
82
INTEODUCTOKY.
116
wherein
is
[83
{x, y)^ is
we have
p = ^{n-l){n-2)-t-<=i.Z. 2-1=2,
and therefore (in accordance with Chap. III. 23, 24, etc.) there exists a rational function
having any three prescribed points as poles of the first order. Let us attempt to express
not
this function in the form ^/^, wherein ^, V' are curves, of degree m, (wt > 1), which do
vanish at the double point. Beside the three prescribed poles A^, A^, A, of the function,
yj,
will intersect
curse
of the general
m=3
4m 3
i/^
and
shall
x and
is
made
is
As
allowed to lapse.
may
be
in the
finite for
y.
curve which passes through each double point and cusp of the fundamental curve/
is
a corre-
/=/-(l-;c2)(l-Pa;2) = 0,
we
this
tion
common
by z=
+ kx^,
namely there
is
a point of
For example
If
by the parabola
y=ax^+bx + c,
which, by the same transformation as that above, reduces to
z=ax'^ + bxz+cz',
and
it is
all
hntp
of them,
1.
"We shall see in this chapter that we can obtain these results in a somewhat different
way: the equation y''=(\-x^ (1-i^x^ is a good example of those in which it is not
convenient to regard the equation as a particular case of a curve of degree equal to the
is
always allowable,
it is
often cumbersome.
Prove that the theorem, that the intersections with / of a variable curve 4f are
determined by all but p of them, may be extended to the case where / has multiple points
Ex.
1.
This process
is
84]
117
^|f
of/
point
x.
point at infinity
adjoint ciu've
may
+k
of/.)
as a sextic.
Shew that the singular
be regarded as the limit of eight double points, and that a general
is
\f+>^+y{a:, l)''+l = 0.
{X,
Ex.
number
2.
Shew that
3.
y^= [x,
1)2^ +
is
{:c,\T^^P+y{x,\r*^-^ = 0.
For further information on this subject consult Salmon, Higher Plane Curves (Dublin,
1879), pp. 42
48, and the references given in this volume, 9 note, 93, 97, 112 note,
119.
= y^ + tf'-'{x,
f{y, x)
so that
then
0-
infinite
is
if
so that
o-
=;,'
+1
o-
+1
we put x =
is
7;
Let
x.
]),_,
+ ,-ir+'-*.(l,
is
f)^,+
o-
^.
is
= 0,
be the dimension of y
is
is finite
when x
is
...+^^m-i)r+i)-An-.
l)x
(a;,
F^n,
+ ...+y{a:,
an integral function of
is
thus
\)^^
(1,
f)^_^__
quantities
Xi, X,/2, ...,
Ex.
1.
X_i/(n-l), \nln.
/+yV(^,
the dimension of
\\->ryofi{x,
as an integral function of
\\^=^{x, 1)6=0
is 3.
equation becomes
>;
is
2.
a=o
In fact yi=rij^=ylx
satisfies
the
equation
yi^+yi'^^, i)3+3'i(^. i)4+(^. 1)5=0
and
is finite
Ex.
surface
when | =
oo
or
x=0.
Shew that in the case in which the equation associated with the Riemann
contains y to a degree equal to the highest aggregate degree which occurs, o-=0.
2.
[84
HOMOGENEOUS VARIABLES.
JJ8
equation {7,, f ) =
the surface in association with the
?=0. The original equation
and shall speak of the infinite places as given by
may
writing x - c for x, c being a constant.
is practically unaffected by
of
sheets
n
the
that at x = Q,
therefore suppose the equation so written
places
the
as
places ^ =
the surface are distinct; and may speak of the
we
shall consider
We
x) = 0, F{t], ?) =0,
the simultaneous use of the equations f{y,
accordance with the view,
we shall be better able to formulate our results in
x='x> are regarded as exactly
hitherto always adopted, whereby the places
But it should be noticed that both these equations
places.
By
85.
any
like
finite
homogeneous variables,
regarded as particular cases of another in which
put
x^tolz,
For
used.
y = ujz'^'; we obtain
of a particular kind*, are
maybe
f{y,
2)
and
it
=2-"'"+''
U{u\
z),
U{u\
clear that
is
to,
<o,
where
z) is
= y,
by writing u = 7}, = 1, z = ^.
We may speak of tu, ^ as forms, of degree 1, and suppose that they do not
become infinite, the values a; = being replaced by the values = 0. When
z are replaced by to, tz, t being any quantity whatever, u is replaced by
We may therefore speak of m as a, form
t^+^u, y and x remaining unaltered.
o,
= x, z=l,
and
changed into
is
F{r}, f)
ft),
of degree
<j
U{u;
Similarly
(n(o-+i)
1.
when u,
CO,
m, z) is a
toa,
+ l),
being multiplied by
tz respectively.
That there
notation
may be made
We
clear.
undergo when
by homogeneous forms, may be fitly given here
Instead of/(y, x) we shall have TJ{ri; eo, z) which is equal to ^"'''"'""/(y, x);
instead of the integral
instead of/' (y) we shall have U' (u) = 0'""" c+Dy (y)
an
integral
form
of
degree tj + 1, equal
dimension
function^ gi, of
n + 1,
gi
=
it
is
to see that the
since
n
easy
X (Ti+ 1)
+^ 1,
to z'^i^^gi, will arise
2:"'+'^
equal
A
to
In
determinanti A(l, ^1,..., ^_,) is
(1, g^,..., r_,).
accordance with 48, Chap. IV. the former determinant will have a factor
Some examples
( 42).
It
may
ig
nsed by Heuael.
+ Chap. IV.
X Chap. IV.
42.
43.
when
<r=0.
HOMOGENEOUS VARIABLES.
86]
(caczY corresponding to a
finite
^1, ...,
</),(a:,
y)
is
119
we
63,
calculate
the form ^i
...,
a, z) from
(u,
that
fzdo)
adz
/t~'
^0 {u,
z)
tr {u)
wherein
/u.
z"-(h;
0'"-^) '-+"
z ) gr
{%h,
"i.
y).
^i)
'
Q>,
is
,i,-'<f>o(x.
+ Xfi^'- ^r (u,
az^ co^z
(a;,
equal to
yO
zz,{x-x^
thus equal to
is
If (y)'
where \ = /j,Zijz = (bx a)/{bxi we put
If in this
= 0, we
a;
a).
to be part of the expression of au integral of the third kind (Chap. IV. p. 67).
But
if
integral
is
exactly what
x=\j{^a)
The
1.
45, p. 67,
can similarly
be written
j'^wl^"'
Ex.
U{u;
2.
<u,
'")"''*^(''' "'
2)=0
1)25 +
2,
+(-,^r-''' *-.(,
')+
wherein y
is
of dimension
f>
-, ^)].
+ l,
the equation
is
m2
= (u,
z)2p+2
We
shall
are integral in
x and
t(y. ^)
*
= r-'(^.
l)^+...+y-^-'(a;, l)^^+...
If
we regard
infinities.
{x,
1)^_^.
as throughout undetermined,
we may
120
we
If herein
sion of
term
= 'nj^''+\x=\j^,
as an integral function of x,
2/
where
write y
(77,
shall obtain
an integral polynomial in
is
we
<r+
rj
[86
i/r
x)
^- "
(tj,
f),
is
and
(y,
i=o,
1,
{n-\)
ff is
{n-l-i){a + l)+fii.
Thus G
and
f)
(97,
is
is
The dimension
f.
i/r
(y, x),
of the integral
speak of
for f
as the grade of
be distinct, that
polynomial "^
f=
yjr
is,
(y,
if
f ), of order n
(17,
a;).
I"
0.
G may
low grade
yjr
x) are necessarily
(y,
the same.
of grade
tj
(jj)
may be
integer.
is easily
^ (y.
00)
of grade (n
l)(cr + l) + r-2
seen to be
= 2/"-'
(x, !),_,
x^ {y--'(x,
wherein the
be entirely absent
first line is to
first line is
to be absent if r
absent
= 0.
if
'j-
0,
{a:,
. .
1,
and the
first
if
!),_,+ ...+
= 0,
+(x. 1),^,,
the
first
(x, 1)^_,
,.+_,},
term of the
line is to be
(n-l)a+n-S + r
contains
nr
terms, this being
still
true
if
+ ^ (n - 1) (n - 2 + W0-)
= but when r = 0, the
o-
general polynomial
of grade
iin-l){n-2 + n<r)
terms.
This
is
obtained for r
*
when
The number
ff=0.
not the
in the
number
> 0.
is
way
numbers
for the
common
case
GENERALIZED ZEROS.
87]
121
Further, putting
f (y. *) = f-(-')<'-i-3)-' ^
(r,,
f ),
[(
l)cr
+ n 3 + r]n
the total
is also
number
is clear
Since
fi.
number
yjr
of zeros of
sum
of the
^ {v> f
is
If
?i (n.
by a change
1)
+ w (r -
number
is
ft(?i
{r),
y}r
We
{y, x).
(y,
is
where
(y, x)
yfr
+ /i(r-3).
'^,
(;,
i.
The
ii.
Ex.
iii.
generalized
1) ((r
+ l) +
number
We
x=
oo
it
is
this
extremely
and
speak
to
-i^
{y,
introduce
"V^ (77,
{i),
^^^
'^j
(77,
^f
^) is equal
zeros of -^ (y, x)
+ l) + n(r-2).
)t
{r-2)
of zeros of/'
is
{y), for
which
= 2,
is
n {n 1)
(cr
1).
of grade d,
Its generalized
in x.
x),
f-(-i)'r-(n-3)-ir-i) \^^
(^_ ^)^ ^j^e
co
But, since
7i(7i-l)(o-
Ex.
fi,
an integral polynomial
is
Ex.
by
x
2).
the
integral
=
a.t
shall find it
of expression,
artificiality
f
equal to
of finite zeros of
l)(o-+ l)
yfr
x)
of
= as
- 1) (o- +
at f
then
of the
of infinities
number
*^^'
of zeros
is
nd.
x)
is
rv^dx
f
J
is finite
at this place.
the condition
is
If
(y)
(y,
x) dx
I.
2,
3)
ADJOINT POLYNOMIALS.
122
be
shall
or again
at the place;
finite
[87
expression
(a;
g) yr
(y,
x)
f'{y)
be zero at the place, to the
shall
number
where /'
+w
of sheets that
(y)
is
v|r(j/,
x),
the condition
not zero.
And
if
wind
at the place.
w +1
being
is
that i^
(y, x)
therefore
is
therefore
(y),
We
is
at a finite place
order at least.
As a
the
first
ineffective at
places
all
of zero index.
If
</r
{y,
x) be of grade
(/i
1) cr + w 3 + r, and
= 0,
i/r
{y, x)
is
that, at all
the function
F'iv)'
should be zero, to the
first
order at least.
It is easily seen
that this
is
r--^dx
should be finite at the place considered.
When
= 0,
is satisfied
f (y) =
or F'
11
places where
the polynomial
x) is said to be
which contains a
simple factor corresponding to every finite value of x for which/' (y) vanishes,
and if
denote the number of these factors, it is immediately seen that the
If
adjoint.
(a;
a)
(j;)
i/r
(y,
polynomial i^
is
zero,
{y,
x)
is
where /'
(y)
or
i?"(r,)=0.
Ex.
i.
x0,
and w=0.
is
at each of
Cf. Forsyth,
Theory of Functions,
p. 169.
88]
The general
123
y.
ii.
3^ -
[( 1
+ -t") ^^ + 1] + 'S^^-^ =
At
branch places.
The dimension
is
f>
of y is
1,
(?i
l)(r+ 3 + 1
or
2,
is*
Ay'^-^By+C-^x\Py + Ex-^F\.
In order that this
C=0
that
may
and F=0.
By-\-Ay'^ +
and
if
we put
xjt] for
it is sufficient
x and
for
l/i;
Dxy + Ex\
i;
-,
Bji+A+Dx-irEj^,
which
is
\-{\+B)a?-ri^-\-hV=0.
Compare
Ex.
83.
iii.
y2
= (.r_a)3,
is
{x,
or if
= 0.
is 1.)
Ex.
ii),
we
is
(n
l)(o-
1),
2 (i + w) = n (w and
1)^-1
/ denote 2t and
W denote
2m',
1)
(<r
+ 1),
all
finite
/+
W = n{n-\){<T +
Hence, as f
Tf
we can
2n +
2j9
\).
- 2,
infer
p = Hwshewing that /
is
Farther
if
1) (n
+ no-) -
^7,
an even integer.
denote the number of zeros of an adjoint polynomial
=0
adjoint polynomial vanishes, at a place where /' {y) or F' {r}) vanishes, to an
order equal to twice the index of the place J, we have the equation
X + I = n(n-l){a+l) + n{r-
2).
* 86 preceding.
number X.
.t),
or
(tj,
f),
vanishes to order
+ X,
will furnish
124
[88
Thus, as
I
= n(n-l){<7 + l)-2{n-l)-2p,
we have
X = nr + 2p-2;
and
this is true
when r =
These important
Pliicker's equations*
0.
may be
results
o-
= 0.
x. i. The number of terms in the general polynomial of grade {nV)(r + ti-Z + r was
proved to be J (?i-l)(M-2 + ra<r) + nc- 1 or ^(ra-1) (n-2 + reo-), according as r > or )-=0.
This number may therefore be expressed as ^ + ^1+ 7ir-l or p + \I in these two cases.
Ex.
ii.
homogeneous form
((r+ 1) in a> and 4, and the form A (1, g^, ... ^-i)
of degree W.
The quotient A(l, u, ..., M"-')/A(l, gi, ... ^_i) is ( 43) an integral form
It is the square of an integral
in 0), 2, which, by an equation proved here, is of degree 7.
A (1,
...
?, tfi,
M""') is of degree
n (-
I)
iii.
It
in a, z together is ^/.
43
b,
Exx.
and
I, 2,
48
also
Harkness and
Morley, Theory of Functions, pp. 269, 270, 272, or Kronecker's original paper, Crelle,
that if for y we take the function
wherein
X, Xj,
polynomial
A (1>
y,
...
y->)= v^A
X, Xj, ...
t.
finite) order,
91)
the
g_,),
have any factor which is also a factor of A(l, ^j, ...,gn-i)- And the inference can be
madet that for this dependent variable y, there is no place at which the index is greater
than ^, and no value of ^ for which two places occur at which /'(y), or F'{ij), is zero.
R be any rational function, and suppose, first, that none of the finite
R are at places where /'(y) = 0. Let be any integral polynomial,
i/r
finite pole of R, to
II (x
by
fi.
Then
- a),
for
which contains a
which/' (3/) vanishes,
M(y,*) = Mi2t//'(2/)
*
89]
125
is finite
function
(c-3/.)(c-y,)...(c-y)
1=1
~ yi
^^
^^^_ ^^^
is
X and,
an
equation of the form
1.
Hence we have
in c, of degree n
therefore, expressible as
i=i
yi
wherein A, A^,
...,
An-i are
Therefore, putting c
replacing yi by
finite
y,
= yi,
integi'al
x.
we have the
R^}r
polynomials in
(y, x),
and
result
= y-^A, +
y^'-'A,
...+yAn-, + A_,
R = ^ly}r,
'b
y.
Since
(a;
'^,
like
yjr,
is
o) ^ _ ( a)
i/r
where/'
(y) vanishes.
^/n(a;-a)'.
126
y,
which
is
[89
be represented as before by
(?/)
+1
w,
it is
is
which (< + l)(w+l) > i + w, or, for which i > (i l)/(w + 1). Hence the
denominator Tl {x a)* only contains factors corresponding to places at which
the index | i
at which
is
(2/)
all
is
y"~', form a fundamental system of integral functions for which the condition of
dimensions
is satisfied.
p + n1 by
the
It
if
is
sum
of the indices at
clear that if
to find an integi'al
the places
all
.r
I,
y,
...
3/""' is greater
it
\R
is
polynomial in
a;
than
= 00
is
possible
an integral
function.
90.
polynomial
of grade (n
1)
have poles
polynomial
for
so that
yjr
any places f =
0, it is
adjoint at infinity.
= 0,
R^
+ s,
(nl)a+n'3+r, and
the
so that
it
tj
and
f.
will generally
places f =
;
also convenient, as a rule, to choose
is finite
=
When
R = ^-'l^r,
and
@,
be of grade
-^
a+n
at
all
where F'
both
is
i]r
so that f^/l"(7?)
(17)
infinite
and F'
^^^^^
^^^^^^
yfr
vanishes at a
(ij)
^^^.^p^
(17) is
zero at
Then,
since
f=00, and
that
R can
wherein i
be written in a form
is
an integral polynomial in
tj
|.
namely, the
I"-', and therefore r s is not negative
which occurs in the expression of a rational function in the
Hence @i =
polynomial
and
^-POLYNOMIALS.
91]
is
127
may
regard
^ and
i/r
as of the
same grade.
tfr
which
is adjoint
^fr
91.
We may
other than
Then, the
y"-^A
dv
+ y"-^A^ +
dxwherein A^,
by
^,
and
+ An-i
...+yAn-i
f'iy)
...,
let its
d^~^
But, as a function of
as a function of x.
dx
f,
f-("-i)<'-(-wi"(,;)
F'iri)'
Thus by a
repetition of the
rj
and
f.
is
or
negative.
is
y,
of grade
CLtV
(n l)<r
+n3
or less.
It
is
clearly adjoint
at
all
finite
1)
o-
places,
and,
3, it is
Conversely,
That
vaniaheB at
is, if
f=
it is
the polynomial be
<//,
of grade (n-
1)
if
<f)
IN ADJOINT POLYNOMIAL.
128
grade (n
1)
o-
+ 3,
?i
which
adjoint at
is
finite
all
and
[91
integral
-dx.
/;
is
an integral of the
Corollary.
(?i
We
first
kind.
linearly
first
kind.
We
can therefore
make
the inference
We have
92.
is
of the form
Vr_3+r
We
= 3/"-' {X,
assume
shall
l)r-2
in
+ 3/"-= {X,
l),_i
. . .
+ y (,
many independent
grade (
l)o- + 3.
?i
(n l)a + n 3 +
+ {X,
1V_,
+ afA|r_3.
1),_,
l)cr
+m3+r
is
equivalent
r contains nr
1 +p
terms.
has 2p
= 0.
F'(v)
Hence, the 2p 2 zeros of the differential dv (Chap. II. 21) are the
zeros of the polynomial /' (y) dvjdx, exclusive of those where /' {y) = 0, or
F'ir,)
= 0.
It is in fact
Ex.
i.
</>
vanishes.
f {y)=0,
(p
vanishes i times,
Ay + Bx+C.
The
places.
values.
have special
92]
Ex.
ii.
/(y,
particiilar case
129
is
a; =0 at which y = 0.
For general values of the coeflBcients in the
equation these are not branch places and f'{y) vanishes to the first order at each the
index at each place is therefore \i where i = l, and the condition for adjointness of the
general polynomial of grade 1, is that it shall vanish once at each of these places. These
conditions are equivalent to one condition only, that C=0. Hence, as there are no other
places where the index is greater than zero, the general integral of the first kind is
{Ay + Bx)dxlf'{y)
/
and
f>
x=0,
=2
y0
;
Ay + Bx
the polynomial
at
2p- 2
vanishes in
which /'(^) = 0.
jEx. iii.
In general when the equation of the surface represents a plane curve with a
double point, the condition of adjointness at the places which correspond to this double
point, is the one condition that the adjoint polynomial vanish at the double point*.
Ex.
iv.
y+/(^,
there
l)i+y(^.
l)2
y^+y^x, l\+y{x,
l)i
+ (^,
+ (x,
1)4=0,
1),=0,
only one place at infinity and the index there, in both cases,
is
Shew
is 1.
is
in all
cases
i(a-l)(|^|-'--l).
-
where
means the
than
r/a,
deficiency is given
by
^=i(r-l)(a-l)-/',
where
/'
Cf.
denotes the
sum
of the indices at
iv. p.
430.
is
all finite
The
method
practical
416)
is
often of
great use.
Ex.
V.
is
an adjoint polynomial
Ex.
+ {x,
of grade {n l)<r + n 3.
(x, iy'-^<f>i
...
l)^n-i-i<^_,
can prove in exactly the same way as in the text that an integral of the
the former like
y^), (a;/, y,'), at
be written in
can
constant,
a
being
and at the latter hke -C\og{x-Xj'), C
We
vi.
Clog{x-Xi)
the form
P_
f"
"
^-](x-x{)\x-x,')f'{y)'
where
is
The sum
y, of grade
to an ordinary fc-ple
of the indices at the k places of the surface corresponding
Ex. i.
the index at each of the places is in fact i{k-l). Cf 83,
;
is Jft (* -
+ Chap. V. 64.
1)
130
x=Xi where y
not equal to
is
x=Xi where y
Putting
y^'.
ifr
[92
is
form
in the
V.=(:r-0{Cy-i+C,y-H... + C_i)-(^-^i)(Coy-+C,y-H...+C"_i)
e_j,
Co, ...,
{n-l)a+n, and
R
at least, that
C',
is
...,
(iJo2/"~'
and = C'.
zero
Further,
if
(aJ-Oy""!
follows, since
it
"
is
of grade
+ TO + l
(^-^i')
the equation associated with the surface
of grade (M-l)(r
be written
Xi (^) denote
all
r + Qir'^ + - + Qi,
it follows,
is
not infinite at
are the
y-i +
;^,(a;i).y-2+X2(^i)-y"' +
where
(:ir,
x^
1)20-+. ..
^l)-(^>0+y"-2(^, l),r-l+y""^(^,
+ (^,
!)(-,)
,r
+ n-3].
denotes
[y
so that
- + Xn-i(^i)-
dx
-P=jy,/T;J(-2^,
(c, ;rj)
= {x^
a;),
and
general polynomial
"
'
{x, x^')
which
for ^p
of grade
(-
(%)]/(^ - ^1).
l)(r
( l)<r+?i-l contains 2
+ m 3.
yfr
terms.
The
yjf
vanishes in the 2m
practical determination of an integral of the third kind here sketched is often very
useful
Ux.
condition that
vii.
Prove that
if
it
y"
r-')=Q (1>
>/,
g,
>
S-.-i),
and
if,
X-i
respectively
we denote
W.
Xn-2
Xi
W.
O Q
be denoted by
the elements
1> 1.
yi. Vi^
S*!""'.
by
^1, 2j
>
*'2l ""SJ
* (^) + *i
(*) .9i
(^1)
+ +*- 1
>
gv. - 1
K).
in 46, is equal to
This takes the form Uiij + ...-|-Mi obtained in Ex. vi. when c = and c=l, namely
is a constant.
This condition will be satisfied when the index is zero at all finite
when Q
and
infinite places.
92]
Ex.
131
viii.
is
that
it
^=0.
at the place
adjoint provided
B0
Cdx ry^j^yx,S^)x^r)
is
/ML(il+J^')+^^+fo+i)l.
x x^
X-Xi
W\-
Jj
Ex. ix. There is a very important generalization* of the method of Ex. vi. for forming
an integral of the third kind. Let
be any positive integer. Let a general non-adjoint
polynomial of grade n be chosen so as to vanish in the two infinities of the integral, which
fi.
we suppose,
will
coefficients,
and
...
ffjf
dx
-li
is
The proof
is
on the
form of an
integral of the first kind in 91, with reference to the investigation in 89.
Further as we know
( 16)
that dPjdx
of the form
is
C {dPldx\ + Xi {dv-^jdx) +
where
C, X,,
...
. . .
+ Xp (dv^/dx),
i/r,
infinities,
which is
but two of the zeros of a non-adjoint polynomial
of grade
/i,
is
of
the form
where
^(^ is
a particular form of
form of
Ex. X. Shew that if in Ex. ix. one or both of the infinities of the integral be places
where /'(y) = 0, the condition for Z is that it vanish to the first order in each place.
Ex.
xi.
(2/y
^)3+(y. ^)2=0,
* Given, for <r = 0, /i=l, in Clebsch and Gordan, Abel. Punctionen (Leipzig, 1866),
Noether, " Abel. Difierentialausdracke," Math. Annal. i. 37, p. 432.
p. 22,
and
t Counting zeros which occur for x = oo or supposing all the zeros to be at finite places.
Zeros which occur at a = eo are to be obtained by considering f^t, which is an integral polynomial
,
in I
and
r,
( 86).
92
132
[92
which the dimension of y is 1, let us form the integral of the third kind with its
two places ^ = 0, 2/=0 by the rules of Exs. ix. and x.; taking /i = l, the
general polynomial of grade 1 which vanishes at the two places in question is \x-'ry.y.
The general polynomial of grade m - 3 + /i, or 2, is of the form ax^ -\-'by'^ -\- 'i.hxy + %gx +2fy-\-c.
In order that this may be adjoint, c must vanish in order that it may vanish at the two
for
infinities at the
it
must reduce
form
to the
{\x-\-^iy){Ax+By+C).
Hence the
Ex.
UAx + By + C)dxjf'{y).
is
xii.
6 ft
p.
(Cf
6 ft p. 19.)
19 in a similar way.
Ex.
which
xiii.
is
we suppose
simplicity
these values of
all
to be finite,
KR
function {x-aj)...{x-ar)
is
of the
and
is
expressible in
the form
l)r+(^, l),_T,_i5'i +
(.r,
form
this
(r+l) + (r Ti) +
contains
efficients are
..
(2?,
lV_^^_j_i5'-i;
these co-
each of the
11
k^
where
is
not
infinite,
not
is
and so
infinite,
on.
first
valent to only
93.
Let
nr-Q-(T+l)
.4i, ...
Aq
independent conditions.
be
Riemann
surface.
We
which
a condition which
they are the poles, each being of the first order*. This is
is always satisfied f when Q>p.
The general rational function in question
of the form
\ + \Z:,+ ...+\gZg,
wherein \,
\,...,\
whose
rational functions
The number q
is
Zi,...,
is
Zg are definite
Aq.
connected with
Q by
an equation
Q-q = p-T-l,
where t
-1-
1 is}
the
number
of linearly
form
/i,0(a;)-f
*
We speak
Cf.
Chap.
as
if
III.
+ /ipilp{x),
This
is unimportant.
X Chap. III. 27, 37.
93J
This aggregate
is
x, of
133
We
kind.
which are
left
Aq.
vanish in 4,
Now we
proved that
have
{nl)a + n S+r be
if
yfr,
grade
of
its
other
zeros being 5,
S^
wherein X,
= \^ + \,^,+
Xj, ...,
some of them
(nl)a + n
contains nr
+p
arbitrary coefficients,
of grade (n
and
these, in
l)cr
this
special
...
Bj^,
+ w 3+r
case,
by the
+ \%,
...,
...,
B^
for ^,
(t+1).
And, calling these places B,, ..., Bji the residual of the
Aq,
because they are the remaining zeros of the adjoint
places Ai, ...,
polynomial sjr of grade (nl)a- + n 3 + r which vanishes in A,,..., Aq,
or
J?
we have the
result
2Q
Aq, have the strength ^ (t+I) as determinators of adjoint polynomials of grade {nl)(r + n 3 + r.
Ai,...,
S 92.
Q <7ir - 1 +p.
+ 88.
EQUIVALENT
SPECIAL SETS.
134
there are
[93
SETS.
polynomial
iJ-1 of the places B^,...,B such that every adjoint
of grade
For an
place.
vanishing at these places, vanishes at the remaining
(_l)o. + n_3 +
three
through
drawn
be
curve
example of this case we may cite the theorem If a cubic
intersections
mne
remaining
the
curve,
quartic
coUinear points A A^, A, of a plane
cubic through a proper set of eight of them necessanly
i, ..., 5g are such that every
be chosen.
passes through the ninth. In general any set of eight of them may
r,
When
be of grade
+1
(re
( _ 1) o- + n -
is
1)
o-
+w-
3 contains
i/r
itself to
3.
arbitrary coefficients,
that
T+1
When
places A,,
...,
Aq,
so that the
(w
l)o-
+ -3
vanish in
Q-q
as deter-
2p-2-Q
the characters
of the
1)
o-
+ 3.
In
this case
Q-R = 2{q-T),
Q + R = 2p-2,
and
reciprocal*.
Ex. When the strength of a set 4,, ... A^, wherein Q<p, as determinators of adjoint
polynomials of grade {n-\)(r + n-Z, is equal to their number, so that the number of
linearly independent adjoint polynomials of grade (7i-l)<r + n-3 which vanish in the
,
is
residual zeros of
It is
94.
known
that the
number
of placesf of the
c, is
Riemann surface
the same as the
number
of places at
which
has
sets
Math.
AfiTial. vii. p.
283 (1873).
t In this Article, when a rational function g is said to have the value c at a place, it ia
intended that g -c ie zero of the first order at the place. A place where g -c is zero of the k-th
order is regarded as arising by the coalescence of k places where g is equal to c.
CORESIDUAL
95]
For
let
135
SETS.
J.^ be denoted
by G, and be given by
where i/o, ..., k, are arbitrary constants, and fr,, ... (?, are particular functions
whose poles are among A^, .... Aq of which one, say G^, may be taken
to be the function {g c')/(g c).
Then if G' denote any function whatever having poles Ai, ..., A'q, and not elsewhere infinite, the function
G' {g c')/(g c) is one whose poles are at .4,
Aq; thus G' {g c')/(gf c)
can be expressed in the form
,
G'
for
proper values of
iff
- C')/(ff - C) = Vo+flG^^
/^/
Since this
is
is
c
^g-c
q
c')
...,
Ai,
...,
A'q,
obvious.
If the symbol oo
95.
"g-c
g-c
+ VgGg,
grade (n
1) a- + w 3 + r,
of the form
^ = \yjr + X,% +
+ W,
which vanishes in the places Bi, ..., Bjt, gives rise to cjo* sets of places,
constituted by the zeros of ^ other than Bi, ...,Bg, each set consisting of,
say,
places.
We
sets.
is
a residual
clear then that any set, A^', ..., A'q, which is co-residual with
Aq, is equivalent with .4,,..., Aq, in the sense of the last article;
for these two sets are respectively the zeros and poles of the same rational
function in fact if yjr be the polynomial vanishing in Bi, ..., Bj^, A^, ..., Aq,
and ^ the polynomial vanishing in B^, ..., Bj^, A^, ..., A'q, the rational
function S-/-^ has A^, ..., A'q for zeros and A^, ..., Aq for poles. Hence
by the preceding article it follows that the number q+1 oi linear, arbitrary,
It
is
Ai,...,
coefficients in
Ai,
...,
Aq,
is
its
poles at
SETS.
THEOREM OF CJORESIDUAL
136
have
its
sets of places
number
A'q.
...,
multiplicity, this
[95
sets as poles
+ n-S.
of grade (n
i/r,
1)
<j--t-
- 3 4- r,
be
/'(?/)
polynomial,
have Bi,
is
i/r',
...,
of grade (n
B'r for
l)a + n S-i-r',
residual zeros,
its
R being
...,
5V, Aj,
...,
...,
Aq,
+ 2p 2 Q, it
n 3 -f r',
(n 1) o-
equal to nr'
-|-
A'q.
...,
-4^ as its
poles can be written as the quotient of two adjoint polynomials, of which the
denominator
and be of
is
it
must vanish
sufficiently
this.
function ^Z'^, whose zeros are A^', ..., A'q, can be written as the quotient of
two polynomials of which i/r' is the denominator, namely in the form ^V'^'-
The polynomial
...,
as stated.
Hence, not only are equivalent sets necessarily co-residual, but co-residual
sets are necessarily equivalent, independently of their residual*.
The equivalence
97.
of the representations
^/ijr, y/^/'',
form
where /=0 is the equation associated with the Riemann surface and
is an integral polynomial in x and y. Reverting to the phraseology of the theory of plane curves, it can in
fact be
shewn that
if
three curves
/ and
i(f
intersect in
Now
= 0, K=0,
in the case under consideration in the text, if the only multiple points of
H+fi
so that in particular
exist curves ^'
/=0,
is
all
That theory
is
ascribed to Sylvester;
of.
ensures that
be
ifr
98]
137
has a multiple point of order >(; 1 at every multiple point of/ of order k. The adjointnesa
of the polynomials ^, ^' ensures that the compound curve ^^' has a multiple point of
order 2 (it 1) or k + k \ \ at every multiple point of / of order k. Further, the curve
J...,j1j, B^,
A^,
...
there
Bk',
for
passes through
Aq.
...,
fully
is
from which it follows that the curve ^' is adjoint at the multiple points of / and passes
through the remaining intersections of / and ^\f'', namely through A\, ...,A'q and
This is the result of the text.
ffj, ... B'g..
,
In case of greater complication in the multiple points of /, there is need for more care
in the application of the theorem here quoted from the algebraic theory of plane curves.
But
this
351
theorem
may
reader
is
of great importance.
it
the
i.
p.
311 ;
Also papers by Noether, Voss, Bertini, Brill, Baker in
the Math. Annal. xvii, xxvii, xxxiv, xxxix, xlii respectively. See also Grassmann, Die
Ausdehnmigslehre von 1844 (Leipzig, 1878), p. 225. Chasles, Compt. Rendus, xli. (1853).
de Jonquiferes, Mem. par divers savants, xvi. (1858).
p.
From
98.
ixiii. p.
269.
vii. p.
sets, of
set, Sq, of
is less
than p,
= 2p 2 Q common
to
ii
1)
c- -(-
n 3,
-|-
arbitrary coeflBcients)
the
may be
lot, it follows,
may be determined
which have
places,
the
lot,
and in
Hence
R other places.
also,
function, of
when Q
poles, contains
more than
Qp+l
arbitrary
(n-l)o--f-w-3.
i.
is cvpressible
-t-
1,
is expressible
as the quotient
l)a + n 3.
(re-l)o-+i-3.
Ex. ii. For a general quartic curve, co-residual sets of 4 places with multiplicity 1 are
determined by variable conies having 4 given zeros ; but co-residual sets of 4 places with
POSSIBLE DEPENDENCE
138
[98
1, i.e.
by
straight lines.
iii.
The equation of a plane quintic curve with two double points, can be written
are cubics passing through the double points and
form ^<S'-^'S=0, where ^,
seven other common points, and S, S' are conies passing through the double points and
two other common points.
Ex.
in the
Ex.
itself in
to
When r + l
iv.
places, there
being equal
the remaining q places Ap^r, ... , ^e- Let 8^ be a residual of Sg,
Then, regarding S^ and Jp _,.,..., .4 g, together, as forming a residual of
Jp_T-i, it follows ( 93) that there is only one adjoint polynomial of grade
ip-'i-Q.
ilj, ...,
in Sg
and in Ap^,,
...
...
Hence there
Af^.
exists
no
rational
Ap_r-i.
Sg and Ap^r,
It
...
A^^
as
common
Compare
zeros.
26,
Chap. III.
III.,
that
if
j, ... Bt'+i be any t' + I independent places, t being less than t, there exists no rational
function having poles in S^ and JBi, ... Br+i. In fact r + 1 (r' + l) linearly independent
,
<?(,
and
jBj, ...,
Br'+i.
Let
S^,,
where
iJ'
strength of S^, and 5,, ...,Br'+i together; hence every adjoint polynomial of grade
(re l)(r + re 3 which vanishes in S,, vanishes also in B^, ... Br+i.
Now every rational
,
function having Sg and ^1, ... , .Br' +1 as poles, could be expressed as the quotient of two
adjoint polynomials of grade (re-l)(r+re 3 having Sg, as common zeros; since each of
these polynomials will also have ,, ... , .Bt'+i as zeros, the result is clear.
We
is
But
purpose
it
;
it
We
determined hy
are
p + s-1
places,
are
common
zeros of
The formnla
is
n-S
Q-p places.
forming a set
t'
-f-
Q-p - s + 1
or
Q-q=p-(T + l);
1)
o-
+ n - 3 would
q-s + 1
if
q were
places
Q and
by
o-g_,+,.
therefore T
(n-1) a + n -3;
+ l=p,
all
places, contrary to
adjoint poly-
what
is
proved
OF PLACES OF A
100]
139
SET.
the strength
follows
it
...,
Ag^i from
t'
Sji, a-qs+i
<7-p+,_i,
+ w-3+r;
and
determinators
s 1 places
+ n S + r with
we choose
if
Sg, which
zeros in
Ai,
93) that
...,
prescribed to
determinate.
=q + p,
as follows:
is
Consider
a lot of co-
places, in
set
t'
+1
quartic curve, determined by variable cubic curves having seven fixed zeros,
on a plane
is
generally
determined by prescribing 2 places of the set. But if there be one of the sets for which
3 of the five places are coUinear, then the other two places do not determine this set
we
in
an elementary way.
100.
adjoint polynomial,
yjr^,
of grade
Qi
{n
1)
Sq^,
+n S + Vi,
so that
+ iJ=wn + 2p-2.
Suppose
that
places,
jVr._^
Ur-r^X,_r other
grade (n l)a + n
of the polynomials
S+r. A
2 vanish
i^i
and
The product
polynomial of grade (n l)(r + n
grade (nl)<r + n 3 + r which
an adjoint
and Tr-n-
will vanish in all the places forming Sq,, Tr-r Ur-r, together,
the polynomial to have a sufficient number of these
choose
provided
the set Sq, into_two parts, one, T, consisting of
Divide
places as zeros.
the other U consisting of p-{Nr-r,- Xr-r,)
places,
+ {Nr-r,-Xr-r,)
vanishes in
Sg
we
Qi-p
For nr + p-2 is the number of independent zeros necessary to determine an adjoint polynomial of grade (n - l)ir + n - 3 + r.
be not independent of the others already chosen.
t More if the I - 1 places Ai, ..., A,_i
*
cayley's theorem.
140
places.
The
places,
where
sets
and
[100
Qip + Nr-r,,
or
Q p,
= n (r
for Nr-r,
86, Ex.
r^), (|
iii.)
if
1)
cr
+ n ~ S + r,
1)
U
cr
and
Ur-r^ together
+ n 3,
the general
which vanishes in
Sji, will
however,
If,
t'
polynomials of grade (n
adjoint
1) cr + n
(r ri)
vanish in the places U, the products of these with the non-adjoint polynomial
give
t'
+1
l)a + n3
vanishing in
vanishes in S^,
determination
Since
now
and
The
(n
it,
which
complete
we may drop
but
number
+n3+r
p (Nr-r, Xr-r)
1) o- + w 3 + ri, will
the
{nl)a-
( 99).
the consideration of
all
+ 7i 3 + r, which
vanishes in
have a multiplicity
-I-
-I-
We
where
have noticed
Nr-r,
= n(r-
Xr-r.
( 86,
iii.)
that
7-1 ),
^i?ttI+'
For <r = 0,
be equal to
Ex.
.+'-i(-+i)(^';)]-i.
greatest integer in
therefore, the
x.
number jp iW_,_
-|-
Xr-r,
is
immediately seen to
i(7-l)(7-2)-i/,
where y = n-(r r-j), and ^I
finite and infinite places ( 88).
is
the
sum
caylet's theorem.
101]
141
prescribe t
+1
or,
unassigned zeros
if
if t -\-l>0
it is
to
t'
assigned zeros.
For example
points
(p=4)
let
let
Take r=2,
= l.
points, be denoted
by A.
vanishing in
plicity
or
t'+H-2, where
+1
is
zero
if
/>! is
be seen to be r
>
itself.
is
In this case, in
a prescription of
will
3.)
(ii)
quintic, with
^3.2 2,
r=4,
ry
= Z.
Then an
adjoint sextic
t'
this quartic will then, of itself, pass through the unassigned zero.
fact,
all
+ 1+^4,
t'
A and
*i
all
but
or
1,
make
If
sufficient
1^
the sextic,
y\r,
made
to pass
first
p. 25)
101.
In the following
articles
of this
further restriction
i.
was added by
In chapter
adjoint polynomial of grade (w l)o- + 3asa ^-polynomial.
set
of
places
constituted
that
the
by
the poles
seen
have
we
23)
III. (5
'
of a rational function, is such that one of them depends upon the others ;
such
that
every
</)-polynomial
vanishthe
set
place
of
one
there
is
thus (5 27)
'
ing in the other places, vanishes also in this. Conversely when a set of
places is such that every <^-pol3momial vanishing in all but one of the places,
vanishes of necessity also in the remaining place, this remaining place
Or on some
of them.
When
a set
/S is
(x-x)
(x
(x
-y^-
142
[101
+ /i'^
-(-
vyfr,
</)-polynomials
place
and
wherein X,
such that
(iCi)
ft,
are
vanishing in the
places.
We
desire
For
^,
(^,
now
p-3
all
and
arbitrary constants
p-3
are
to find
given places
(x) ^ (x^)
be equal to the ratios
yfr (x^).
fundamental
of
the
help
with
expressed,
thus
equations
the
two
From
equation of the surface, we can eliminate x^, and obtain an equation for a;,, so
that the problem is in general a determinate one and has a finite number of
the ratios
solutions
for Xi is
<^ {x^)
^ {x^)
-f (xi)
4p(p 3)*,
Hence there
and
Eiemann
p1
^3
number of
positions
exist on the
determined from
(f>
them
places such
-)-
= 2, we
obtain,
by the Eiemann-
sets.
are a finite
to
may be
An
elliptic, in
zerof.
* This resalt is given in Clebsch
xvii.) giyes
of the i (p - 1) (p - 2) sets, of p - 3 places, which can be selected from the p - 1 places formed
by S and A, constitutes, with , a set of p - 2 places, and, in accordance with the hypothesis
allowed, each of these sets determines another place. It is assumed that the p-2 places S, and
the place B, can be o chosen that the Up -1) (p- 2) other places, thus determined, are different
from each other and from the p places constituted by S, A and B together. Since the places S are
Each
S and B
is
BY (/(-POLYNOMIALS.
102]
Then
it will
last article,
finding a place
pS
be possible to choose
common
is
143
number
finite
p3
S and
have
(xi),
by
be
be a place not coinciding with any of the positions possible
for a;,, and not determined by 8.
Let (/>, ^ be two linearly independent
^-polynomials vanishing in S and A. Then the general (/i-polynomial vanishing in S and A is of the form X<f> + fj.'^, X and /t being arbitrary constants,
another
Let
so chosen.
zero
let
S,
^i
and the general (/(-polynomial vanishing in the places S only can be written
in a form \(f> + fi^ + infr, wherein f is an arbitrary constant and ^Ir is a
(/(-polynomial so chosen as not to vanish at the place A.
Consider now the rational functions* z=<plyjr, s =%/y}r, each of the
l)th order.
They both vanish at the place A.
(p
-t-
These functions
(s,
z)
= 0,
will
obtained by eliminating
will
place of the
the number being the same for every place of the new surface^f-; since there
is only one place of the old surface at which both z and s are zero, namely
the place which was denoted by j4, it follows that there is only one place of
the old surface corresponding to any place of the new surface. Hence each
of X,
y can be expressed
as rational functions of
z-<^
(f>,
si|f
= ^,
{s,
s, z,
z)
= Ol.
Since a linear function, Xz+fts + v, equal to {\(t>+H-'^ + >^)l'^t vanishes* at the variable
+ V') namely in jo-t-l places, it follows that the equation
z)
= may
p+ 1
the number
it
surface
is
when
always hyperelliptic.
every
p-2
may
be remarked that
It
(cf.
poles,
we can
103).
We
when we
In case p = 2 or 1 the
p-2
III. 31)
places
that in
general a rational function of order p does exist. The reader may prove that for a hyperelliptic
any rational function of order p.
surface whose deficiency is an odd number there does not exist
* It must be borne in mind that, in dealing with a rational function expressed as a ratio of
by the zeros of the denominator; some
two adjoint polynomials, we speak of its poles as all given
may be at a;= (cf. 86), and in that case their existence is to be shewn by considering
instead of the polynomial, ^, of grade /i, the polynomial in ij and {, given by f**^. Or we
= we may, in the text, have ( 92,
use homogeneous variables ( 85). For instance, forp 3,
= 1. Then <p -.^ : = l : V i=<'> ^ 'i and ^ has a zero at = oo .
of these
( 84),
may
Ex. i.)<p=x,^=y, ^
t Chap. I. 4.
X Or by the
<l'
I.
THE
144
Sp-3
IRREMOVEABLE CONSTANTS.
[102
of its double points will, therefore*, be ^p {p - 1) -p, or \p {p - 3), though it is not shewn
here that they occur aa simple double points. These double points are the transformations of the pairs of places, (xj), (x^), on the old surface, which were such that every
(^-polynomial, vanishing in
thep-Z
and
fixed places S,
Since a double point of a curve requires one condition among its coefficients, and the
number of coefficients that can be introduced or destroyed, in the equation of a curve, by
has
general linear transformation of the coordinates is 8, it follows that a curve of order
^m{m+Z)-(8 + K)-8,
and
places, S, of each of
z is
new
new
linear functions of
^,
<f>,
i^,
instead of
i" order to
4>t ^j
there are, in the
>^>
p-3
surface
(s,
Hence
z) efiectively
3(p + l)-9+p-{p-Z),
that
is, 3;;
(Chap.
I.
103.
intrinsic constants
this is in
7).
The p
places
S may
III. 31) {p
l)p{p + 1)
infinite only
<f),
The p
vanish in the
p2
places
variable intei-sections of
it
will,
for
= 0, = 0.
We
X<j)
+ fj,^,
which
7ijz
+ fis = 0,
which pass
^:
enquire
||
1, bo that ir=0.
And we have remarked that
responding to a double point of the curve, so that S + it = 4Si
of dimension
i is
ef.
1 at
182.
order.
II
103]
145
the differential of the fun&tion ^/^, of order p, vanishes to the second order,
namely * 2^ + 2p 2. Among these tangents, however, there is one which
There
therefore,
are,
3p
points, counting
Of the
other tangents.
p 1
as
'ip
tangents.
distinct
tangent
thus obtained, there are 3^ 3 distinct cross ratios, formed from the
3^3 distinct sets of four of them, and these cross ratios are independent of
lines
any
We
and
z.
3jl>
draw through
it
first
double points,
^p^ + |jo+
2.
-i-
there are five other constants in the equation of the curve, namely, those
0.
a linear transformation.
The reader
7,
Chap.
I.
will
is
The
g.
new
surface,
]}
and the
poles of
loss of definiteness
and
conviction.
p2
clearly
fails
for
<j>
and
pth order.
meet
the curve again in 0, eight other tangents to the curve can be drawn from 0. The cross
ratios of the six independent sets of four tangents, which can be formed from these nine
tangents, determine the curve completely save for constants which can be altered by
projection.
More
drawn.
generally, from
The
any point
is
projection.
* Chap. I. 6.
+ Cayley, Collected Works,
B.
vol. vi. p. 6.
Brill n. Noether,
Math. Annal.
t.
vii. p.
303.
10
SPECIAL SETS.
146
104.
It
is
common
^r
= ^/<f>, where
are
^,
<^
take
last Article to
And
dependent
[104
may be convenient
it
to take another
variable.
In the process of Article 102, the fixed zeros of the polynomials used
p S in number, and a double infinity of (^-polynomials vanish in the
places of the
set.
These two processes are capable of extension. If we can find a set Sq,
Q places, in which just (t + 1 =) 3 (^-polynomials vanish, and if the places
Sq be such that these three (^-polynomials have no other common zero, while
the problem of finding a further place x^ such that the two ^-polynomials
vanishing in Sq and cc^ have another common zero x^, is capable of only a
finite number of solutions, then we can extend the process of Article 102
of
sheets.
unity,
and the equation such as represents a curve of order 2p 2 Q. If, therefore, we can find sets Sq in which Q > p - 3, the new surface will have a
less number of sheets, and therefore, in general, a simpler form of equation,
than the surface obtained in 102.
Similarly, if
(t
-I-
we can
find
=) 2 (^-polynomials, say
set,
^ and
Sq,
(f>,
common
zeros
of
to obtain a
Q>p-2
in 103.
We
105.
what are the conditions that t -fany Q places Oi, ..., ag.
where
equations
(i=l,
\^(|>^iai)+...+\p<f,J,{ai)=0,
p-r-l
2,
...,Q)
>p- t - 1, and
p-r-l
-(p-T-1)
(^j
(ap_,_,+)
= m,
<f>j
a from
(a,)
1
+...
to
+ mp__,
Q -(p -t same
for
<f>j
(j=l,
{ap_r-i),
1),
the values of
every value of
j.
2, ...,
wij, ...,
The
set,
p)
mj_,_i
of p, of
105]
SPECIAL SETS.
a has any
147
1 equations,
of the form
<^l(l),
</)i(ap_T-i)
^J)-r-l(ai),
^j>_T-l (l)-T-l)
<^p-T-i+t (oi),
arising for
A;
1, 2,
. .
</>i
<^j>_T-l (ap-T-l+ir)
<^p-r-i+t (a^,_^_i),
^p_T_i+i (ap_T_i+)
1.
is
=0,
(ftp-T-i+a)
1) such equations*
1) equations
it is
in
The determination
<?
1) other places,
such
places.
of the q{T-\-\) places,
possible to determine 5 (t
}(r + l)
/>
t 1 [ ^(t + I)],
pT 1
places S, T,
places S,
T are
all
(?
j(i-+l)
places,
or qr, places, T, to
may
be
add
to
or
5,
other
common
zeros.
The^ t 1
Q=p- 1,
r-)-l
= 2,
it
follows that q
=\
and
Q q{T + \) p Z, and
p\ places in which
The problem of
two 0-polynomials vanish, and, of these, p Z places are arbitrary.
determining the other two places may be conceived of as the problem of determining
p T 1 [ J (t-)-1)], or one, other place, to add to the/) 3 places, .such that all 0polynomials vanishing in the resulting p 2 places, which are independent determinators
We have already seen reason for
of (^-polynomials, may have ^=1 other common zero.
believing that, when thejo 3 places are given, the other two places can be determined in
hence, from the theory here given,
I;)
(p- 3)
it
we can determine
follows that
ways.
To every
set of
of
lot of sets
places, the
greater than
o
arbitrary.
Q _ ^ (t
-t-
the
Q-qij+l),
number,
of
> 9.
From
this,
the original
set,
which are
^ 5 (^
*
-t-
1) (t
1),
Q>1+P-^\-y
...,
function.
102
NORMAL EQUATION.
148
Putting
Q - g (t +
i5
1)
= g + a, we
=2p- 2-
obtain
= (T+l)(9+l)+,
set
Q, places,
[105
1+P^ + -^-
Q=
which T
If
now we determine,
=3
in accordance
(^-polynomials vanish,
it
common
(^-polynomials vanishing in Sq
(/)-polynomial vanishing in
-yjr
s is of
dimension
-I-
regard to
Since
or Stt
1 in
1,
or Stt
z.
value of 22
is
>
-f-
Thus writing p
^p.
-f-
is
= Sir,
+2
tt
in all cases.
From
R=p + 2
follows
Tr
Q=p
4^
-tt
thus
q= Q - p +
tt
1,
And, in a surface
3m+p-9
3(;)-7r
and
+ 2)-f-p-9-(p-37r)
or
3jo
Ex.
= 5,
is
i.
The values
of
n-
for
p=4,
5 are
1, 1
respectively, an(i
p w + 2,
Chap.
I.).
in those cases,
6 respectively.
Hence a quintic curve with two double points (p=4), can be transformed into a
this will also have two double points, in general, since the deficiency must be
\maltered.
We determine a set consisting of Q, =1, quite arbitrary place. Let the
quintic
ANOTHER FORM.
107]
149
general conic through this place, and the two double points, be X<^ + fi^ + i/\//- = 0. Then the
formulae of transformation are ^ = <^/ii', s='^lij,. As in the text, we may suppose (p,
to have another common point, in which yfr does not vanish.
Er.
ii.
in genonil,
^(6
1) (6
2) 5 = 5
(;^
= 5)
double points.
if \<t) + fi^ + v\(r be the general conic through the two points and the double point,
the equations of transformation are c = (/>,i//-, s='d/-\jf.
jwiuts
Ki:
iii.
Show
p ir + i
by this methotl to
p = 6,
= 6,
are i-espectively
But, as remarked
107.
which t +
for
We
places,
if
2, to
9
8.
we oan
also
make
ii
places.
It is
arbitrary places,
;^
where
8,
8.
from
( 104),
7,
7,
f^
= ^q
or i(<?
thus obtainable
is
[x)[-2p-l-B-q-l)'
^'
lX according as
q is
denotes
\(X-l)...(\-.'+l)V!.
For instAnce with 7?=/', q = 0, the series reduces to one term, whose value
which is clearly right while, when B=p-l, 3 = 1, the series reduces to
is ^)
1,
or i/)^>-3), as in 101,
When p
and
is
10-2, p.
even and
IS
144.
R = ^p + 1, q = ^p-l.
be summed,
equal to
2
When p
is
summed, and
odd and
is
jR
|p-l,^|ip-l ip+1i (P
1)
1.
= i (P - 1) "
1-
equal to
Now
let \<f>
ij.'^
places,
residual to one of these sets of
r =
with
surface
we may transform the
^>
MatK
p. 296.
150
[107
The function
the constants X,
the curve
2R + 2p-2
of the
The
Chap.
7,
be written equal to
If then
p
may take*
when dz vanishes
Compare
quite arbitrary.
is
/x^
fi
R = p-Tr + 1,
which are
I.
that
is
or 2Tr
2tt,
is
ip
+
+l
R > t + pt/(t-{-
according as
1,
is
or
1).
even or odd, we
as
even or odd.
Hence, when
sets oi
^p+
which have
even,
is
then be determined in 2
|p 3 common
places,
we can determine a
+ /t^,
zeros.
place,
|jd
^p +
\</>
with a given
and
therefore
no two of these sets can have a
1,
Let the sets co-residual to
place common without coinciding altogether.
being chosen
these two sets be given by \(}> + /x^ = 0, X'0' -1- /u,'^' = 0, ^ and
have no other common zero.
so as to vanish in A
we assume that
1 places will
not be co-residual
<f>'
<f>,
adopted.
is
= <^/^, s =
(f)
(f>'/^'
will
enter to a degree
^p
1,
and, since
^p+1, no
further
is possible.
p+2
it
will
{Z, l)"'"'
wherein
By
m=^p +
!x.
+ (2,
If
it
fjL<f>2
^p + 1
= 0,
v<}>3
ylr,
The
^='f>il^.
*
1)'
we have a
+ ...-(-(z,
it is
108.
l)'"-'
'l.
F=<^,/t.
2=<f>,/^fr,
ThuB, for perfectly general surfaces of deficiency p, no rational function exists of order less
Biemann, Oetam. Werke (1876), p. 101.
p. 460.
1<)9]
and eliminating
151
number
To a
\X +ftY+vZ+l.
F=
point
= X,,
Fj, Z=Zi of the curve in space, will correspond the
places of the surface, other than the fixed zeros of
^,, (f) <^^, f, at which
and
it is
is
given by
jP
TT
p=
+ 3.
iir,
or
47r
1,
or iir
2,
or 47r
3,
may be taken
For instance with*/) =4, R=6, taking a plane curve with double points
x = tx3 ,y = and x=Q, y = <x>, given by
equal
at the places
\<))i+ii4>2
B places
are, in
number,
2p-2-6, =0.
A'=xi/,
Z)
F,
of transformation are
Y=x, Z=y,
X= YZ,
X^{Y,
of which the first
set of
notation
we may put
z=
(j)i/->fr,
A set
of
Z = 0,
may
Riemann
^'J-'fr.
Z\+A=0,
is
It
is
is
surface.
clear
a curve of order
It has already
practical
method
* Since p must be
5 (t + 1) (? + 1), this is the first case to which the theory applies.
+ It is easy to shew that this is the general adjoint polynomial of degree n - 3. We may also
shew that the integrals, jxydxlf'{y), etc., are finite, or use the method given Camb. Phil. Tram.
zv. iv. p. 413, there being no finite multiple points.
[109
HYPERELLIPTIC CASE.
152
any place
find
5-
2p
Q=
Putting
denote hy S.
we
Suppose
54).
p>l.
that
first
we
shall
=p-T- 1,
andT+l=p-l
1,
()i
places
to
) o-
- 2
ft
will contain
+ p1
We may
zeros of
<^i
oi
?i
+p
two zeros of
Then the
<^i.
may be
s=0
of the
associated
new
values of X,
of
z,
namely,
= 0,
^+
is
of the
new
If
<})i/(f>2,
-\jrilifr^,
and z together,
v,
and therefore
the zeros of
if
if
-y^^
are
= 0,
p+
in the equation
new
(s,
Thus
The highest
= 0, is the same
surface.
1 in z.
z)
jjls -'t
v,
for arbitrary
y\r^
(5,
z) in s
and z together
will
where
assume that
shall
= -^il-\jr^,
be
form
fx,
functions
z)
(s,
and we
S,
-^i
z)
number
to
(fi^/cpi,
aggregate degree in
as the
2p
is
(s,
prescribed
suppose
?i
is
jjL^
fi^yp-.,, fj.^,
places.
surface
in
all,
vanish in a set
fii-\frj
a, /8,
+ s/3 + 7 = 0,
we put
cx
= sa + ^13,
p+
1.
form
a = id' - 07,
which
Ex.
is
A plane
and
A',
(ii)
and three
fixed
When p
is
or
0,
2p-2
variable places)
110]
have no variable
Ml Ma.
.
which vanishes in
other
common
being a function
^+2
n+p
zero
transformation z
in
zeros.
arbitrary,
/"^
places,
4!
beside
these
n+p
ilrj
we obtain an equation
+ s(^,
use the
is
of the form *
lW +
(^,
l)p+.
= 0,
immediate.
the general conic through the double points and a further point A,
also to vanish at
we may
fixed places,
4!
Ex.
153
Then we may
any point B.
let ^iV'i
and
i/'i
+ /^2''^2"''F3^3
'^^
being chosen
= ^i/>i'3, s = 1^-2/^3
places
except
and define
Ci, ..., Cp
but we
(pi
all
of
^w
...,
<t)p,
Ci, ...,Cj,
Let any
of degree
p{p + l)...(p
/x,
be
+fjL-l)jfi\
terms.
In a polynomial
^f'"
<j>,:<l>2- ....<f>p
hep -2
= l -.z-z'--
-z^';
H'Pi = 'f>3l<f>i
. . .
<f>pl<t>p-l
* Further developmentB are given by Clebsch, Crellc, t. 64, pp. 43, 210. For this subject and
this Chapter, the reader may also consult Clebschfor many other matters dealt with in
(Paris 1883), t. ni.
Geometrie
la
sur
Leions
Lindemann-Benoist,
+ Chap. n. 21.
LINEARLY INDEPENDENT
154
^p{p + l) (p
reduced to
[110
2),
But we can
at most.
in
<|),,
fact
...,
(ftp
shew,
whether the surface be hyperelliptic or not, that there are not more than
3(p 1) linearly independent products of pairs of <pi, ..., <f)p. For consider
the 4 (p 2) places in which any general quadratic polynomial, <i>^^^, vanishes.
If
(f>i(j)j
<f>i^jl^'-^
among the
zeros* of
<I>'^'
many
among
only 4<{p
~1)p + 1,
number
the same way it
more than
precisely
=3(/>
this
But
111.
<^i, ...,
^p
can
it
<I><^',
more than
of the polynomials
contains
Hence there
<^i,
...,
are not
...,
(2/x
<f>p.
In
l){p l)
(ftp.
linearly
1)
1) zeros of
arbitrary constants.
1),
(2/i l)(p
the 4 (p
linearly
as there
..., <^p.
<f>i,
<!>'''',
independent products of
/j,
of the
polynomials
ft
..., <f>p.
^i<I>f"
is
quadratic in
<f>^,
..., (\>p.
</)2<I>'l'l
It contains
2p terms.
<^2(/)i
But
clearly these
occurs both in
terms
0i<E>i'i
and
(/)2<I'''".
for this,
shews that
But
^'i'
since these
vanishes in the
zeros of ^^
which
is
* Here, as in all similar cases, the zeros of the polynomial are its generalised zeros
is
regarded as of
oi
p-2
its specified
</>,.
2 places)
when
it
grade.
t Precisely, the theorem is true when the surface ia sufficiently general to allow the existence
places such that the general ^-polynomial, vanishing in them, is of the form X^j +
zt^,,
\ and ^ being arbitrary constants, and ^j, (p^ having no common zero other than the p - 2
places.
We have already given a proof that this is always the case when the surface is not
hyperelliptic ( 102).
PRODUCTS OF ^-POLYNOMIALS.
Ill]
155
vanishes in these
-1
contains 2p
linearly
</)i,
..
^p.
<^i,
independent.
...
-I-
-I-
Xj,<f>j,l
wherein X,, ..., \p are arbitrary constants. Herein <f>3(^3<f)3 + ... +'kp4>p)
cannot contain any terms <f>3{\3'<p3+ ... + \p'ipp) which are equal to terms
already occurring in the part
+ <^2*'"',
<^i<t"i
p 2 common zeros
that \<pi +
is the
vanish in the
hypothesis
these
3p
fitf).,
2 places.
of
<^i
or else
and
<^2
\3'<^3
'^^'^
-I-
...
would
\p'<f)p
2pl+p 2,
contains
or
As
<f>i, ...,(fip.
number
3
is
the
3p
3,
we have proved
that a greater
exist,
have
a multiplicity
2p3
the
that ^'='
2p l
of the form
common
in a polynomial
(fjp,
2p
is
^=2,
zero,
^i<l>i^'
of terms, ^2^''''. i
^1^*^',
...,
-I-
places,
first
at most equal
contains at least
of the
degree in
and
If then
arbitrary coefficients.
it
<^i, <^2,
follows
...,
(f>p.
be ^-polynomials
the reduction in the number, 2(3^ 3), of terms
^2*!'''^', which arises in consequence of the occurrence
it
^2'^''"".
5p 6
places
in
= 3.
it
is
with one
/a
p 1,
to p.
if ^1,
to
(f)^
terms,
</>i^"'',
occurring
</>,<!>"''
</>i,
<f)2<i>'^^^
...,
(/)j,.
Hence taking ^3, and a quadratic polynomial O"''', such as do not vanish
in the common zero of (pi, <^2, it follows that a cubic polynomial with at least
5p-5 linearly independent products, is given by
c^a^l^'
We
<^2*'l^'
-I-
^3^"'''-
in the cases
/*
= 2,
/t
= 3,
the polynomial
Assume now
contains (2/4 l)(jo 1) linearly independent products.
and
that *<^-'i
terms,
^^-^>
independent
(2/i-3)(p-l)
contains
that
OW
*
From
93)
Q-R = 2{q-r),
Q=p-2, R=p, t=1, we obtain q = 0.
From g-g=i)-(T + l), putting T + 1=0 (because 2j)-l>2i)- 2) Q = 2p-1, q=p-l.
putting
156
contains
(2fj,
5) {p
independent terms.
I)
[Ill
variable zeros;
when
places,
is
fj.>3,
{2fj,
- 5) (p -
1)
1,
which by hypothesis
^'"-''i
has
qrip.-i)
its
+
have no common
polynomial,
(^,<1)'''-"
<^2, is
is
^j'*''''-^',
the same
in
which
possible.
If then, in a
degree in
of the /xth
<^24>''''-",
polynomial
(^1,
...,
(f>p,
wherein
be of the form
\if(u2)_
(j)o'^V'i^'''\
2{2f,-3)ip-l)-(2,,-5)(p-l), =(2/x-l)(p-l),
+ <p^^''i^-^^ as we have
that {2/x l){p 1) is the
number
exists, it follows
of the polynomials
/u,
c^,,
...,
<f>p.
is
And+
ibid.
t.
xxvi.
"Ueber
p. 143,
die
Normalcurven
furjc>
= 5,
6, 7."
is
If
= i/^i
-^3
V'4,
When
/x>
1.
for
113]
157
then the general rational function with poles at the sixteen zeros of a polynomial, St'PI, of
contains 14 homogeneously entering arbitrary con\^i, 1/^21 '/'sj V'4>
stants.
Now there are only ten terms in the general polynomial *(^l, of the second order
in yJAi, ... \^4 and these are equivalent to only nine linearly independent terms, because
the second order in
the form
Hence the
*(2)/*<2).
The
113.
110
to in
112,
which
two
dimensions.
Let
(?i
(f>i,
1)
he
Ci, ..., Cp
all
of
...,
cr -I- ji
c,
<f)p
places on the
p independent
...,
except
Cp
suppose,
defined,
3,
Let
Cj.
Xi,
We may
x^
...
Xp
is
(f>i
...
<f>2
'
<j>p-
all of X,, ...,Xp are zero, and, since only the ratios of these quantities are
defined,
we may suppose
Hence we may
that none of
interpret
x^,
...
them become
infinite.
homogeneous coordinates
we may call this the point x.
Xp as the
p-l dimensions
Corresponding then to the one-dimensionality constituted by the original
Riemann surface, we shall have a curve, in space of p-l dimensions. Its
of a point in space oi
order,
</>i(c)
<^2(c)
<^j,
(c)
= <^i (c')
infer
<t>,ic)
...
that the
first
(i>p{c').
(^-polynomials
corre-
same mutual
where X,
fi
c'
is
definite,
and
F^'
are normal
zeros.
Hence the
T*',
a form
x,:x,:
...
Xp=l
Chap.
z'
II. 21.
-.
,
^^-'.
:
158
[113
To any point x of the space curve corresponds, therefore, not only the place (, z) of the
Eiemann surface, but equally the place ( - , z). The space curve may be regarded as a
doubled curve of order js
Leipzig, 1890,
t. i.
(Cf. Klein,
l.
Vorles.
iib. d.
Theorie der
ellvp.
Modvlfunctionen,
p. 569.)
p=
3,
the curve,
case, j) = 4,
C^p-.^,
the ordinary
is
^p{p \-\){^p
^p {p + V) {jp + 2) (bp
and
is
5),
5,
a sextic
2),
1,
linearly
If,
= 4, we
equation f
l\_
= Z-XY=0,
by
l\ = Z^{LX+MY)+Z{aX^ + '2hXY+hr'^) +
XYhy
surfaces,
PX^+QX^Y+RXY'^
+ SY^ + AX^ + 'iHXY+BY-^ + CX+DY+\=0.
Vi = 0, V.^ = 0, T'4 = 0, can be obtained from ^'3 =
r,=0,
Z, respectively in,
(i)
the coefficient of
by
XY
one another. Other cubic surfaces can be obtained from ^3=0 by replacing
by .Z'in
two of its terms simultaneously for instance, if we replace
by Z in the coefficients of
h and H, we obtain a surface of which the equation is F,- U^+ Vi^O. Similarly all
others than ^3=0, Fi = 0, ...
I'4 = 0, are linearly deducible from these.
XY
its
order
when
contains
5 given points.
(/x
3,
(/Lt
arbitrary,
homogeneously
+ 3,
*<''>,
3)-(6/i-.5)
entering,
111 preceding.
Cf. 108.
X Where
(ix,
v) is
l)/>'!.
115]
fi
C/i
159
^^
^/^
+ ^~^^ -(2m-1)(p-1).
(^-3,
3) -(6/. -3),
6fi
or
Denote the
equations by ?7i = 0,
Ur = 0.
=
surface, <I>'^'
0, vanishing in the
.
wherein Xj, ..., \., \, fi are arbitrary constants, and fJ is a surface of order fi,
other than ^i^*, which vanishes in the 6/i 5 points, and does not wholly
contain the curve.
same
are the
The
X'^''''
+ ^Z7=
<!>''''
right
the
involving
=0
4!
is
arbitrary
of
constants.
Ex.
There are sixteen of the surfaces \'irM + nU'=0 which touch the sextic
i.
Ex.
z=
Uj-iW,
is
(in points
fixed points).
differential, dz, of
the rational
ii.
by
We
115.
place.
But
the
k-\
* 111.
is an immediate
+ The canonical curve discussed by Klein, Math. Annal. t. 36, p. 24,
includes other cases also.
generalisation of the curve Cjp_j here explained. But it
Higher Plane Curves (Dublin 1879), p. 22, "on an apparent
X See the note in Salmon,
there given, which include a reference
contradiction in the Theory of Cunres" and the references
consideration of curves in space see Appendix I. to
further
For
1748.
date
Euler
of
paper
by
to a
160
r,
others,
We
take
first
being independent of
r.
m and
The curve
[115
n,
Um =
say
0, t/
= 0.
is
such singularities
only
as
\mn (m
7-mn
For when,
firstly,
>m
-(-
4-
4)
is
1.
the intersections of
1,
J7,.
with
the
^r
^ in
'
^n
riii
^ )n^ n
rn
'
'
r~^mn ^^ ^j
wherein F,_,, F^^i, Fr_,_ are general polynomials whose highest aggregate
order in the coordinates is that given by their suffixes.
Hence, in analogy
with the argument given in
(r
3) coefficients in Ur,
3,
/f=(r-m-|-3, 3) + (r to
zero,
however
may
81, it
we can reduce a
71 -t-
3,
number, K, given by
certain
3)-f
(r-
m-n
-I-
3, 3),
is
we can
u u (V
And we may
of Fr_i_.
(r
m.
+ 3,
ji
Tf,._,_
^-o
T,.-,n-n are
TJ
JT
to zero.
Hence the
effective
2 (r
711
equations by which
coefficients in
-I-
3,
Ur to (r
in number.
3)
3,
equal to zero.
(r
3,
3)
(r
its coefficients
- m + 3,
3)
- (r - M
-I-
put
U^
are effectively
3,
3)
-1-
(r-
- m - rH- 3,
3)
116]
THEOREM OF SEQUENCE.
more than
certainly not
that
is
161
number
the
of effective coefficients
this.
rmn \vin {m + n-
4).
- \mn (m + -
r>m
4)
m + n -2
or
is
m + n - 3,
namely, than
(r
By
3,
3)
- (r-- m + 3,
number
is
3)
- (r - n+
3, 3).
equal to
S)/3l
result
If
when
^n {m +
number
of this
Um = 0,
17^
- 4) +
Hence
Corollary.
r,
(r
+ 3,
3)
mnr
it
when
follows,
{r+'i, 3)
rmn
upon
by P, the result
+P1
m + n 4.
r?nn
P+1
is,
that
the inter-
the prescription
> rmn
P + 1,
that
linearly independent
i.
P of
surfaces of order
P= 1
every surface of
order r drawn through 4r quite arbitrary points of the curve entirely contains the curve
mined by4r
This
is
the
of them.
number
When
= 2,
r,
the
number
{r
+ S,Z)-rmn + P-l
is
equal to
2.
ii.
the result of
116.
order
111.
11
162
m > 3,
?i
Now we can
(m + M - 1
>
3)
- (n -
- (m - 1,
3)
3),
3)
2 or
= J mn (m + w - 4) +
1,
=^m?i(m + -
number
4)
P is the
number
concerned
The
others.
total
The analogy
other words,
in
P1
as before
of these polynomials of
is
number of intersections
mn (m + n 4), = 2P 2.
order m + 4 with the (^-polyis
ji
obvious.
If now, the
117.
is
are
= P.
1,
equal to P.
Hence
m > 3 and w = 3,
(m + n-1, 3)-(m-l,
Um+n-i
firstly
3.
is
Here we
suppose
[116
space being denoted by X^, X^, X^, X,, the symbol [i,j] denote the Jacobian
d{Um,
+ dX X, + dX X, + dX X, + dX,)
^'^'^'^dxr '^dxr
^~r
'
'^
-^'dX.^^'dXj^'Wl^^'
and the similar equations holding
same
dX,-
denote
it
as the ratios
is in
fact constituted
U^V^,
U3V1
results,
U1V3,
we
UiV^
U^Vi
U^Vi
W4D1
U^Vi
MjWj,
respectively
U^Vi
by
U1V3,
row
u = itiZ,
-I-
-1-
W4X4, du
etc.,
^17]
we
Vi
FINITE,
determinant
163
164
Ex.
i.
[117
n,
vfe
i>n-i {udv-vdu)l{uvf),
u^x^+u^x^,
where/(a;i, x^, Xs)=0 is the homogeneous equation of the curve, u^UiX^ +
and {uvf) denotes a determinant of three rows.
etc.,
Shew
ii.
i/,
where
X, Xj, ...
X, are arbitrary
[
yj/-
Retaining
118.
udv vdu
still
0, v'
of the form
is
u=
= v.
ZU-V=0,
SU'
= 0,
-V'
U,n
0,
Un =
0,
wherein
z,
Riemann
v/u, s
v'/u, of the
Riemann
surface
to
any point
will
have
mn
sheets,
the places, at which z has any value, being those which correspond to the
places,
intersects this
+ 2p 2
branch places,
These are the places where dz is zero
of the second order.
Thus they correspond to the places, on the curve in
space, where udv vdu is zero to the second order.
We have seen that these
are given as the intersections of this curve with the surface (mw 17^ Z7) = 0,
of order m -(- n 2
their number is therefore mn (?n + n 2) = 2m + 2P 2.
Hence the number P, obtained for the curve in space, is equal to the
curve.
Riemann
deficiency
The same
order,
and
And
of the
u', v'
surface
when
with which
u, v
it
reversibly related.
is
Riemann
it
same
as those
in
We may
is
a particular case
is
that
^^1
REFERENCES.
Ex.
Prove that
if
be such that
d^, 63
J 55
+ e^ + e^ = 0,
formT Z
r"*^*"'"''
=y V6^c [(^ - e,f ~ (e, - e,) {e, - e,)] V^^^ [(x - e,)^ - (e, - e,) {e, - e,)]
:
Fina
X,
terms of X,
in
[i
F,
.^,
T in
(2-3)-i7V6^+e2(e3-e,)r/V?^ + e3
As
the form
t.
{e^-e^)ZI^'^i:^'\lx
Zl'^'^TTb
e^)
Tjy.
xix. p. 507.
already remarked
is
16, p. 47
Clebsch, Crelle, t. 63, p. 229 Clebsch, Crelle, t.
Salmon, Solid
64, p. 43
Geometry (Dublin, 1882), p. 308
White, Math. Annul, t. 36, p. 597
Cayley, Collected
Woris, passim. For the more general case, in connexion however with
an extension of the
theory of this volume to the case of two independent variables,
the
;
may
p. 293,
and
Journal de
Cayley,
I'Ecole Polyt.
Collected
and
1885, 1886
Ex.
t.
Works,
t.
(1882), p.
lii.
Vol.
v.
p.
613,
t.
2,
136 (1883)
Halphen,
Noether, Ahh. der Akad. zu Berlin (1882)
\
etc.
t.
ii.
p.
Journ.
de Math.
1889.
Prove that
i.
where(r,
;j)
+ i2 + ...+TOi_,-i-l),
aU the values
= l,
2, ,..,
{k-
1),
to every pair of
two
unequal numbers chosen from the series mj, m^, ..., m^^.^, and so on. Hence infer that
of the intersections of a general curve in space of k dimensions, which is determined as the
complete locus common to k\ algebraic surfaces of orders m^, m^, ,.., ii_i, with a
surface of order
r, all
but
JiiTO2..,wii_, (nij-l-TO2-l-...+Ji_j
are determined
by the others.
k=Z.
The
result is
known
i; 1) + 1
to hold for
k=2.
We
Ex.
ii.
With the notation and hypotheses employed in Salmon's Solid Oeometry (1882),
291) (see also a note by Cayley, Quarterly Journal, t. vii., or Collected Works,
517), where m is the degree of a curve in space, ?i is its class, namely the number
Chap. XII.
Vol. V. p.
(p.
number of
its
line,
j3
is its
number of osculating
the number of points through
a
is
the
which four consecutive planes pass, x the number of points of intersections of non-consecu-
[119
EXAMPLKS.
166
an arbitrary plane, y the number of planes containing two nonpass through an arbitrary point, h the number of chords of the
which
consecutive tangents
curve which can be drawn through an arbitrary point, g the number of lines of intersection
of
of two non-consecutive osculating planes which lie in an arbitrary plane, ^ the number
tive tangents
which
lie in
tangent lines of the curve which contain three consecutive points, prove, by using PlUcker's
equations (Salmon, Higher Plane Curves, 1879, p. 65) for the plane curve traced on any
plane by the intersections, with this plane, of the tangent lines of the cui-\'e in space, that
the equations hold,
(1)
?i
= )-(r-l)-2j-3TO-3^,
(3)
r=n{n-\)-2g-Za,
(2)
= 3r(;--2)-6:r-8(m-(-Sy),
(4)
m+'is = Zn{n-'i)-Qg-ScL,
/Ji-l=|)-(r-3)-.i'-m-^ = |M(M-3)-^-a
jUj
(A),
in space
(5)
(6)
= OT(m-l)-2/i-30,
= 3m(m-2)-6A-8/3,
^-)-)i
(7)
(8)
^ = 3?- (r-2)-6y-8
(^-t-n),
(^-(-),
p^-\=^m{m-^)-h-^=^r{r-Z)-y-n-'^
(B),
From
the equations
(1)
and
we can
(7)
infer
therefore
y-^n = x-\-w..
Hence jOi=p2.
Ex.
For the non-siugular curve which is the complete intersection of two algebraic
ji, v, prove (cf Salmon, Solid Geometry, pp. 308, 309) that in the notation
iii.
surfaces of orders
of Ex.
ii.
here,
/3
= 0,
m=iiv, ?-=^i'(/j4-j'-2),
This
is
Ex.
the
ii.
h=^
(ii-l) (v-1).
/ii/
iv.
ii.,
by
p,
(5)
and (B)
that
6(p-l)=m(jn-7)-2A-(-2r=3 {r+p-2m).
Hence shew that if, through a curve C of order m, lying on a surface S of order /i, we
draw a surface of order , cutting the surface S again in a curve C" of order m', and if
p, p' denote the values of jo for these curves C, C" respectively, then
m'(pi
+ v-4:)-{2p'-2)=m(ii+v-4:)-(2p-2)
Salmon, pp. 311, 312). Shew that each of these numbers is equal to the number,
of points in which the curves C,
intersect, and interpret geometrically the relation
(see
i,
i+r+S=m{fi. + v-2).
Ex.
or
V.
il+p',
curve
C are
If in Ex.
iv.
a surface
<^
of order
/t -)-
- 4 be drawn through
(/*
-I-
4) m' p' + 1,
of the points of the curve C", prove that, so far as its intersections with the
C entirely.
</)
contains effectively
terms.
<*
119]
EXAMPLES.
Ex.
167
vi.
vii.
i/
Then
C"
is
coordinate planes can be chosen so that the homogeneous coordinates of the points of the
cubic can be expressed in the form
T=\
ff^
ffi,
6 being a variable parameter. For instance (using Cartesian coordinates) the polar planes
of a fixed point (X'Y'Z') in regard to quadrics confocal with Z2/a+ Y'^lb + Z'^lc = \ are the
osculating planes of such a cubic curve, the coordinates of whose points are expressible in
the form
Ex.
a
For the quintic curve of intersection of a quadric and a cubic surface having
we obtain, from Ex. iv., putting m'=\, p' = Q, m = 5, that /) = 2 the
viii.
common
generator
results of Exx.
be taken to be
S,
T of the
denote polynomials
where the Greek letters aj, oj,
Hence, if tr be defined by the equation,
. . .
we
obtain
of
I,
o-2
= (f,
l)e; f,
tr
first
degree in x, y,
and
and
x, y, z are rational
functions
<r.
Ex. ix. Prove that if the sextic intersection of a cubic surface and a quadric surface,
break up into a quartic curve and a curve of the second order, the numbers p, p' for these
curves are p = l, ^^' = or p=0, p'= - 1 according as the curve of the second order is a
plane curve or
is
lines.
In analogy with Ex. iv., shew that the deficiencies of two non-singular plane
curves of orders m, m! are connected by the equation
Ex.
X.
TO(m-Hm'-3)-(2p-2)=7rem' = OT'(i-fm'-3)-(2/-2),
order m-(-m'-3, be drawn
and further in analogy with Ex. v. that if a plane curve, of
order m\ only ^ - 1 of its
of
curve
the
of
points
7'
-^'
-1- 1 independent
3) to'
through (m -f
m can be prescribed.
of the theory of curves in space with the subject
connexion
Further indications of the
I.
Appendix
in
found
be
will
of this chapter
intersections with the curve of order
[120
CHAPTER
VII.
Transcendental uniform
We
have shewn in Chapter II. ( 18, 19, 20), that all the fundamental functions are obtainable from the normal elementary integral of the
The actual expression of this integral for any given form of
third kind.
120.
fundamental equation,
is
and
forms,
for
is
numerical results
defined
it
may be more
algebraically.
Of such
integral functions (Chap. IV. 45, 46), the other expressed in the terms of
ix.).
"
for
Q*'
z,
"
Q^'
"
has place
therefore also the character of the integral of the second kind, E^'"
was found
which
here shewn
z,Ci,
how
It
... Cp).
it is
it
to obtain
vanishes to the
first
function.
*
of the third
may compare
Clebsoh
Noether, Math. Annul, xxxvii. (1890), pp. 442, 448; also Cayley, Avier. Journal,
v. (1882), p. 173.
122]
NOTATION.
121.
the
169
first
esimal at
X,
and
Dui,
let
integrals in 'regard to
denote any
Let
Riemann
denote the
infinit-
difi"erential coefficients of
all of
t,
a.
the
Let
c,, ...,Cj,
no linear
finite.
X^Dui
where
+ \pDup
the places
all
and
places*;
vanishes in
Then
all
&>{
let
of
thus
expressible
Du/
for its
kind
We
.
numerator
(?-,
i-th row,
( 18,
the
differs
from the
JDiil, ...,
...,
i(j,
i('^,
are
u,
(x)
as
v^'"
this chapter
Hi (x) dtx- In
Hi (ci) +
we
the form
notation
for
the
normal
Ff =
integi-al
Wj (x) dtx.
2irl J a
J a
be denoted by
Ci_j,
we
elements
m^'", ...,
for its
now
Thus
there exists
in the
z,
like
The
no
ith.
r=l
The
- ;\ U
infinitesimal at d.
infinite at
(t,
kindf
first
two determinants,
"
denominator only
u'
Consider
122.
C,-,,-
it''"
can express
Of (x)
jr-
as
given.
first
(Oi(x)
but such
1 of these
Ci, ..., Cp
(x) is
denominator has
function
is
and, at
Ci,
like Wi
{z)t~\
t,.
it
is also
being the
..., c,.
C,
is
u';'
on the
left side
p.
9 (1890),
Neumann,
259.
A FUNDAMENTAL FUNCTION
170
first
At the
kind.
[122
I'th
period
ni{z)- I
which, as remarked
121),
is
<Or{z) iliiCr).
Hence
also zero.
We
function of x.
^{x, a; z, Ci,
..., Cp).
Cj
49.
Cp
...,
it
shall
form
+\pnp(x)
\n,(x) +
vanishes in
all
of them.
where g
is
first
Hence (Chap.
III. 37)
B are
of the form Ag + B,
These constants can be
Cp is
constants.
jo
= l,
if
we use
z, is
1,
and
so that
a.
u,
v,
y^,
z, c,
we may
etc.,
rr=-^[!r(--)-f(-'')-^(-)].'iW=|^.
and obtain
^{x,a;
^{x,a;
z,
2, Ci)
Ci)=-^{f(M-j))-f(-y,)-f(a-n) + f (a-y,)},
g)'(a-t.)
+ g)'(a-yi )-|
and any doubly periodic function can be expre.ssed linearly by functions of this form,
same value occurs for yj and different values for v. (Cf. 49, Chap. IV.)
in which the
Since
123.
follows
that
= ("i-n'!'"
\dz
'
<=
(Oi{z),
wi{z)/-j~
jT,
is
-j-
is
I dtz'
Throughout
dt
is
an expression as /(z)
z,
= jtH^'",
,
-tt
is
when a
p+
WITH
^^*J
/
^ (*, a;
^,
...,
Cp)!
171
dz
a rational function of
is
POLES.
that o>i(z)
co,(z),
It
z.
is
a rational function
is
Hence
(x,
^jr
a;
z,
Cp)/
...,
At the
place
x,
and r-
r-"
is
",
z, yjr (o,,
n^;
J-
z,
log
namely
t,,
infinite only
..., Cp) is
infinite only
, is
is infinite like
when
like
at
is
when
or at
a.
At
plax^e a.
the place x,
a;
(x,
yjr
z,
c,, ...
of the
The
infinitesimal.
a;
z, Ci,
Cp)
...,
like
~ w +1
J^
'
^^^'^ Wi
at the places x,
place, like
-^,
1,
W2
by
dt
^
J
Wi
wind
only
it is infinite like
^^^^
z, is infinite
neighbourhood of the
and at the ^
place a
a respectively; and
will
be
infinite at
the value of
yjr(x,
a;
w+
z, Ci,
every branch
1
...,
the
Cp)
number
when
is
The
given below
( 130).
124.
From
the function
ylr(x,
/>(x,
Ml (X, z)
wherein
function of
is
z,
z, c,, ...,
place,
Cp)
..,..., <^)<.
,
an arbitrary
=e
we obtain a
function,
nr;;-lFrrr
r^i
x.
a;
=e
a;
^ = {w + X)V'
(of.
z.
When
2, 3; Chap.
I.).
z is
Ample
^=1
when w + l
sheets
wind
172
It is
(i)
[124
kind
it
when
aip (^)
Hi (cj,),
which
may
accrue
z, c,
do not
equal to zero.
to n^' "
liri,
The
(ii)
(iii)
The
(iv)
The
function
first
We may
and
is
z,
Cj,
like e^i
therefore speak of
'c,
c.
tc^
Ci, ...,
being the
Cp as essential
125.
we
give an application.
simple poles at
^'E{x,^:)
is
an uniform function of
at
Ci it is infinite like
Now,
in fact, log
x,
Eix.^mY
which becomes
F (x)
is
also
an uniform function of x
integral
V.'
ld\ogF(x),
fF'
\
(x)
dx,
~^'
'"^
C; is
"j r^;".
on the Riemann surface which does not enclose any of the places
certainly zero,
Cp]
for it is only
1,
Hence the
Ci, ...
a constant multiple of
equal to
2 V
r=l
Or B
'
f
I
'
dt
taken
t'
Ci
round
Ci,
and
is,
each of
Cp
^K"+
+^rr'
were chosen
125J
wherein
\i, ...,
\iflr(cO+
Hence we can
...
values of
= 0,
Xp be
r=l,2,
,p,
zero.
infer that
. .
r=l
a constant
173
this constant is
z.
functions of
'
E{x,a,)
E{x,oi^)
where a,, ..., 0^ are the poles and ySi, ..., /3m the zeros of the function.
have given the proof in the case in which the poles and zeros are of the
But
order.
We
first
Further, the zeros and poles of a rational function are such that
2 F,''
This
We
F, ^
2
is
remark that
Ex.
ii.
is
to be considered in
i.
= 1,2,...,^,
z)
by means of
and upper
Riemann
Ex.
r=l
r=l
j,
= 0,
x.a
n,'
these paths
/.v-za-c\
=log
(^
and
must not
(x-s){a-c)
T-w
h
{x, .)=^_-^^_-j
\
y,
we
have
^(.r,
a;
z, Cj, ...,
\o^~E{x,z)={\{x,a;
z,
^^izi-)
=log
^
c,)=
c^, ...,
Cp)dz
=-j
4^zr)+(._y)
tr{u-y) a-{a-v)
[f
dv {({u-v)-((,u-yi)-aa-v) + C{a-y{)}
(-y,)-f (a-yx)],
and therefore
(T(u-v)<r(a-y) (v-y)[((-yt)-i(a-yX
K,
^(^.-) = -(_y),(a-.)*
,
Ex.
Prove, if a',
iii.
(f
E{x,c'}E(a;z)
say
z,
Ex
Y
...
The
iv
...
,
y such
Cp ( 94, 96,
f,
Chap. VI.).
iK^, f
^. ci,
...
Cp)
/'
C'^.
f>
yi.
yp)>
MODIFICATION
174
[125
the zeros of either factor are the poles of the other, and the product is therefore a constant.
To find the value of this constant, let x approach to the place z. Then the product
becomes equal
to
- 1^-^
t:,
[7)^1^ {x, z
yi,
f,
...
yp)]j:_j
It is clear
D^^{x, a;
Cp)
2, Cj,
a.
^(^, f
Cp)>K') 2; fj-yii
2, ci> .
,yp)=-A'K2. ;
=
where a
is
f,
yu
vp)
--OfV'(f. ;
yp are subject to
z,
.-, Cp),
^. ci,
Hence
W [x
also ii
i, c^,
Cp)
. .
D^\^{x, a;
z,
Cp)-r^'"],
Cj, ...,
we have
'^(2;
^i-. V.
^{x,a;
Ci, ...,
2,
Cp)^{z, a;
a;
places
E (x,
of the
...,Cp),
yi, ., yp-
f,
that
z, o^,
f;
..., Cp)>/.(.j;,
Riemann
one place,
surface,
a;
(,
y^, ...,yp).
zero, of
c,
of the surface.
limit,
when the
of
fliipo) in
coefficient
denoted by
that
o'l,
infinite at only
The
;i,
z, c,
z)
126.
(x,
iv.,
yp) = -i;'K
f, yj, ...,
a rational function
yjr
yp)= W'Cf
..., Cp
z, Cj,
ri, !
f.
Of (c),
exists, infinite
by
or simply
numbers
c,
fl^.'"'.
k^,
...,
We
h^,
c,
at the place
c,
will
be
III. S 28)
...,
The periods
kp.
of a
wherein
approaches
c,
'
limit,
c,
fj.
all
-Xpii'^"-'',
vanish
when k
all
Gunther,
Crelle, cix. p.
(i
is
= i,2
p).
'
does
have care as
to the
199 (1892).
of the form
when
to the infinitesimal at
ki,
when
c is a
branch
place, it is necessary to
127]
when k
127.
is
\i, ...,
Consider
now the
function
175
all
176
128.
now R(x) be
If
the places
[128
first
...,
order at
\p so that
the difference
R{x)-\-<fr(x, a;
is
and
c,
z^,
...,
Zm',
a; z^,
c)
is infinite at c like
%-i <;* +
- {A,
Xm^{x,
a; Zi,c)
z^, c)-^.J^|r(a;,
+ A^
%-l
t-""),
where
Ai=\^a)i{z;}+
only infinity
1,
by
that given
is
this ex-
is
a constant; and
R{x)=A+\^-^{x,
a;
z^, c)
+ X^i/r (,
a;
Hence
p).
A+A,D]'-'^V1'"'+
wherein
(i=
+X,Wi(^'m),
...,
Ap
be
zero.
z.^,c),
whereby any rational function with poles of the first order is expressed by
means of the function i^{x, a; z, c). It is immediately seen that the
equations Ai
the
= Q=
...
number given by
When some
= Ap
the
R (x)
Xj, ...,X, to
III. | 37).
a;
B^'yjrix,
Ex.
a;
D^i/r(a;,
z, c),
If ai(x), ...,5p(^)
notation of
z, c),
(.:),
...,wp{x) in 121,
and the
J=l
and that
ytr{x,a;
2, c)
= >Ka,-,
^{^,a;
From
129.
z,
^, c^, ...,
Ci, ...,Cp)
= -f
k.-l
c)-2a,i(z) D^'
yjfix.a; c,Ci,...,Cp)
E{x,z) =
a;
(;jr,
c)
z,
z,
=e
I
c)
we derive a
''
r=]
-
&),.(^) dt^,
Cj, c).
function of
a;,
given by
'
properties
(i)
It is
an uniform function of x
'
r=l
130]
The
(ii)
the place z
(iii)
and
is
The function
is infinite
.1
before
zeros at
j3i, ...
/3,
\K-i
-
v'-'
c,
and there
like
t:''r
e'='
tc
As
177
o^, and
...,
when some
of the places
...,
are coincident.
...,
'
0
""
130.
We pass now to the algebraical expression of the functions which
have been described here*. We have already (Chap. IV.
49) given the
expression of the function ^{x, a; z, d, ..., Cp) in the case when all the
places a,
still
Chap. IV.,
If,
86,
Chap. VL).
y when
a;
= 2, we
-i
4>o {s,
z)+ 2
/>^
z) gr {y, x)
(s,
{z-x)f(s)
and use the integrands o)i(a;), ..., o)p{x) defined in
^{x, a; z, c^,..., Cp), which was given in
put into the form
byt
{z, x),
rational expression oi
^{x,a;
= (z,x)- {z,
z,Ci, ...,Cp)
a)
- 2
m, {z)
[(cf,
x)
121,
49, can
the
be
(a, a)].
r=l
is
known
z,
(Klein, Math. Annal. xxxvi. p. 9 (1890); Gunther, Crelle, cix. p. 199 (1892)) that
of the expression
p. 386;
\j/(x,
a;
x.,
lectures.
is
aware.
Bolza,
GStting. Nachr.
1894,
p.
268;
235 (1895), and in the Jahresbericht iter Deuts. Math.Vereinigung, Bd. iii. (Nov. 1894), pp. 403 436. But it does not appear how far the last of these
is to be regarded as authoritative
and it has not been used here. The reader is recommended
to consult the later volumes of Weierstrass's works.
Weierstrass, Geiamm. Werke, Bd.
ii.
p.
( 45).
It will
be adhered
to.
12
178
(^^
x)-^, wherein
dz
by
limit
this
is
x)
{z,
-j-
[130
finite*
z, is
we denote
if
for
the places
at
ylr{x,a;
= [{z,
x)
- {z,
dz
^-
)]
tOr{z)
Cif
Ci
From
.^
pf.
/
{X, z),
>
= [(z,
the manner of
x)
{[(c. .)
_/:::- s
""^"'^
I><*''"^
z, c)
wherein D*'"^
we immediately obtain
^{x,a;
of the
all
x)
dc
- (a,
a)]
-^
finite positions of
Cp.
this expression
131.
[(Cf,
{z,
- (c,
a)]
a)]
=e
it
^-
||
45)
rr[(o..^)-(c.a)]g
"'
r=i
(z)
a,,
= limit^.^
B^^ {[(c,
[(|-)'^"
m)
- (c,
|[(r, x)
a)]
- (?,
^|
a)]
g|]
49,
fij (),
. .
. ,
integrands
(^1
dz
(z)
f^)di'
Z(f)i
(z)
JU^
dz
di'
z^'~'
'
^ (z)
/'(s)
dz
dt'
<f>i
(z)
/>)
dz
'
di'
place of D,i{z),
...,
!^p{z), the
for
precisely as in
In regard to
...,
fip(c),
^r^j[(c,^)-(c,a)]||,=i)^p:-.
regarded as a function of
x,
when x
is
near to
c,
as
Hence,
* /'
87).
('?).
also, it follows
"when
Or the
Ti
is
result
very nearly
may
g, vanishes to order
be seen from the formula
(z,x)-(z,a)=~
(Chap. IV.
5 45).
+ w, and
P^'l
132]
EXAMPLES.
Ex.
132.
i.
/=(^,
if
place
be
Cj
(Cj
x,
at the place
is
1)4,
a be
respectively o,
h,
respectively, then
Cj, c?j
when
(a)
179
d^) is not
a branch place
<o,
(z)
= ^^ u
^\^ ^+y
"-^^
dt'
'is(z~x)
'
and
(.r,
a;
fh_;^,h _
\_2s{z-x),
2s
0) when
"+^
c,)=\ ^
z,
(Ci, c?,) is
dzFs+y _ s + b _ di+y
dt\_z-x z-a
Cj - ^
a branch
place, in the
x = c, + fi,y=At + ...,^,iz)=^^
dzV
d,
2s{z-a)_\dt
d^
Cj
+ bl
- aJ
d,+b
Zd^ic^-a)}
'
neighbourhood of which
icx)^ = ]irait
^s at
d,+y
dtl2dj{Ci-.v)
at
^^^^.2t=^j^^^,
2At{Cj^-x)
A(Ci-x)
o{
and
^{x,a;
e,)4 ^+y, -
z,
\2s{z-x)
_
If
dz (s+y
2s
dt\z-x
(*, z)
Ex.
ii.
a;
shall
When
Cj, Cj
2s'
y
c^-x
^, = Umit
c^-aj
^^ (*> ^)
'^
a,
of -i-
A{c^-a)]
'
branch place,
in the neighbour-
=i.
2^,
il/{x,a;
z, c)
when
c is
y=Aiti+..., and
,.
^^
h_\
(
y
2sdt\A(c^-x)
For the case {p = 2) where the surface is associated with the equation y^=f{x),
is an integral function of x of the sixth order, we shall form the function
Cj, Cj) for the case where c-^, c^ are branch places, so that
f{ci)=f[c,^=0, and
i/r(x,
z,
s+b
z-a
2s^
where f{x)
-i^l dz_Adz
2s{z-a)\dt
Co
a;
A-,dz
^-<h -^idz
r,
we have
.422=/'(c2),
.,dc,
'"'(^)=rf2JS'
and
V(^.. ^.Ci,C2;-|_2,(2_^j
2,(2_ajrfi
2ia!4ci-C2Ui-a;
c.-aj
+ izip_
C2-Ci\C2-^
When
so that
c is
x=c+t\ y=At+Bfi+...,
we may, by
V.
58,
definition of the
forms
3rd,
...,
(p-(-l)th
<i(z), aiiiz) (
127)
Ml
c^-aj)
Ex.
ii.).
In the
by the
any linearly independent integrals of the first kind. In the case
now under consideration we may replace them by the differential coefficients of the
differential coefficients of
integrals
r-
122
HYPEKELLIPTIC CASE,
180
= limits =c
X dx
limits =t
2^ di
2y dt
"
^ \^y di)
\2y dt)'
A^
(-f)
z
cfo\
and
WX2= "'""**=
&
Hence
Further
dz
f2^
sdU
':\^dij'
,(.)
'
2s dt
it
'
\dz
limit;,^^
_2 /
"a
ZAt
-j^
2At
"
At + Bt^ +...)' {dt) \At+Bt3+.
A
_2S
Hence uj (2)
[132
= [4 + S(.-c)]lJ,
_2
/^
/'i
2s dt
dx\
"[iy dt)
.)
THE FUNCTION
132]
In
fact, if
ti
IS
FUNDAMENTAL.
181
A. [(c
'] M
cO
c^,
J + 1]}^^
place
Ci,
Now
with
its
the elementary integral of the second kind obtained in Chap. IV. ( 45, 47)
pole at a place
c,
when
z is
is
E^''=
dzD^l^z,
c),
whether
be
z is
<- = />Z>.[(.,o)J].
Hence the
is
coefficient of
t^
when x
is
near to
Cj,
r=l
This
is
equal to
z.
We shall see
iv.
its
pole at
Ec
-^e,
( 135,
its
Cj,
pole
Equation
x.).
tion of X,
when x
place, the
is
'^
<i(2)
tc"'
first
kind
If
tc
+ -"+-4t^<j>(2).
</'
c.
tc
.?i{^K^^'^)S)-J:-<^^^"f
where
/"i, i is
the limit,
when the
place
del
c of the expression
l^i-M
^.{*-[<."S]+}'
Among
which H
I:
is
is
the coefficients of these positive powers of t only those are important for
/fcj,
/&!,..., Jkp.
...,
ip, is expressible
I^'
",
when
by those of
2)*.-ilt-,...,2)J--'C%
of which the indices k^-l, k^-l,
of
(Chap. III.
28).
...,
rational function
182
Ex.
V.
Example
z, c)
[132
whose expression
is
given in
ii.,
A+B{z c)
2s
z-c .dz\
It' tf'
Jt't.'^s
dz 1
dz
V A
-n
~\
dz
,dz^
dz
dz ^,r
.r
C\
r\,
r\
If in Ex. v.
vi.
i''5'=X/'i''-|-;i/'/
which
is
known
t,
fi
and
t^
be denoted by
+ integrals
26
+ rational
Chap. V.
function of
57,
Ex.
ii.),
{s, z)
X and
fi
being
constants.
Prove, in fact,
if
We
133.
for
of which
>|f (a;,
a;
,1
z, Cj,
..., Cp),
i^
(a;,
a;
z,
in
c),
one
explicit
algebraical form.
The
cases of the
to give the
From
ylr(x,a:
we
infer,
z,
..., Cj,)
[(_z,
122, 130)
a:}-{z,
a)]^- i
wi{z)[{Ci,
x)-
(a, a)]^^^
rr-[(^,^)-(^.)]^
(i)
by HI' \ that
p
^r=.s-K-)^r
1=1
(")
OF THE FUNCTION
133]
The
function
is
fl'^'"
same periods
that
fl"^' "
J- is a rational function of
r^' "
T^ "
X)
\_{z,
- (z,
J+
1^
in regard to
(iii)
an arbitrary
c is
z,
place,. and
J^'^
is
algebraically
The equation
".
(ii)
shews
0,,
(z)
H::
"
(iii)
z.
we obtain
z,
nfz,c; = P^'x.a,
l+i V:
function of
but
z,
z,
a)]
as T*'
t- as a rational function of
where
183
i/r.
H'"
"
(iv),
\
/>
C;
is
( 45, 46).
Since the integral of the second kind ^'", obtained in Chap. IV.
45, 46), is
changing X and
equal to D^P^'^,
z,
and
a and
also
c,
and then
differentiating in regard to
z,
+ 2
E^
Ff
BjLc.
"
\(z,
X)
(iii)
The
which
diflference of
everywhere
is
r^
I + 1^
[o,,
{z)
/It
t- as a rational function of
finite,
the equation
(iv)
An
p. 120.
From
first
z.
But
it
should be
They
we have
second kind.
(vi),
z.
equation (v) expresses the difference of E^'" and r^'" in this way.
From
(v),
to express F^'",
{z, a)]
=An;t.a, =^jn,,,,
Vf
this equation
is
=1
on the right
'
may
184
first
Ci,j
and
[133
we
C'i,j respectively,
may
" are
loops of the
first
But
134.
"^
immediately obtainable.
These equations
kind.
all
they connect functions which are algebraically derivable with functions whose
definition
loops.
We
proceed further
certain
them by
algebraically.
where
is
an arbitrary
place.
z.
= D^Hl-\
And
if,
as in 132, Ex.
iii.,
we use
the
by
we have
Mi,
and
E']."
(v),
putting therein
dxD^{x,
c,)
-Dx
c, for z,
'a
d^hT^d..
we
we obtain
If in this equation
in equation (v),
(a;.
Cr)
(Cr, a;)
+ 2
Ai,r Wi {x\
r;-=^r-+|7r-{i).,[(.,c.)|]-A[fe,.)J']+j_^,,,w};
135]
and thus,
if
we
185
by the equation
i{Mr,i- Mi r)tOr{^)]
-I r=l
(viii),
we have
rr = ^' " +
2
i=\
or
r=l
f=l
i^
"
(ix)
by equation
\eX'- [{cu
+i 1 1
The equation
we obtain an equation
i=l...p
-'
Ptc +
z,
z)-(cu
(ix)' is
also equal to
I']
(rrvr- rr Fr)w,.-ifr,.)
(T- "
c)]
(viii),
Q^' " is
(ixy.
such that
&''
+^ I I
(ix),
[Wi
when none
(a;)
0),.
{Z)
of the places x,
z, c^, ...,
i]
Cp are
(x),
1=1 r=l
t'^-'^-t
^"^>
+I i
{"' ^^^
Ik
[I ^"' ''^ - rz
(a;)
Wr {Z)
Q*^ "
= g^ ^,L we
infer
\hi (^''^)-a^^^^'^)]}
{Mi, r-Mr,i]
(xi),
i=l r=l
CANONICAL INTEGRAL
186
This
fications.
[135
is
funda-
The
The equation
(si)
Equation
It should
By
be noticed, in equation
(xi),
^.
For an expression in
..., Cp.
( 137,
Riemann normal
twice.
11^'
17.
Ex.
integrals
ff^'
'',
^'
vi.).
The method
and the particulai-
".
some of
his
136.
From
the equation
r^-''
(viii)'
we have
+i I
2(A,i + A)^r
Gj"'
t=i
From
this equation,
we
(viii)',
= ^}r{x,
which result
may be
infer that
a;
z,
(xii),
(?r=a(^)r;"' + [(.,.)-(.,a)]|-|^a(.)[(c^)-(ca)]^,
the equation (xii) has another importance
canonical
integral
of the
if
G^'
we
",
Q^'" an elementary
call
=D^Q':^, an elementary
thus T/te
any place
Q),
+ (rational
function of ,
z, c,
. .
Cp)
/^
E (x, z),
in the form
h (x,
z)
=e
-1
(xin).
1^'^]
If
we put
and
187
gives
(xii)
therefore, also
(xvi),
and
^'
^)
{^''
which
is
I) - ^' ((*'
")
'- "i
(-)
^. ^et"]
(Xvii)
(xi).
'
Ex.
Prove that the most general elementary integral of the third kind, with
z and c, and vanishing at the place a, which is unaltered
when
are interchanged and also a and c, is of the form
137.
Its mfijiities
X, z
'
t=i
'
i.
at the places
wherein m,
Ex.
ii.
= l r=l
i.
be denoted by
be denoted by G^'
"
,
prove that
^(x,a;
Ex.
iii.
If,
in particular,
"
(f.'
= ^^"-
2 <.(2)^'".
'
=l
be given by
1=1 r=l
prove that
'
This
^{x,a;
Ex.
is
r=l
2, Cj,
..., Cp),
z,
of the coefficient of
<;
in
the expansion of
Cj (
132,
iii.).
The
^'",
of this e.xample, in
that its calculation does not require the determination of the limits denoted by
Ex.
iv.
1)4,
is
of the form
i/i,
j.
188
if
Cj
be denoted by a;=c,,
_y
= rfi,
^)
+ i/"^' (2)
[137
(X, Z)
dz^'^'''
and calculate
"
5'2=/(j;)
if
dx
^) =
/7^'
where,
- 1; {Z,
"
P y+/(j
f''
[/'"> (2)
xi,
(^ - Zf],
in the form
^) cdr rfz
;,
{^ -
_ J_
f^
dx
f'
dz
metrical expression
a;2(cB,02
Ex.
where
= limit, =0
/(.r) is
when the
place
(-J
V.
1= -/' (ci)/4/(cj).
J/,,
(/>
[--), wherein
= 2)
in
Cj is
a;
c^, c^
is
t^,
and
in
J;=C2-t-^2^
^^ -je22-ci= "^'^''-^'c^-c.^sdi'
and
^'''''
dt-Ai^':^z' ^''^-A,{c,~c,)
infer that
[,o,{x).{z)-^,{x).,{z)][M,-M,]^-^^ (x-z)
Supposing X and
^r 4.
where
.dj'',
^V! f[/'fa)+/'(^)}(^-''l)-2/(^)
^l
A^ have
(Cj),
that
z2 _ [/fa)+/'(z)][^-C2]-2/(2) ^-Cil
C,-C2
(^-Cl)^
(ix)
l^^^'^.
(c^)
(0-C2)2
Ci-C2J'
respectively.
where,
if
/(^) be
/(jr, z)
denotes the
expression
j;'(a( i'+Scti 22
and i,
Ex.
J/i iV^
vi.
Let
R {x)
c,
and
c^.
137]
sum
to the
infinite,
we
shall
(cf.
all
t,
R {x)
will contain
and Lfi
-J-
(;f)
-^
_j
-r-
means the
in terms of
coefficient of
-j-
form U{x)
is
'
theorem
^[^wJl- =
infinite,
189
ifr.
R {x).
If however
when x
is
{x) be of the
V{x), wherein U{x), V{x) are rational functions of x, whose poles are at
finite places of
the surface, there will be no terms arising from the infinite places of the
surface.
= 'KI.
2. i.
-,
x)= I
[a,i
a;
denote fixed
Cp)^^/'(|, a; x,
finite places
prove,
by
that
D,^{x,
z)-I).-^(?,
where
tc
Shew when
\
z,
E^/'
{z,
x)^
},
'c,
and
[ylr(x,
as equation
^jrix, z) is
coefficient of
{X)
=1
regarded as a function of
x, in
the neighbour-
Cj.
all
(xii)
z),
obtained in the
is
the same
text.
holds.
Ex.
131),
(Cf.
iii.
p. 180.)
vii.
we
^{,x,a;
Ex.
By comparison
standing by i)*"'
"
"
a)]
B,r = (I^I^-'
HD'^.
*
and, when z
^Ur-[B': b\
Gunther, CreUe,
is
-((.,
cix. p. 206.
.)
J + y]^^,
MODIFICATION
190
wherein
t^
133136
<"=
XI-
Ff "i)**"'<-
C?^
^*.-l^.,..^^..-l|^.a_^^^_
when
c is
(ii),
=1
wherein,
so that
l<i(^)i>r'<-''
1
Pi'
c,
[137
TTX.a
\-^c,
^x,
a'
^^-^
T^
Fr
first
"c
^X^a
T^*.-l
)]g + ^|^5,
.)-(c,
(iv),
(Vii),
to be interpreted as
is
'
-^2
2 (iri-iV,)F;'"<.,(z),
i=l T=l
is
equal to
rr"-i2
Ff V(2)(fir,i + -Bi,r)
(viii);
1=1 r=l
and,
finally,
..m = n,.-i2
F,'
F,.
t=l *=!
which, clearly,
is
such that
(i + 5,)
defined
^ {x,
z, c)
(ix),
by the equation
v:- ")
2r ,
(ix)'.
a; z,c)^(f;'- So.,
[z)
d]''^
"
CT,'
(xii).
137]
The
191
is to
i>.^.c:= A ((-.
.)
A ((0,4:)}
c)
z,
as a function of
in the
t''/\k
<0i (z)]
Ki, r
(x).
c,
so that (Ex.
iv.
132)
where Pj,
the
sum
is
Ni,
of the coefficients of the first negative powers of the infinitesimals in the expan-
yfri^,
a;
(2,
z,
c)
D^ V'
(I,
x,
c),
at all
places
where negative
^ (^, a
-Ox
2, c)
V
V'
- uj (z)
>
is
L^.
(A),
(a;)]
one of the
p numbers
2 I ^i{x),^r{^){Nr,i + Ni,r)=B,^i,[z,a;
A^Cm-i i=lr=\
x,c)+ l,^i{x)L^M
(B)
t=l
thus an elementary integral of the third kind, permitting interchange of argument and
parameter, is obtained immediately from the function i/r (^, a z, e) by integrating the
right hand of equation (B) in regard to x and z.
;
Prove
that
also,
if
a)]J,
K^r-
2 o,dz)jf'-'Ey
(XV)
1=1
^'(('' ^)
Ex.
viii.
equation
To
" - i (^)
(xvii).
is
wherein /(t)
is
is
(xi)
of Ex.
vii.
When
x, z are
c,
putting
prove that
D,
((x, z)
t^
and
t^.
GENERAL STATEMENT
192
(see Ex.
and thence
ii.
[137
132) that
R(x~z) dx
when
Also,
3-
function
M+
+i\^<2
z is
Ci)-^(Ci,
(2,
77*
^)
be a place near to
if Cj
c,
dz
^^^
c,
+ .__^sothat
prove that
,r
^~
UH/'(2)1[2-c]-2/(^)
2^(2-cf
3 ^''(2-c)[^H^$(2-c)]+/'(2).(2-c)M^-ie(^-c)]-2/(2)[2^'-ig(2-c)]
~
'iA^s{z-cf
where f{x,
{
is
z)
(xi)
of Ex.
vii.,
(2),
where
oj,
v.
the brackets
vii.,
^jf^x,
a;
z, c)
132.
ii.
The
138.
canonical
prove that
,=<,, <.
found in Ex.
'
formulBB in 133
136 enable us to express the form of a
third kind, in the most general case
and to
integral of the
Let
kind
fij
(Cf)
M^' ",
. .
w*' "
denote Dxuf
denote any
"
be denoted by
by
^,
fj,i
(x).
Ci
Cp
(i,
j)th.
element
is
121) to define
Then we
toi (a;),
clearly
...,u)p(x).
have
Wi{x)=Vi,ifij{x)+
* Since
u",
...,
u'
i/i,p/ip(a;)
matrix
/i"'
(i=l,
cj, ...,tp
2, ...,p).
are independent
138]
Ui^j
a,-,j,
193
in
which
,-.
i^'"
and
call
v,,iir';
v,,iHi
them fundamental
integrals
u^"^,
integi'als
...
...
,.^,,/r;"- 2
k, <" +
+ a,,^u';\
...
Thus by taking
i/,-,i=l.
fjbi{x)
'
,u''".
...
"
F^'"
p
for
the row of
...
quantities i^'",
^^^'"t find v
-t^'",
H'''" stands
where
...
in
is
fact equal
to
the
137)
r=l
'
;=l
=1
= 2
, j;, a
a)j{z)nc.
-2^ S.S .2
'
,=1
r=l
.?
134, 136).
if ", we have
r=\ s=
1=1
'
x,a
a,,jMr
fisiz),
=1
r=l =1
and
this is
places, the
functions
an important
if ",
functions
.
result.
if"
independent
Thus the
c^.
i^' " do not depend upon the places c,,c ...,
voL
See for instance Cayley, ColUcted Works,
volume, where other references are given,
'
ii.
p. 475,
II. to
the present
13
B.
194
we
dz
B.
infer
"(^,.)|]=A^r-^.^;'
D.
{2,X)
dt
Now
Rp (x) be any
[138
rational functions of
W D,Lf
it is
"]
(17),
if i?i(a;)
x such that
I [fii(x)D,Lr-l^i{^)D.L7'']= i[/,i(x)Ri{z)-^H{z)Biia>)l
1=1
i=l
then
-Rj (x)
Hence
a.
the equation (17) furnishes a method of calculating the integrals Li'", when-
ever
it is
possible to put the left-hand side into the form of the right-hand
side.
The equation
dx\
D.{{x,zf^+l^,,i{x)D,Lr',
is
x and
z.
This
expression
is
also
equal to
dx\
p
l.ar,.IJ.r{x)^.{z)
t
A((^,^)^)+i>zHx'--2
dt)
r=l =1
and, therefore, to
y
DiT^
/.
-2 2 2
a,,,^(a;)/i,().
r=l s = \
tix.cr^c + i
134,
Mj
Li
ix.)
=llj_c
1=1
"
-2 S 2
fflr,
Mr
r=l =1
2 I
{A,,
r=l =1
+ A,, ,)
"
-2 1 I
,=1 ,=1
o, .^' %:
'
functions Li
are found.
The formula
,dz
r=l=l
serves to express any integral of the second kind in terms of the integrals
.
EXAMPLK OF HYPERELLIPTIC
139]
Ex.
i.
195
CASE.
/(a:) is a rational
polynomial of order
2/i
+ 2,
the function
Rin^
^^^
d ( V \
V
(|-2)'rf\y(|-^)y'
iJ(|)
[
/'(g)
2/(1)
2y4(|-:r)(|-2)
(|
is
=0,
- a:)2
(^
'
ij).
g^(.,X)-g-(^,.)=22(^--fc)X,,-,,(^2^2i-2^2-j'
where
^'
represent in turn every pair of unequal numbers from 0, 1, 2, ..., 2p, whose
not greater than 2p, i/ being greater than k, and the coefficients X are given by the
fact that
sum
^,
is
Hence, a set of integrals of the second kind associated with the integrals of the
dx
r^
.
JaJ' }a~Y'
given ^
by
.
IS
^7=
J
and a canonical
is
is
(3^)^
[^
x'-^dx
}a
first
kind
'
X,,,a't + 1-V).r',
(i=l,2,...,p);
given by
+,!i^
^)^....^^j
(*+'
t!l
equal to
.J
J 2y+
' az ax
/:/
which
'
2
k=i
a^y
Ja}c2s2yl
This
xdx
/==
is clearly
The value
23
(z, a^)
gi.
xie*[2->^ + \^^iix+z)]
{x-zf
2y
symmetric in x and
of 5-
2
t=o
{x, z)
z.
is
p. 49.
/^dx
'dx
\^x^2.\^x^
I:
^^
functions
'^y
is
y^=ix^-ff^-g3.
X,
139.
i;*'", ...
7n>r()
which, since
fl,-
{x)
Then the
Wp'"
= V,
'i-rriDv^ ",
M*' "
Hh
elliptic
Mj'
...
Up
^i(a;)+
are equivalent to
= 2 (V
''f'
"
+ Vp^p")*,
at the
first
integral
u;
"
at the second
132
IQQ
that
so
Xt,
Q)ij = \j.
be denoted by
27?;
...
...
be denoted by
will
",
namely
We
D,j (Ci).
be denoted by
= 4 (ai, lOJi.j +
+
j = 4
277'i,
and
ft)/,
. . .
+ a;,
{i,
(o'j,,
7]
2ato,
ri'
2ao}'
{Ci) is
/tt,
the
(i,
we denote
infer that
fij (Ci)
The
prove
functions.
,
Cjj,
to
and
fl, {x),
hence we
l,2, ...,p)
i/p, i
flj (Cp)),
we
2h\'V'' "
...
obtain
is
may put
AX.
2'7rifJi
Up ",...,
ttiA
v^'
we have
h,
= A-rri,
=
...
iri.
Thus
Vp'"'
are
connected
by the
= ttiV- ".
four equations
2Ao)
= iri,
2ha>'
= irir,
rj
= 2aa>,
t)
= 2ao)' h
(A)
L'
therefore have
(i/i, i
i^~"A by
the matrix
...
the
by A.
2AX^ =
Ci,
{x)
iXr
If then
will
is
{i,j
j)
Trifi
and
Lf " we
J,
wherein
the
(ttj,
of the
all
Now
...+HpOip,j),
27;i,j
similarly the
77;
H^;"
By
will
a.,-,^
The matrix
17'.
H'^:
w = \ and
vup
will
v',;j
shall write
i-' " at
j and
we
The two
[139
/i,
t,
which
in
general
unsymmetrieal, to be
141]
equation hu"'
"
= ttot'*-
the
first
aiij
",
...
197
^hm
equation,
= iri,
by the
similarly the
other
mined
Vt,j-
T7i,j,
matrices
a,
given quantities
is
thus only
V-
^p {p +
1) quantities,
p(p + l) +p^.
l>
given quantities.
+2' (P
1)].
There are
therefore, presumably,
= ^F' - P'
we can
in fact
One
of these forms
wi)
is
= T)u>,
^'r{
= rjm,
first
ijai
wr) =
equation
is
^iri
= to
t]
(B)
rj' ay,
^p{pl)
equivalent to the
equations
{a>r, i Vr, j
Vr.if^
r,
j)
= 0,
r=l
in which
equation
is
equations.
right
The second
1, 2, ... p, j=l,2, ... ,p, and i is not equal to j.
similarly equivalent to ^p (p - 1) equations, and the third to pThe total number of relations thus obtained is therefore the
number p^+p(pl),
In
this
Weierstrass's equations.
VV'
is
(C)
Riemann's
These equations are distinguished from the equations (A) as
equations.
deduced
The equations (B) and (C) are entirely equivalent either .set can be
141.
way of obtaining the set (B) is
from the equations (A) or from the other set. A natural
;
natural
way
integration.
recalling that a
= 2aaa
= a,
w'
matrix,
,=ft say, a symmetrical
5n'=25aa)' MA = 25a(i>T-
ija=ijaT=^T=^T,
Hence
and because
5'
= t,
m'lj'
Tfflij'
= o)t, t = t,
t|3t
^ irir,
A(b
= /3t ^jri.
make use
of the
a.'
And
(B).
i5,=|3, S,'
it
= ^r-i,
i;
= (S)-(3,
<o-'a,'=x, so
if
= y(o)')"'i
that <ok =
v'
o.',
^nd thence
iS'
and therefore
(i)m'
is
the matrix k
and thus
<u'ai,
r)a>
ioj]
and therefore
\'ni
-^ jn',
<ir}T]'u>=^K^
leading to
is
K^K-y = inV,
(C).
Further
which
= i'"'
K,
(i)-ia)'=m'(oi)"',
is
s' {B)"'3-r(S')~'<
S, 5'(ii)-i
K/3-yK-'=ijri, or
which
we obtain*
matrices,
Hence,
= o.T,
that
the equations (B), putting i5,=|3, i5','=v, so
From
and
[141
198
is
\iri^,
thus also
Finally (u'v
and thus
a'ri
which
is
We
-<o^'=lin,
therefore, rjU'-ri'm,
(C).
have deduced both the equations (B) and (C) from the equations (A).
to deduce the equations (B) from the equations (C).
similar
Other methods of obtaining the equations (B) and (C) are explained in the Examples
which follow ( 142, Exx. ii v).
142.
Ex.i.
integrals given
r/^i".
='lJi-e,
'\
by the equation
If
fj''i<^
+---+h!i-"cp
>
( 139),
divided by the
are a set of fundamental integrals of the second kind associated with the set of integrals
of the first kind inivf
", ...
iirivp' ",
z>.((..4;)-^^((^.4:)
(a,;
The determinant
independent.
mine an
(of.
J^
(X)
n.
of the matrix w,
The determinant
D. K\ - 0,, (z) D,
\i^
.., ^'
no periods
231, p. 440).
K^ ")
are linearly
we conld
of the
deter-
second kind
142]
Prove that the fimctioD
haa only one period, namely at the tth period loop of the
A?" "
we have
in fact
Shew that
Ex.
equal to
is
...,
For the
1.
2t)p'
A,'
Ex.
sets
a;
...,
V=-i-
From
ii.
199
and
(C).
we deduce
i.
a
2,ri 2
^i
"
"
, '
A,
(j
z,c ,x,a.
-,
A;
)= HS
I,
.!,
L,
(,
z,
c .x, a,
u.
L^
).
1=1
t=l
z separately to pass, on the dissected Riemann surface, respecfrom one side to the other* of the rth period loop of the first kind, and from one
tively
side to the other of the sth period loop of the first kind,
we
p
2
-4
((aj,ri;j,,-?7i,,<Bi,,),
t=l
which
the
is
(r,
- 4 (ui; - ^w).
5?;
- r)<a vanishes
(Ex.
i.).
right hand.
Supposing X to pass from one side to the other of the rth period loops of the first kind,
r from one side to the other of the sth period loop of the second kind, we similarly
prove that Sij' 7<b' has the same value for the functions on the two sides of the equation,
and therefore, as we see by considering the functions on the left hand, has the value Jm'.
and
While,
we
both:p and
if
pass from one side to the other of period loops of the second kind
5'ij'
= ^'a>'.
We thus
Ex.
If
iii.
of the first
t/f"
"
. .
,
.
V^ "
be respectively denoted by f, f
and IT^'"
roimd
|',
fi,,, fi,,-,
,
...
?7^'
" at
U^'
d W^'
",
taken positively
all
p
r=l
which
Ex.
is
iv.
If
the function
".
,
Lf+Riix)
is
infinite
To
Vr,i [(Cr, X)
- (Cr,
a)] -rj
that side for which the periods count positively (see the diagram, 18).
200
Prove that
if
in the
"
[142
c,
then
where Ai^
are as defined in
,, i/j, ,
Ex.
Cj
t-p
. .
V!''
i.
all
is zero.
V.
134,
and
iii.
(a)
U'''^
(0)
f/-."=J--",
(y)
r"'
"
by taking
iv.,
"
rf'
that
a>5'
= a'a,
= wf
vi.
we can
Shew
Ex.
on
the
fij(6r)
We
143.
yjr (o),
...
Z,'""
z, c),
(c), ...
...
D'^
tt^'
fipic),
"
(6j)=l,
/i,
^j
c.
independent places
that
Z/'j'
vii.
^,
c,
h.^,
6;,
= 0,
...
b^, just as
prove that
v,.,i
(a-),
..-
= o>r{bi)
Mp(-^'). '^e
ti)i(:c), ...
from
Cj, ...
Cp,
so
and that
E (,
z) to
(undissected)
Riemann
rational
surface.
include, but
Such functions
that they
functions, in
may
are
possess essential
singularities.
Consider
first
is infinite
i/r
{x,
F{z)
dz
\
z,
i/r
c)
{x,
;^
a;
is
a rational function of
z, c), -TT
dz, or
z,
F{z)ifr
the integral
{x,
a;
z,
Acta Hath.
i.
sum
of its
c) dt^,
its value.
But
vi,
143]
where F{z)
infinite,
is
201
ip-
(,
a;
z,
c) is
infinite.
Now, when
is
^r
is in
a rational function of
we can put
z,
ylr{x,a;
where tm
z,c)= 2
m,
Dl^y\t{x, a;
c),
is
Thus the
integral
F(z)yfr
the value
ui, is
{x,
27n
where
J!t
a;
-j-
z, c) dt^,
i)^Vr(a;, a; m,
c),
of the integral
place m.
When
like tx
tx
is
in the
x,
x,
ylr
and
(x,
a;
z, c) is
infinite
therefore, taken
round
JF{x)ylr{x, a;
z,
c)dt^
gives
2-7riF(x).
a,
2TriF(a).
in the form
= F{a) - i Y^Dr^ix.a;
r=0
vi, c),
"
LL
the places a and c being arbitrary (but not in the neighbourhood of the
place m).
p = Q,
^{x, a;
mY
F{z)
s,
c)
=-
and
wherein
^ =
2lTt
A similar
finite
dz,
result can be obfciined for the case of a single valued function with only a
number
say of order
/ (z
/i,
of essential singularities.
the integral
jt[^
F{z)
dz,
When
is
only a pole,
^ (x,
m,
(?)
will terminate.
202
[144
We
144.
to
which fi{x)
functions, of
is infinite
i?" (a;),
we can
find
c),
way
otherwise infinite only at the places Ci, Cj, ..., and in such a
place
of
the
neighbourhood
a.
that the difference F{x) -fi {x) is finite in the
and
Since fi{x) is a rational function, infinite only at the place d,
put
we
can
comes
to
c,
z
when
infinite
become
\|r (a;, a
e, c) does not
which
is
wherein
places
when
and
sufficiently near to
is
^|r{x,a, z,c)=
wherein
tc is
T^ Dl >}r(x,a;ci,
J^
(A)
c),
is
Also,
c,
and x
s [Dl i/f
"S,
Dl ir{x,a;
wherein Rkix)
z, c)
Thus
c.
we can put
c,
z, c)]z=c,
also,
when
is
near to
S^ <* R {x),
is
not near to
is
(a;,
and Ar a constant.
c,
(B),
is
c.
There are p values of k which do not enter on the right hand for it can
easily be seen that if A:,, ..., kp denote the orders of non-existent rational
functions infinite only at the place c, each of the functions
;
[B";
-'
yjrix.a;
vanishes identically.
must
z,
c)],=
Of the
places
be the places
let these
"'
ir{x,a;
z,
may be
lie
[Z)f
an
infinite
number
then
will
c,
within which z
be denoted by M.
valid,
s will
c],=,
finite and,
M;
when i>s,
we have
If
-v/r
{x,
a; Ci,c)=
*
t^
Ri,ii(x),
k=
wherein U is the value of tc, in the equation (B), when z is at c,. Hence also,
from the equation (A), wherein there are only a finite number of terms on
the right hand, we can put
fi(x)-fUa)=l
t1Si,tix),
(C),
ft=0
wherein
Sj^t is
*
a rational function,
so that c
is
what we may
>
s,
and x
is
...
c.
(HaufuDgsstelle).
144]
It is
function
a;
yfr(x,
The
c.
a certain neighbourhood of
Let now
is
The
the investigation.
c) in
z,
be any
We may
c.
203
a;
outside
is
call this
such
convergent
let
+ E+2 + f+3 +
/tj
for i
> s,
the
terms
x outside M'
> s, by
we
are, in
modulus,
less
than
ei,
when
the equation
have, for i
Thus the
> s.
series
2 [/i(a>)-Ma)]+ i
gc{x)
i=+l
1=1
The function
which
is
is
not unique
if
the place
infinite only at
any
(x) denote
yjr
the addition of
c,
^ (x)
to the function
obtained will result in a function also having the general character required
As here determined the function
in the enunciation of the theorem.
vanishes at the arbitrary place a
but that
0, 1, a, l
rational
integers,
{x-p-qa)-\
z is
...
let
p+qa,
the
an immaterial condition.
is
wherein
is
<
....
^= "^
+ ^3- + -"ii- +
Also
=
x-rrii
=
when
m-i is
nij,
Wj,
...,
Here
when
+ a,
is
j;
-ir
a -mi
/x-a
?i
\
|
vii, c)
x^-d^
5-
5-
o-mtj
<
Ix.
wij^
m.i''
<
nh
+))
/
204
Now
is
the series
convergent.
where
all
[144
(i is
x^ a^
a?,
3^
<
r-+...
fi,
values of
= 0.
TOj
x a
+-
1_1+
This function
145.
We
f (a|
is
c,
is infinite at
1, m).
a place
Ci
like
is
- 2 j^Dlf{x,a;
this
^-" 1
and
may be
Ci,c),
fi {a).
not necessary
be known, or even that
there should exist rational functions infinite only at the places C; in the
that the expressions of the rational functions/;
assigned way.
All that
is
necessary
C;,
is
144, it is
(a;)
should be assigned.
single-valued
Ci,
c.,,
function, having
...,
be as in 144.
the places
...,
which
Cj,
c.^,
is infinite
...,
We
can construct a
as zeros, of assigned
c,
where
147]
E {x, z) = J><^'-
205
^' "'''''
When
c.
is
near to
we can put
B, log E (x,
z)='^\ W, ^
r=0
(07,
c)],=
z,
1^
and
therefore,
when
Ci
is
near to
c,
and x
is
c,
we
can put
wherein
place
c,
i2,i(a;) is
and
ti
Let the
s+
1,
convergent, let
Then the
fi,i
+ f+2 +
>
ti
Ri, k ()
> s,
<e,-
series
i\i\ogE (x,
a)
+ I
1=1
(\i log
i=+l
E (x,
rm)- i
Ri, * (x)
ti
k=0
only at the
and, taking
fj+l
is
infinite
is
144.
numbers
is
fii,
absolutely and
uniformly convergent.
[E (x,
Ci)Y'
With the
is
results obtained in
well-known results
144
The
ii.
i.
p.
66, or
Mathem.
they
following results possess the interest that they are given by Abel
(Abel, (Euvres Cumplkes, Christiania, 1881,
;
146, the
46 and
vol.
ii.
p. 46.)
206
Ex.
^ ix)
If
i.
and
/(:<;)
be a rational polynomial in
be a rational function of
x,
x,
=n
(a;
[147
+ oj)^t,
= 'S.yta^ + S
^-
then
{x-z)cj>{z)
'^
{z-x)^{z) "
fc.'^*;
directly
^w
by noticing that
if
<p{x,
z)=
^(2)(^-z)
then
0(Z,^)^<^(^, Z) =
is
Denoting
a rational function of X.
we
{{}l-x){l-z)^{X-z){X-xf
^-^j^
it
Ex.
ii.
prove that
wherein
A^^i', is
This result
may
-^ (x) is
the product of
all
^W=TS^*(^'^)i*("'^)
as in the last example.
Ex.
iii.
result put
i= J, and
ii.
</)(a;)=[i/r (.i-)]".
In the
These results are extended by Abel to the case of linear differential equaFurther development is given by Jacobi, Crelle xxxii. p. 194, and by
tions.
p. 177.
148]
CHAPTER
VIII.
The problem
logarithm
as a
sum
generally,
jdxR(a;,y,y
wherein
parameter, and
as a
sum
....y*),
jdxR{x,
But an
and logarithms of
rational
'J cut?
+hx + c).
\dxR{x,
cannot, in general, be expressed
s/aaf
-\-
ha?
+ ca? + dx + e)
by means of rational
or logarithmic functions
such integrals lead in fact to the introduction of other transcendental functions than the logarithm,
namely
to elliptic functions
and
it
appears that
the nearest approach to the simplicity of the case, in which the subject
of integration
exist for the
is
a rational function,
sums of
is
equation
f^'
dx
Jo V'(l-ar')(l
7a-ar')(l
/'"^
('"^2
.
dx
->^''"io
io V(l -ar'Xl - AV)
-/t'ar')
["'
Jo
^
'^{l-a^)(l-k'x'}
_.,
INTRODUCTORY.
208
[148
provided
^3 (1
On
- kWx^) =
:^
V(l-a;/)(l-fe'') +
statement
and
it is
it is
x,
V(l-a;,^)(l - k^^).
x, to
introduce a quantity
y,
such that
y^-{\-af){\-l<?x')
and
any value of
to regard y, for
or negative sign
in fact
x,
0,
x=\, a;=r,
and back
value, through
we can
A/
'(0,1)
i(o, 1)
2/i
i(o,
3/2
Then the
ys
1)
The
where the limits specify the value of y as well as the value of x.
theorem holds when, in the first two integrals the variables (x, y) are taken
through any continuous succession of simultaneous values, from the lower to
the upper limits, the variables in the last integral being, at every stage of
the integration, defined by the equations
The quantity y
introduced
is
is
called
/fc').
its
R {x, y) be
= (1 - ao') (1 - AV) we
x, y,
the
of similar integrals
R{x,y)dx +
J
{"i,
R{x,y)dx
J (a ,b )
m m'
bd
can be expressed by rational functions of (a;i, y,), ..., (x^, ym), and logarithms
of such rational functions, with the addition of an integral
/(itm+l, J/+i)
^{x, y)
(oi, 61),
y,), ...,
{Xm, ym)
is
y,
(a:,, y,),
...,
quite arbitrary
dx.
but in the
..., (a;^,
ym+\ are
INTRODUCTORY STATEMENT.
1-50]
209
mined, and it is understood that the relations are preserved at all stages of
the integration, so that for example a+,, b^+j are respectively taken to be
the same rational functions of (a,, b,), ..., (a^, 6^). The question of what
alteration
is
observed,
is
of integration
Riemann
is
not
R {x, y) dx
/;
consideration of the
is
This question
altered.
is fully
treated in the
called
period loops.
149.
Abel's theorem
may
function of
may be
x defined by an equation
f{y, x)
wherein
be enunciated as follows:
the equation
A^
of the form
= y" + A,y"-' +
+ 4 =
x,
0,
side of
R(x,y)dx+
limits, is expressible
R{x,y)dx,
by
y,), ...,
i^m, Vm), and logarithms of such rational functions, with the addition of the
.sum of a certain number, k, of integrals,
wherein
z^, ...,
R{x,y)dx-
R(x,y)dx,
.rj, 3/1,
...,
2/,,,
as the
Vm-
(m, ym)
may
be supposed to hold at
all
stages
of the
integration;
in
particular they determine the lower limits of the last k integrals from the
The number k does not
arbitrary lower limits of the first vi integrals.
depend upon m, nor upon the form of the rational function R (x, y) and in
general it does not depend upon the values of (a;,, jr,), ..., (a',, .V,,,), but only
upon the fundamental equation which determines y in terms of x.
;
150.
equations determining
is
are
regarded as variable.
By
some
or all of the
continuous variation
of these
14
210
0,
t)0,
{y,
[150
x and y which
satisfy
vary continuously.
will also
both
Then,
if
in
number
...,
the equations
terms of
Pj
Xr)(a*
/(s,-, Zi)
= 0,
6(si,
...,x,, y,.
2i) =
As a matter
Finally, from
we can determine
s,-
rationally in
:<, ym, which appear on the right-hand side of the equation which
2/i.
expresses Abel's theorem, are rational functions of the variable coefficients in
i.
e{y,x).
151.
When
6{y,x)
is
form Let
(a
b,),
...,(, 6g) be
sequence
Q-q=p-r-l,
R{x,y)dx+
is
'
-1-
-(,
R{x,y)dx^
-\
(a:,, y,),
and logarithms of
sum
''
y)
R(x,y)dx
3/,)
R(x
"
'"''^
R(x,y)dx
152]
herein
it is
The
places (oi,
function
Z {x)
vanishes in
are
{x^,
bi),
...,(ag, 6^)
if d^ {y,
and
3/1),
may
...,{xg, yg),(zi,s,)
(a"i, 2/1),
211
may be
x),
same
{y,
rational
when
x)
when
its
it
zeros
function will be of the kind, hitherto regarded as special, which takes the
same value
Riemann
a;)
definiteness,
dec
we can
negative powers of
t,
in
express
R (x, y) -n
in
series
of positive
is
and
denoted by.
|m-lt^+|m-2^ +
then, if P^'
at
^, 7,
"
infinitesimal at
f,
the places
infinities
'''"B{x,y)dx-A.F^-;-A.Ef'-A,D,E-^-
-A^D^'E^^
la, b]
in the expansion of
R {x,
2/)
;^
>
is finite
we obtain
+ A,,B;--E;-'1
(a, 61
wherein
SA,
is
7, c
same
for
[l^^
2\2
if
u''"
denoting the^pkce
riZ(.,2/)<fo=S(A.
J
and
(a, 6),
first
C,, ....
+^.i>r)i1;; + (?><"+
+ ^.2),+
kind, a
+ GpK'"-
The consideration
of the
sum
r R{a:,y)dx+
wherein
o^, ...,
the places
(x,. y,),
...,
"
R{x,y)dx,
h,), ...,
s,). ...,
{Uq, Iq),
and
(^p_- Sp-.-0,
is
x,,
...,Xq denote
thus reduced to
M^'-'''
{i=l,2,...,p.)
+u'.''"',
Ex.
i.
By
+[{x,
where (cf
46,
ir'"Vi(^,y)+...+(^,
II., it
follows that
any
l)''"-'"^</)n-i(^,y)]//'(2')
Chap. IV.)
I)
at the place
|.
ii.
Xl[(l, ^)-(ll, )]
X,, Xj,... being constants,
We
153.
+ >^2[(l2.
'^)-(f2.
)]+-.
in
can be
(cf 130,
Chap. VII.).
the
conventions that the upper limits are coresidual with the lower limits, and
that the
paths of integration are such that at every stage the variables are
paths of integration
Riemann
all
values of
i,
154]
213
a,, ... a,, are the poles, will contain q + l arbitrary linear coefficients,
entering homogeneously, and the assignation of q of the zeros, say Xi, ...,,,
places
will
...,
...,
t=l J a
determi-
aq.
R(x, y)dx
(i,
of Abel's theorem
is
'S,
= s A,hgZ(^) + A,^^ +
where Z' (f ) =
2^, =
because
Oi, ...,
D(Z (f)
Oq and
the term
(Chap.
x^, ...
20
II. p.
154.
can be omitted
a rational function of
is
Xg.
Z (f)
Herein
(8)).
a simplification
first
a,, ...,
Riemaon
Let
trace on the
all different;
be the value of
places at which
places as
fi
following the
in the
/j,
Z
Z
first.
fj,
The simultaneous
q>
1,
is
a^, ...,
Uq;
but
it
particular one.
fi,
dl
Z-fi dx'
I,
= \R (x,
sum
II. p.
20
(S)
Chap. VII.
the infinitesimal
zero.
la accordance with
Of such
t,
is
first
the
137,
(Z - nY^dlldt,
places where
Z is
occur,
Ex.
vi.)
in
terms of
is
equal to
equal to ^.
then the
the
sum
We
of the
^d^ldt\dt)'
* Sets coresidual with
Avith
\dfiJ,
\d(Ji}q'
sets
214.
provided
Z /xhe
provided
dZ
the convention
made
this
sum
is
we may regard
we
satisfied.
equal to the
is,
In accordance with
first order.
Hence
[154
sum
{Z /i)~' dl/dt
we may write
where dl/dt
is
in the form
/dl\
/dl\
fdl
\dfx,Ji
ydfji/Q
\dt
Z fiJt-i'
many
purposes
infinite.
In
right
i^'",
in
regard to
fi
first
from
oo
kind, the
to 0,
we
obtain
+w*fi'"e
?t^"'''+
In case
ft,
hand
is
= 0.
= P^' "
(A)
and
say,
Z be
not equal to
equal to
+
Z{^)-^^
Z{r^)-f.'
-
hence, integrating,
while,
if
Z (x)
fx.,
respectively denoted by
^1
i'.
and
"^ Q>
a'q
we have
^'
= log Z{^)-fM Z{y)-
Z(^)-v/Z(y)-u_
Idt
which
is
^ Z{a;)-v
be
REMARKS.
155]
215
155.
by the path of
we may
fi
the
it,
in the integi-al
by
There
no need to extend the equation (B) to the case where one of the
left passes through either f or y, since in that
case a corresponding infinite term enters on both sides of the equation.
is
But
it is
should be coincident
may be removed.
immaterial, and
is
limits
And
a'l, ...
if
two
x^
(or
more) of the places at which Z takes any value, /m, should coincide, the
equations (A) and (B) can be formed each as the sum of two equations in
to Z = /x and from
which the course of integration is respectively from Z
Z= to Z = 0, and the final outcome can only be that the order in which the
upper limits x^, ..., Xq are associated with the lower limits ai, ..., a^ may
undergo a change. But in the general case we may equally put, for example,
=x
fj.
rx,
dl+
a,
a-i
rxx
rxi
dl+
dl+
dl,=
>
J x^
a,
C^i
C'
dT+
J a,
x,
r^2
dl=
/"a^i
dl+
a,
dl,
.'
Oj
with proper conventions as to the paths hence the condition that dZ shall
not be zero to higher than the first order at any stage of the integi-ation may
be discarded also, with a certain loss of definiteness. The most general form
;
of equation (A),
when each
of the
is
of
course
+ u^fi-''8 = iVia),-,i +
''+
It*-"
where
...,
(,-,i,
sufficient
w'f.p are
integers, independent of
+ i/pto,-,^ + ilfi'w',-,i +
We
i.
+ Mp(o'i^j
...,
(C)
...,
is
in
If,
equation (B),
we
substitute
we
for
it.s
rational
shall obtain
in x and y, of
where, now, e^{x), e,{x) are any two polynomials, integral
zeros,
and e,{x) has
x^^ for
which, beside common zeros, e^{x) ha.s x^
If in this equation we suppose any of the coefficients in
for zeros.
e ix)
is,
for shortness,
common
zeros of e^{x)
{y, x).
CASE
.j^Q
and
.,
<?,(.)
remain
fixed,
,Xchangmg
...
[15^
IN
thereby to
^.
+ d^.,
..., :r,
0= (.)
+ d^, we
+ S0. (.0,the
shall obtain
,^fi:..,= 81og^^
'^,...
before given, in
shghtly more general than any equation
with
^.. ..., x
coresidual
places .,+ da:,, ..., a^, + da; though
value
same
the
has
necessarily snch that the function ^,(^)/^,(^0
of
course
the
in
Abel
them. This general equation is obtained by
which
places
that the
is
not
are
at all ot
his proot
of his theorem.
the equation
For any Abelian integral we have, similarly,
"-which,
-,^-.=[i*'^''4v
may be regarded
also,
Abel's theorem.
a a complete statement of
156.
Z{^)
function
One
grade
(/-
Z (x)
be an integral polynomial, of
which is adjoint at all places
1) cr + - 8 (cf Chap. VI. 86, 91),
double point of
two, say A, A', wliich correspond to an ordinary
case of this
may be
noticed
if
f {y, x)
except those
-'
= 0,
the integral
iy)
Hence, it
been
have
A'
the places A,
will
Z{x),
= 0,(x)! e,{x),
its infinities.
in these
we
Z (x)
will
and
shall obtain
y*"""
kind
for
is
result
+
:
if,
third kind.
8,
of double points,
83,
p+S
integrals as
if
will
For example,
let a^,
...,
we overlook
(cf
= 0.
aggregates of
Y''Q'<^e
A',
I. p.
101, 93).
Take 5 (= Q
-p + t
-t-
167).
(^-poly-
1) arbitraiy
157]
places
c'l,
..., Cq,
aud
rational function,
places
Let
Ci, ...,
Xi, ...,
Cj, ...,
?(a;),
Cq coresidual with a^
for
rty.
a^, ...,
determinate save
217
a constant multiplier.
Xq be any set of places at which f (*) has the same value, A say,
xq are the zeros of ^{x) A then, as a,, ... a^ are the poles
...
so that
a;,,
of f (x)
A we
and as
i;'(Ci)
have
^(Co)
= 0,
is zero.
Hence, calling the places where a definite rational function has the same
value a set of level points for the function, we can
make
the statement
the
"^^''^ dx,
+ ^-^^ dXn = 0,
dxQ
dxi
Ci, c.
In particular,
when q =
\,
c.^,
Hence,
..., Cp+j,
which no
in
"^^^ dx, +
Ex.
157.
i.
We
+'^-p^
dxj,^,
0.
the integral
;
is
becoming
j,
= cc.
Consider the
rational function
.1,..., L, is of
instance,
being given by
To
/dl
the place
to be considered,
we put x = t
y=J^.(i-i^i''-i^2<-
).
and obtain
EXAMPLES.
218
[157
l+^Aot+
1
^-A
l + .Z-L
and therefore
dl
rf
Z-^
1
1
^ J - ^0
l-^<2T-2 (l_^)2
Hence,
if
u.-i,
...,
.t-'-n.
is
5 (A-A^,)
the zeros of
>/+
(x^,
the^ + 1
have, from
>/i
1-Al
..., ^ij^
-.4)j'^
+Kx+Z = 0;
U {.r) = + A.t;>' + BxP-^ +
using
P^x) for (.v-Xf)
and
-,
+ i,
J/p
..., {x^p+i,
+ 1) are any p + \ of the places {x^,
t/
where
c'l
(1 /i)~^.
l/-,i,
1/1),
(^-i'p +
1/2,,
j),
i),
we
equations
+ Ax,P + Bxii'--' +
+Kxi+L=0,
(i=l,
2,
(p + 1)),
i>+l
and hence,
(*-
a,,
,),
if
61,62,...
be the values of
If in the integral
/ the terra
be
aj, ...,
and
/'(.?;)
= (*' aj)
...
"1+
sum
+7'2P+I'%+1
zero.
The reader
can be deduced,
!t,,
u^,
for
=!
tecomes
1I3
which
when a;=a,,
x''
/'.'
will
1/
we have
-I'l,
...,
*'.jj,
gives
sum
is
157]
HYPERELLIPTIC CASE.
219
kind
first
du dx\
-j:
._,=0
\{Y-b){Z-c) tTT.
dx dtjt-'^''
that the
It is
F- b, Z-c
Z=c, added
are of the
fii-st
to
the
order.
r=
\]a,x~h
\\dx^ Z{x)-v)i'
,j.,
Ex.
iii.
^ {x)
for
integral jwlynomials in
x=b.
.,,,
of which
x = b.
f{x).M^{x)-yHx) = K(x-x;)
where A'
{.v\
some
{x-x^),
f^,
where
C is
f^ <Hf dx _l^
<t>ix)dx
a constant, and
R is
<^W__
descending powers of x
N{x)+M{xWf(^)
(^-')^//(x)
in
y{z)+M{z)Jf{i
^ N{x)-M{x)^f(^)
must be
used consistently.
theorem when
Ex.
iv.
direct proof.
when
zero
is
z is
The
chosen
is
<^ (x) is
result of Ex.
to
iii.
is
of x,
let
(y,
x) be
=6
{i/i,
i.
with
Let/ (y,
.r)
ct
/ (y,
,r)
corresponding
e (y, x) = 0,
x) e (y., x)
finite zeros of
p. 445),
. .
x at the
l (in
r {x)
of order 2p +
is
when f {.r)
/ (x) = 0.
a root of
We explain
valid
is
contain any of the variable coefficients of 6 (y, x). Let ^ be a root of F{.i-)=0, and
then one or more of the places (f, 7),),
1)1, ..., >; be the corresponding values of y
;
are zeros of 6
(I, >,)
if,
[157
220
{y,
into e {y,
.c)
+ Se
.r),
{y,
coefficients of 6 {y,
x)
lix
by a shght change
F {x)
and
-r),
8F{^) + F'(,$)di = 0,
Fo
where F'
{$)
= dF{^)jd$.
Denote now by
U{.v)= 2 6 (y,, .0
e (,, $) se
e (r,i_ I) 6 (,,^1, ^)
&F{^)=sr (|)= 2 e (,
(I)
U (x)
$),
(,,i,
6 (y, x) hB
6 (yi.i, x) 6 {yc^i,
.v)
.i
and
y,
>;)
jrj^yp^^^)
{j/u
x);
=i
then
if
y) be
{x,
i<-
where, on account of 6
{rj,
(f. l)
|)
IJ
d$=-K
we can
''
(^,
we have
write
^^
'^^-i&'' ^'"
and
1=1
= *(!).
say,
only.
(I) being a rational function of |
for all the zeros of 6 {y, x), we have
(j>
this form,
^R{^,r,)dS=-2jrf^^j^y,
herein the
summation on the
right
out,
and the
perfect diflfereutial of a function of the variable coefficients of 6 {y, x), in fact in the
form
\R{x,y)-&loge{y,x)\^_^,
as
we have shewn.
For example, when
x)=y-mx-n, we have
F {x)=x^ + {mx+nY3a.v {mx+n) l,
{y,
Fg(x) = l,
and
i$
V'
SjjibF
fM
(g)
F'(^)
^g)
_
- F' (i) ^^
^(g;)
I
i+,3-^ .fj(^_|,)/;"(g,)'
F{x)
and hence
is
^ix)^_Zm^^
Now
as
easily seen.
^
[x,
4. =
From
xydx
this
Z)
we
T":
= - 38
-r-,
-,
infer
^mn-a ^ fmn-a\
= -3=
+ 3 r-: f)
5-
.Xi-x.^
-.
,
'={x,+x.,
+ x^
,
-^
Abel's
157]
In
221
this
infinite; putting
where
= l,
dx/dt,
or
a+n
m+l
x.
finite
If Y,
V.
am^ + n
m+
"
<o
and we
<-',
aa + n
=\og
{7/
easily find
_3 (a-mn)
m + ufi'"~'
m' + l
'
VZ
for
infinities,
any value of
x,
and [2
values of
for
conjugate values of
x) dljdt,
is
which
is infinite,
it is
easy
(cf.
sum
2 (VZ), for
of the
all finite
(^S^l. = [^(^^)^-'-[^(^^)](-a)-"
sum is to be taken only for the
/ be any Abelian integral, =\R{x,y) dx,
which
is infinite.
Hence
if
we have
if
we assume
we can
obtain
the zeros of 6
{y, x).
This
Vol.
i.
is
p.
'F{i)F'{iy
is recommended to study Abel's paper* which, beside the theorem above,
enquiries first, as to the form necessary for the rational function
important
two
contains
The reader
dl/dx, in order that the right-hand side of the equation of Abel's theorem may reduce to a
equation of Abel's theorem,
constant, next, as to the least number of the integrals in the
vi.
Obtain the
re.sult of
Ex.
i.
( 157)
in Ex. iv.
and published by
Which was presented to the Academy of Sciences of Pari.s in Oct. 1826,
this period many papers were
During
t. vii.).
savants,
divers
par
{Nemoiren
1841
the Academy in
Abel, Minding, Jurgensen, Broch, Kicbelot,
published in Crelle'8 Journal on Abel's theorem, by
I have not examined all these papers with care.
i
xxx.
Crelle,
(See
Bosenhain.
Jacobi and
222
Ex.
vii.
sum
[157
U.v
J
'
is
Ex.
viii.
Crelle,
</>
t.
v.
158
p.
xiv.
wherein
ix.
t.
nova algebraica,
(x),
yjf
(x)
>!,
polynomials in
^,">"''+^IV"'+4'y'^^
j^g
=f{x),
(x),
x,
{[V^TiH^)+V*wTTI)]/2
^/(l)/(v)}=o,
wherein x^,
x^, ^ and TOi, m^, y are coresidual with the roots of <^ {x)=0, and
^, y are the
places conjugate to | and y ; conjugate places being those for which the values of x are
the same.
When
158.
p- t -\
aq,
by
which are given by
Hence the p +
1 relations,
Abel's theorem,
ut'-'"
+M>''fi
nT+
= 0,
+p'ir;'='^og{Z{^)iz{y)],
cannot be independent. We prove now first of all that the last may
be legarded as a consequence of the other p equations. In fact, if x^, ... ,Xq
and Oi, ..., aqbe any two sets of places, such that, for any paths of integration
"^
(i
"'
/"
"'
^fm.
ill'",
M,,
...,
+ M^(Oi,j, +
o>';
a,, ...,
set
+ i/'pO,',-,,
of linearly independent
ilf/o)',-,
of
i,
",
and
a;,
...,xqfor
zeros.
=Gi,,U,
C'i,pV
first
kind,
.so
that
we
are given by
(Ser. iv.
t. iii.
v. vi.).
1-^^]
wherein
w,i<i,j
223
+v?-'''-M, + M/r,,, +
0,-^papj
=0
or
1,
and
we can deduce
''?""
+i)/',r,,.
Herein an integral,
about the place
circuit
Z{x)
Tll\"^^,
x^
And
is
2'"'(''-""'
is
...
+v^'"''0)-2^{M\r,,,+
unity.
Z (x)
i.
It
is
+JtfV,.,)
*,, ...,
it
has poles
is justified.
y, f,
integrals
^ is
y.
(Cf. 6,
Chap.
I.)
equations
M^,
wherein
...,
Mp, M\,
...,
j',
should be
simultaneously true.
St.
exist, yj
n.
If
and y^
hyperelliptic.
^^
th^ poles of a rational function of the second order, and the surface
(Chap. V.
52.)
is
224
]
now
In regard
59.
to the equations
+ >" =
<""'+
express that the
which
ttj,
(ti'
[159
places
the
with
coresidual
are
x^,...,x,(i
o,
places
CA(a;)+
wherein
C,
.
.
C^
+Cpnp(a;),
vanish
which
constants,
are
these
in
places,
then,
and the
equations are therefore equivalent to only p r \ equations, determining
the other places oix^, ...,xq in terms of those assumed. This can be stated
Q _p
also in
T-
-I-
assumed
another way
the
arbitrarily,
differential equations
p-dx^ =
+ axQ
pdx,+
dxj
(i,
{i
= l,2,...,p),
express that the places x^, ...,Xq are coresidual with the places
x^,
+ dxQ
if
are independent
if
a;^
however t
a;,
+ dx^
...,
the form,
C\^+
dx
wherein
Ex.
i.
= 0,
+C~^
di:x
''
as poles of the
x^, ...,x^,
^t
determinable from
1 of
then the
"p
them.
first order,
X,,...,X are the coefficients of the inverses of the infinitesimals in the expansion of
the function in the neighbourhood of these
plaw^es,
-?^iC-
-'^'^C:
p equations
But,
if
+\a,{x,) = Q,
we take consecutive
places
coresidual
tlius, if
the
first
= (J-/i + t + 1)
of
{1
= 1,2,
with
./-j,
we
...,p).
..., .i-^,
and
tj,...,(^
be the
also have
-t-Q,(j:s)5=0;
,, ..., t^
be taken proixjrtional to
we
shall
When
the set
= Zi,/Xj,.
ii.
two double points, there is one finite integral if oSj, ..., 04 represent any 4 coUinear points,
and .c, ,...,T^ represent any other 4 coUinear points, the equation of Abel's theorem is
;
160]
225
but this equation does not express the two relations which
are necessary to ensure that
.r ...,x^ are coUinear; it expresses only that x^,
x^, x^, x^ are on a conic, S, passing
through the double points, or that x^, x^, x^, x^ are the zeros, and Oj,
....o^ are the poles
of the rational function S/LL^, where Z=0 is the line containing
a^, ...,a^ and Z=0 is
the Kne joining the double points.
160.
the
d^,'^+
have algebraical integrals,
that
(^=1.2,... ,p),
+d^;'^^=0'
being
first
kind.
a;,,
...,
Xq con-
with
arbitrary
lots of sets of
fundamental sets of
lot
number
of
all.
arbitrary constants should appear in the integrals of the equations (Chap. VI.
96).
by Ai,...,Aq, ai, ...,ap, so that Ai,...,Ag are arbitrarily assigned beforehand, in any way that is convenient, and the positions of a, ...,ap are the
,
in
places
Xt,,
...,Xq,
entering therein, so
coefficients
!Ki,...,Xp,
these
to
...,Ag).
a^.
The
be
The
equations
expressed
in
its
homogeneous arbitrary
an
for
arbitrary
multiplier,
we may suppose p
terms of
resulting equations
g' -I-
determined save
a,
B.
ratios of the
that
From
-4i,
is
is
may
poles in
rti,
ctp
and
does so
as an arbitrary constant.
of the places
and
Up
Oj
it
x^, ...,xq,
Xp+i,...,x^,
say
(and
o^,
^,
. .
this
-4,
and eliminating
function vanishes in
EXAMPLES OF
226
For example,
for
Q=p + l,
if -^
[160
{x,a; z,Ci,...,
Cp)
rational function which has poles of the first order in the places z,Ci,...,Cp,
taken
which function
1,
>lr{ai,A; Xp+i,Xi,...,Xp)
and a complete
A,
The
of these integrals
first
given by
is
= 0=
(xi,Xp+t;
v/r
=-<fr{ap,A;
Xp),
ajp+j, arj,
by
= =
=-^
is in fact
(Xp,Xp+i; A, a,,.
Qp).
the equation
= 0,
dx.p+i
du
dup
dxi
dx2
dP
dP
dup
dx,
dP
dx.'P+i
wherein
dx^
P = P^" ^,
...,
the equation
156)
and {A,
a^
sets.
Ex.
i.
For
p = l,
...,
differential equation
where
= (a,
6^
Shew
1)4.
infinity.)
also that this integral expresses that the places (x,, y^),
(x,^,
y^,
(a,
b),
are the
are varied.
Ex. ii. For p = '2, the fundamental equation being y^={x,l\=\^x^ + ..., using the
form of the function yfr (x,a; z,Ci,...,Cp) given in Ex. ii. 132, Chap. VII., and putting
the place
'
1/1
Vi.
y%
y-i.
Vi
Vz
the integral
Vi
Vs
{x,-a)F'(x,)^{x,-a)F'{x,)^{Xj-a)F'{x^)^F{ay
wherein
/'(j')
:-\,
(a, 6) is
161]
system of complete
By
227
(a,
6)
we
obtain a
integrals.
Shew that this integral is obtained by eliminating p, q, r from the equations which
express that the places {x^,y;), {x^,y^),
(^3,^3), (a, 6) are zeros of the polynomial
y \3?+pa^+qx+r.
Ex.
iii.
{x,
y\
in
is
of the
is
given by
(2, 3, 4) t/,
where
form
+ (3,
1,
4)
,3,4)
U^ + {\,
2, 4)
U^-{1,
2, 3)
U, = 0,
coefficients, etc.,
METHODS OF EXPRESSING
228
(and Ai,
system* of
Now
J-g).
By taking
l,2,
[161
...,
p we
obtain a complete
integrals.
...,
Ag,
Uj, ...,
Up as the arbitrary
quantities
<^-polynomials
Ci,..., Gjt,
. .
. ,
with
bi,...,bp
Gi,...,Oji, in which
have
any
convenient
positions
assigned
beforehand,
so that the
Bjt^p
i
lot of sets coresidual with (7,, ... (7^ depends upon the positions of bi, ..., bp.
B^,
...,
. .
determined
Bfi_p,
coresidual
Q+R
e = M^ + A',^,+
wherein
fj,,
..., fit
Q + Rp.
sideration
is
+ fik%,
and k
is
for shortness
written for
Then an integral of the differential equations under conobtained by expressing that the places
obtained by putting
in turn equal to
1, 2,
of integrals
. .
. ,
is
is
p.
the places
must not be
Ex.
i.
less
.set
is
no
^)
(^,
y\
C-^,
y)i
1, 2, ...,
than p.
f (y,
a
+ (^,
y)-,
(',
y)i
constant
= o,
R,
p.
.save
that
1^1]
2,
229
3,
convenient position.
q=p + \,
we
li+p + l
such an one
variable zeros:
i.s
-^H,
An
+ .,.)-(Xxi>*iJrFxi'^... + Hf = [^-i\F)F{x)^{x),
{x-x^^^),
4, {x)
= ix-b,)
may
{x-b).
are zeros of this polynomial, and u complete system of integrals, in either case, by taking
i in turn equal to 1, 2, ..., />.
We may
in the
form
-y+^i-J' + i + i^i-"-)then
R=p+\,
infinity
we obtain E=\,
...,
+//;
as above.
Ex. iii. The integration in the previous example may be carried out in various ways.
By introducing again a set of fixed places iij, ..., a^. A, coresidual with x-^, ..., x^., a'p + i,
we can draw a particular inference as to the forms of the coefficients F, <?,..., H. For if
U {x) denote X.r p * ' Fx"" +... + (?, and U^ (x) denote what U (x) becomes when arj
.^p + j
, .
and
also
...,
Fg{x) = {x-aj)
Op,
divide
by
(x),
A, the
coefficients F, G,
{x-a,,)
(x-A),
(x)
ii,..., 6p,
+ t<j)
ff being then F,
each
because
then,
namely U{x)= U^
...,
(x),
of
the
the polynomial
where
is
(?,
...,
If,
polynomials
U{x)-Ug
(x)
a variable parameter
we
F=F,+t,
G=Go+tt^,
H=ff+tt^,
and the quantities G-tiF,...,H-tp F a,Te constants in the integration, being unaltered
when the places a;,, ..., :rp + i come to aj, ..., a,, .4. Hence we can formulate the following
result: let the p-\-\ quantities F^, G, ..., H be determined so that the polynomial
-y+ U^ (x) vanishes in the fixed places aj, ..., Op, A. Then denoting {x- aj...(.j;-ap)
(x-A) by
is
an
integral
polynomial;
denote
it
by (,i-2/' X) (j^^+i,
.i.'P-'+
+ g,
so that
230
F{x)
ioT x''
uniquely determined
are
(f),ti,...,tp
+ <p.
+ tiX''-^ +
<p
and the
in
and put
(x)
By
equation.
this
in
[161
t,
which
is
in the
polynomial on the left-hand side of this equation to the coefficients in the polynomial
{fi-2Ff^\) F{x),
of the
^ X^ +
/^l
tin ---
"7-
symmetric functions
Xfj ^
......
"l
Aij Xjf
t,
the place
A may
...,
be given any
Ex.
By
iv.
eliminating
/ij, ...
are connected
From
V.
Crelle,
iii.
integrals.
by a quadratic
Ex.
of
relation,
32, p. 220).
t.
the equation
P +1
Uix) _
yr
F{^) ^rll^X-Xr)F'{x,)
we
infer
/'+XA,=^2'
whore Aj = ai + ...+a?p+j
we have from Ex. ii.
and
js^
hence
if
a be the value of
.r
^^^
values of
(a)
</>
may
Ci,
case
integrals
...,
paper
V.),
p=l.
shewn how
xxiii. (1842), p.
369.
to obtain integrals
1876, p. 337).
Ex.
vi.
By comparing
coefficients of x'^" in
..
where
In this
is also
A,=.'!;i + ... +
C=i^^)i
j;,,
+ j;
(,
ii.,
we
obtain
- Ai),
~''(^i +
-+
^'' +
i)-'^'(^'i
162]
by Ex.
ii.
is
we have the
generality,
where
(cf.
C is
Ex.
Ex.
i.
a constant
\=0, n = 4, which
is
is
an
not a loss of
integral
this is
231
for
/>= 1,
157).
Shew that
vii.
if
+Li;+M,
^?...^.jn'
(Richelot,
Ex.
loc.
cit.)
If
viii.
T-x(i+...+^)-ir(l+...4-Ly^const.
,1:,
0(1,
Oj
equations
(oj-^p-n ) / (np-^i)
(a^-A)
(oi-ap) /
(gj-^i)
[ca-A)
(gp-'^p-n) ^/,
{a^-a^)
S2
"^'"'''^ '
.
wherein the quantities A,...,ap are the values of x at fixed places coresidual with
Xj, ..., oTp+i, Pi is an absolute constant, and ;* is a parameter varying with the places
Take
^1,..., a.p+1.
in turn equal to
1,
if
2,...,
((?i-l)p,Pi(pj-pO + {(?j-l)pitPi(p*-Pi)
(Weierstrass, Collected Wm-ks, Vol.
162.
The proof
i.
+ (G'*-l)PiP>(Pi-Pi) = 0.
p. 267.)
assume that
for
any
value of x, say
finite
x=a,
the curve
is
completely given
= a-K2"'+',
y = P.{U)
2 = QAt.)
= -(-,'+',
2/ = Pi(i)
z = Qi(ti)
a;
wherein w^
l,
= a <fc'.+'
y = Pk{h)
a;
-I-
,^=Qk(tk)
(D)
Qi,..., Pk> Qkj denote power series of integral powers of the variable,
with only a finite number of negative powers, which have a finite radius
of convergence. The values represented by any of these k columns, for all
and Pi,
values of the infinitesimal within the radius of convergence involved, are the
coordinates of all points of the curve which lie within the neighbourhood
of a single place
(cf.
3,
{Wi
Chap.
1)
I.)
the
+ {W^ + 1)+
sum
+ (Wk+l)
232
is
all
values of
x,
and equal to
is
We
n,
[162
we have only
to write - for
a.
sets
rational function of x.
sum
if
(a-,
y, z)
R (x, y, z)
U
And
{x)
=R
sum
at all the
equal to the
a;
= oo
is
+ -R {x,
z^)+
{x, y^,
U (x)
is
-y,
x, y, z,
x,
y, ^).
R (x, y, z) -j-
at all
in the expansion
t~^ in
when x = -.
t
vLt
-^=0,
which holds for any rational function, U (x), of a single variable (as may be
immediately proved by expressing the function in partial fractions in the
ordinary way), enables us to infer, in the case of the curve considered, that
also
ii(^,2/,^)5
By
0.
[R{x]y,z) dx
and therefore
R (x,
/t,
^'"'
of poles of
^'
S]-' =
R {x,
y,
number
z')
is
equal to the
of places where
number
R {x, y, z)
y, z),
'">
y'
z,
when
in terms
<
is
162]
Hence,
this
/ be any
if
theorem
JR
(x, y, z) dx,
we can apply
in the form
^
=0
dtZ-fiJt-'
Z
233
y, z,
it,
'
the theorem
^^""^
and
if
nomials,
+^"^^=(xX^4'^^^^-4-
of the form
is
0.^
=
wherein
^i
Oj, ...,aj;
.
.
Xk are
are
r^
the
places
('^'
^^
y'
Z =0,
at which
or
= 0, and
and the places
or 6^ (x, y, z)
6i{x, y, z)
0,
infinities of dijdt.
for
by Forsyth,
Phil.
Trans., 1883,
finite
is
x = \, we have
y= +'Jx
For
^=
p. 337.
Take
For any
ino-
d, 6^
side
Part
where
infinite values of
putting x = -
.t,
{lx),
z=_i\lx.
we have two
places given by
.1
.1
'=7
For x=l, putting
x=l +
fi,
4 places given by
"=7
we have two
places given
by
y=U + ...
y=i(+...
2=+(l+|i!2+...),
2=-(H-il!2+...)
and, aXx=Q,y==Q,-i = Q,
so that there
is
dx dy dz = it 1 1 or =^1 :-l
a double point with = 0, ?/= +i for tangents.
Consider now 2
1=0
or
=0
:-
.i-
J y^
:.-\-ax
+ by = Qio
those of
+ a'x + h'y = 0.
[162
EXAMPLE.
234
Put
/= [
then
j yz
-^
^, = yzdt
when x
^^'^
'g
-2<=7
to the
sum
+ 6'
-z
o'
- --
log
we
place at x-=0
^dl
log
d(dt
2
one value,
0, has, for
(1+^1:2+...),
the quantity
near to
1 + 6'
t + a'fi+b't{l-ifi...)
+ a'x + b'y
7+^^+6^^^g ^ + a^H6< (l-itK..) =^g TT6
z
If
ia
dt'
rfy)
+ 6'
^_
"^1+6
,
^"
.
^ ^ ^S y+l,
,-i
z-\-a'x+ h'y\
^l+ax+by) t-^'
/1
+ 6'
'ls(i^6'
It is similarly seen that
l + 6\
/ T^b);
co
Now
dxi
dx^
/I
+ 6'
l+6\
and thus
f^'
which
*L^ +
A'Vl--*''
h -^_ = 2
X
'
'JXx
log
163]
235
CHAPTER
IX.
It
163.
is
in the
theory of elliptic
of the integral of the first kind should be taken as independent variable, the
variables,
x and
y,
The question
as dependent.
advantageous,
possible,
if
We
it
may
not be equally
to
linearly
first
if
kind, the
equations
known
Ui,
...
Up be arbitrary
problem
of the places
to Xr
is
to
is
which may be
...,
and
a,,
...,
if
ap be
is
equations
(i=
+u^''^=Ui,
It
Up.
thus
stated
<"'
upon
from
to
= l,2,...,p),
x^; hence
*,
(i
...,
as Jacobi's inversion
,
-M>".
+u'^-'^=-u^'^'-'''^'-
uf"'^+
1, 2,
...,p),
equations,
and
is
not
Ui,
....
Up
164.
It
is
obvious
first
any
a;,,
set of values
...,
that no ^-polynomial
of places, Xi,
...
we should have
-<'"
* Jacobi,
+wf''"' =
0,
p. 55.
(i
= l,
2, ...,2J),
GENERAL EXPLANATION
236
and therefore
having
Chap.
But a further
..., vip,
?)t,,
2o>i^p,
?/ti',
..., vip'
On
i.
Xp
as zeros,
,, ..., Xp.
...,
values of
all
....
III.) to the
Let 2wij,
Xp as poles and /,
a,, ...,
[164
2&)/,
...,
I,
20)/, p
Chap. VIII.).
and
denote
to vary continuously
...,
Up
...,
to the values
Vi,
...,
Vp,
where
Vi
Ui
2iHia,;
= Ui+
and
+ ^vipWi^ p +
same
+ 2m/aj/, p
(i = l,2,...,p),
it
20.1, sny,
then
^m^ai^
"'+...
11^-'"'=
as the places
x^, ...,Xp.
For
Vi-
Zi, ...,
Zp are, in
some
order, the
the
+ wf''^=
M^""'+
And
Ui,
differ
by
a.
be uniquely
X,
Ui,
+ 2n u,+2n..., Up + 2np) =
4,{U..., Up).
are unaltered
<f>{U,
The
functions
when the
variables
U-i
2&>i, r,
2q)2,
..., 2o)p^r
2< 2<
....
2a./,,.
ir
= \,
2,...,p).
The
sign
explained above.
crossing
the
period loops.
But
it
is
often
which this
restriction
-1-M^'"''"=i7i
is
enforced.
In
237
166]
may be
seen as follows
if
were capable of
Up.
...,
This can be
<"'
+u^''^=Ui,
<""'
+ u^'-"-=Ui+2ni,
would
solution, it
x^, ...
by
158,
z^, ...,
Zp is
we should have
follow,
And
Xp.
m'^-'^= m^-"-
+ti!^-'''',
and thence
2ni
27ft,<Ui,
4-
2mp&>f, p
+ 2mj'a)i\ +
2mp'a)/_
j,
+ mpTi^ p = 0,
+ viiTi^ +
TOi
"+
2iri {rn^'vi
(where
everywhere
finite
+ TtipVp
and without
periods.
is
kind),
t)^'
",
...,?;*'''
untrue.
<"+
in this case the integral
all
first
")
+<"'"'=
sum on the
^7^;
left-hand side
is
values; and the particular period which must be added to the right-hand
side to
is
determined by the
166.
admit of
sight apparent
p dependent
why
places
it
is
necessary to take
Xi, ...,Xp.
It
may be
arguments,
It is
?7,,
not at
...,
first
Up, and
equation
wherein
place
a;
* " is
as a function of the
hand
is
this is true
first
We
all
conceivable values of
it is
238
for instance
t7
would not
oo
[166
arise,
to see that
when
definitely determinate
f{U->r 2n)
is
it is
possible
from U.
obtained would
But
= mia, +
+ Tnytop + mi'toi' +
-\-'m^(Op',
...,mp being arbitrary rational integers, and 2i),, ..., 2<ii>p' being the
periods of * " ; and it can be shewn, when ^ > 1, that in general it is
r)ii,
...,
m/
so that fl shall
be within assigned
the function /( U) have infinitesimal periods, but any assigned value of this
function would arise for values of the argument lying within assigned near-
We
existence of infinitesimal periods; for the present such functions are excluded
from consideration.
The
may be
described thus;
all
when the
Ni and N,
integer values of
number exception
;
and
it is
quantities,
of the
to
Similarly
if Oj, aa
iViO,-)- iVjO^,
possible
+ N3a^,
Nibi
be
+ N^bj + Nsbs,
...,
real quantities,
in general, values
are,
and a,b,
c^, ...,
...,c be
(k-l)
Ct
...,
N^ such
...,
N,c^+
that the
(k-l)
quantities
N,a,
are
all
+ Ntat-a, N,b,+
Hence
nil, ....
say
ill
values,
+ Nkbt-b,
+NkCic-c,
taking k
mp, to make p
= 2p,
that
integers
of the quantities
(^1)
prescribed
remaining quantity
D,p approach within assigned nearness of any prescribed value
but the
imaginary part of flp will thereby be determined. We cannot therefore
real part of the
p. 10.
EXISTENCE OF A SOLUTION.
1681
239
...
and
manifest that
it is
we ought
to use the
p new
variables
same number
of
dependent places a;, ajj, .... On the other hand, the proof which has been
given that there can in general only be one set of places Xi, ...,Xp corresponding to given values oi U,, ..., Up would not remain valid in case the
,
sum
of more than
integrals
for it
is
1 assigned poles.
-I-
from the argument here that when p > I an integral of the first kind,
M^", is capable, for given positions of the extreme limits, x, a, of the integration, of
assuming values within assigned nearness of any prescribed value whatever. Though not
It follows
167.
it is
capable of assuming
is
possible values.
all
+jnp(ap+TO,'(Bj'
this
+ ?n,'<Op',
To prove
all
oo
an enumerable aggregate ; the proof is identical with the proof that all rational fractions
form an enumerable aggregate and may then proceed to shew that all values of the form
Since then the
TOj(i)i + m2<2-Hni3<B3 form an enumerable aggregate, and so on, step by step.
aggregate of all conceivable complex values is not an enumerable aggregate, the statement
;
made
is justified.
The reader may consult Harkness and Morley, Theory of Functions, p. 280, Dini,
Theorie der Functionen eirier redlen Griisse (German edition by Luroth and Schepp),
pp. 27, 191, Cantor, Acta Math.
ii.
pp. 363
371, Cantor,
Crelle,
where
also will
be found a theorem
of Poincar^'s to the effect that no multiform analytical function exists whose values are not
enumerable.
168.
^''"-h
(A)
+u^'-'^=Ui,
fii(x),
the determinant of
not vanish
to be the
same
in all the
jo
"
*'
(i=l,2,...,p)
(i,
j)th element
paths of integration
is
a^ to Xi, ...,
fij
(a,)
does
ap to Xp, are
period loops.
When
Uj,
...,
a?i, ...,
k/*i(ai)+T^V'(ai)+
The argument
+k/^(ap) + fe^.'K) +
see
Ui,
CONSTRUCTION OF SOLUTION
240
wherein
is
tr is
derived from
/i,
(x)
by
tr= Vr,.U,+
where,
if
From
differentiation.
(i,
(r
Hence
also, so
of the place
in
Ui,
lies
suflScient,
is
p),
tijiai),
Vij
denotes a homo-
f/^
These
Up be of
...,
c^,
...,
= l,
j)th element
fir' (a:)
we obtain
these equations
+Vr,p
[168
series
Xr, yr
(xi) is
fij
to obtain
a;,,
...,
then
U/,
if
places
a;,',
. .
...,
,
...,
Xp',
such that
Xp,
<" +
<"''"'= U/,
(1
= 1,
2, ...,
p);
<"' +
Since
by
any
finite
the vahies
= l,2,...,p).
(i
for
U/,
all
able, step
+uf^=Ui+
+<'"'"'=f^..-,
we
are thus
equations
(i
= l,2
p),
0, 0,
. .
...,
.ri,
is
small as
we
please.
Hence there
<""+
In order then to obtain places
a;,,
< "'+
Ulm,
...,
Up/m
..., Zp,
+<''"=- Ui/m
...,
lying respectively in
such that
(i=l, 2
p).
+ <""''=
f;-^,
(i
l, 2, ...,_p).
169]
it is
''"+
and
such that
+ TOM^''"' = 0,
+ mu'l"'''+
+u'^''-'^
the places
. .
= l,
(i
2,
...,p);
has been shewn (Chap. VIII. 158), that these equations express only
mp + p places formed of ^,
Zp, each
times repeated and
it
O] ,
241
Op each
(m +
Now, when (m
places which
+ V)p
all
but
{m+
a,, ...,ap, it
at the places
Zp
whose
we may
l)p
places formed of
1 ) times repeated.
algebraically
(m4 \)p
coefficients
Ri,
+Rp = 0,
+ ZP-'iJ, +
...,
Rp
the places
<'"'
is
the set
x^,
been shewn
...,Xp obtained
is
itself in
to see
minant denoted by
vanishes,
is
is
may
37,
ff,, ...,
Up
clear in 164, in
corresponding peculiarity
It
article.
of 99 (Chap. VI.).
= l,2,...,p),
necessarily unique;
how a
{i
( 164),
u,^-'^=Ui,
what way
may
present
Ex.
ii.)
we may
take places
6i,
...,6^, for
which
u"'^'-tin which Vi
f7i
+ Mf '* +
-I-
+M^'''''=Fi,
w^'*".
We
do not enter into the difficulty arising as to the solution of the inversion problem expressed by the equations (A) in the case where Ui,..., Up
have such values that a^, ...,Xp are zeros of a (^-polynomial. This point
are to
is best cleared up by actual examination of the functions which
be obtained to express the solution of the problem (cf* 171, and
*
See also Clebsch and Gordan, Abel. Functnen., pp. 184, 186.
16
EXPRESSION OF SOLUTION
242
Props,
and
xiii.
xv.,
Cor.
iii.,
of Chap. X.).
But
[169
should be noticed that
it
...,
ap do not
is
number
make
the determinant
all
vanish
with
a;,,
...,
the peculiarity
...,
t+1
being the
as^,
We
154, 158.
consider
now how
Let P^' * denote any elementary integral of the third kiud, with
at the arbitrary fixed places
Then
^, y.
hand
in equation (A),
infinities
Xp denote the
it
function
^
is
"^
~-^^T
^-^i.y
...,
further that,
if /* "
a,
and
sum
jXp, Op
T.
third kind
is
adopted.
f,
of the form
+vr+'^'
Vi'''+
where
is
may depend on
f,
7 but
\U^ +
and
T is
of the form
+\pUp + \;
...,
Up.
For definiteness we may therefore suppose that P^' " denotes the integral
of the third kind obtained in Chap. IV.
Then,
are small,
firstly,
x^, ...,
...,
Op,
134).
and
tf,, ...,
Up
\ti[{ai,^)-{auy)]
I
The introduction
of the
function
is,
given by
is
1=1
*
when
( 45.
believe,
1.
p. 302.
due to Weierstrass.
The other
See
Crelle,
lii.
BY ABELIAN INTEGRALS.
170]
243
a,,
It is
expressible as a convergent
in
series
Ui,
...,
...,
Thus
Up.
Up when
U,
...,
also
is
Up are
suflSciently small.
a^i,
...,
<"''
+ul^-'^ =
wf"*'
+ wf""'=i7i,
where
-Uilm,
+ P^7) = l0g^g)-
+m{Pt,y +
P(,y
Z (x)
+Pi.y
of the places
of the places
a;,,
Xp,
...,
(m +
l)th
is
function.
gr^e-^^.-^d)
Z{yy
wherein Tq denotes the sum
^(.y
"^
'^
i.y
sufficiently great
it
is
...,
Up/m.
(7n
+ l)th
Z{^)/Z(<y)
For
is also
let
+\npZmpiie)+\
Z(x)=\iZ,(x)+
wherein Zi(x),
...,
arbitrary constants.
of the
fact
X, Xi,
..-,
\np are
vanishes to the
mp
of the places z
of integral
mth
..., z.
Hence (by
powers of UJm,
...,
168)
Upjm.
...,
X.,
...
Hence Z{^)/Z{y)
is
series
expressible
Up/m.
162
INTRODUCTION TO SOLUTION
244
Hence,
for
finite values of
any
of integral powers of
Up the function e^
C/j, ...,
It is also
f/p.
by series
method of proof adopted, that the series obtained
values
Ui, ..., Up are independent of the range of
J7i,
....
[170
for
is
expressible
the final values are reached from the initial set 0, 0, ...,0; so that the
The function e^
function e^ is a single valued function of E/",, ..., Up.
An
171.
will
if
0, 0,
. .
0.
U,
...,
Up,
be obtained in the next chapter ( 187, Prop. xiii.). We shew here that
be known, the solution of the inversion problem can
that expression
also
K=
it follows,
as here, that
e''^
II.,
ul'^'''
Then
14).
if
+ n^_'';^
is
U,
...,
Up, whose
of e^
is
(Chap. VIIL,
154),
TT^i'Ti
n^t'Tt
Z{Xi)
we obtain *
V{U.
...,
Up, ?7.)...
V{U
Up, ?
we have
X,
when the
values,
we can
values of
y>')-^X-Z{a,)]...[X-Z{ap)]
a well ascertained
said,
f/i, ...,
hence, substituting
for
in
'
turn any
p independent
quantities
Z(x,),...,Z(xp),
and
kind.
of the
been shewn
in 152,
...,
Up-,
f 7)
,
sum
TXa,
jXp, Op
p. 175.
is
obtained, that
172]
BY THETA FUNCTIONS.
The
172.
f,y
which
is
245
i.v
'
character.
consider this
function.
is to be regarded as of a preliminary
appear in the next chapter that it is convenient to
function as expressed in terms of another function, the theta
It will
It
is
manner, which
is
possible to build
in
an d priori
g) (m).
But
may be supposed
derived from
results
which are more easily evident a posteriori. For this reason we proceed now
immediately to the theta functions; formulae connecting these functions
with the algebraical integrals so
and XIV.
[173
CHAPTER
X.
General theory.
The
The
used by Abel
f.
case p = %
For the
were
also
who was
in
which the theta functions are considered in this chapter they were first
any value of p, by Riemann||. Functions which are quotients
of theta functions had been previously considered by Weierstrass, without
any mention of the theta series, for any hyperelliptic case IT. These functions
in
given, for
occur in the
*
memoir of Rosenhain,
Fundamenta Nova
(1829)
Ges. Werke
Bd.
loc. cit.
for
the case
p=
i.,
p. 14,
2.
i.
cendenten,
p. 153, etc.
The equation
t.
i.
(6) p. 225,
p.
343 (1827).
P()=4p()-ff,p(tt)-j3.
t Crelle, xxiv. (1847), p. 277.
II
Mem.
The paper
is
dated 1846.
lii.
(1856);
174]
247
SERIES.
the Riemaiin theta functions are not the naost general form possible.
is
The
174.
In the case ^ = 1, the convergence of the series obtained by Jacobi
depends upon the use of two periods 2a), 2aj', for the integral of the first
kind, such that the ratio jb'/o) has its imaginary part positive. Then the
q= e
quantity
Now
first
"
is,
it is
kind v^ ",
...,
'*'",
so that
if
we
i)^'"
1, 0, ..., t,,
i,
...,Tr,p,
is
+ Tr, r V+
= Ti,V+
all real
positive* for
2t,,
values of the
7iin2+
variables i,
...,
2T,,snrns+
Up.
Hence
for all
rational integer values of n^, ...,np, positive or negative, the quantity e'"*
has
modulus
its
and
being
Kr,,
quantities,
less
real,
ni=-Qo
7i2=-oo
...
+apnp+iJT^
e''i"i+
np=-oo
wherein each of the indices Wi, ..., Up takes every real integer value
independently of the other indices, is e"^, where
i=-(6i7ii
+ bpTip) +
= -(6,ni+
where i^
is large,
bpnp)
(khH,^
rip,
tiy
considered.
greater than 1
positive integer
e^''' is
now the
series
+ L/fi,
is
(l
and the
*
+ -:
series
The proof
is
i/r
has the
finite limit
is
f-"
will
1.
and
,r,.
.^'
<r,.,
"
has no
essentially positive
and
is
+2<ri,nii(,+
'
be convergent or not
convergent or not,
<
"
rii, ...,
fi
for
n,;
be taken
<
be any
^"^<(l+~j
convergent provided
If
n,, ...,np
if fi
be the periods of
<Tun,^+
its
will
is -v/r-" is
fonotion
has
),
and therefore
is
be positive
+-j
which
When
will
+ 2/Ci,2ninj +
say,
-<fr,
is
the values of
for all
-rr
K, and
EXPLANATION OF NOTATION.
248
>^p.
series
is
[174
i.,
Hence the
318.)
is
absolutely convergent.
In what follows we shall write 2Triur in place of a^ and speak otui, ...,Up
+Upnp,
the arguments; we shall denote hy un the quantity !!+
Then the Eiemann
and by tw^ the quadratic ThWi'' +
+ 2Ti2'rh''h +
theta function is defined by the equation
as
@ (u) = 2e=**"+i'^',
where the sign of summation indicates that each of the indices w,, ...,np
to take all positive and negative integral values (including zero),
independently of the others. By what has been proved it follows that @ (w)
is a single-valued, integral, analytical function of the arguments ttj, ..., Up.
is
The notation
is
a row, or column,
letter,
Mj, ...,
It
Mp as
convenient, with
is
i,
...,
Up,
and
u, similarly,
as representing
elements.
its
@ (u),
to consider
given by
@ (w
q, q'),
or
{u, q) = 2e^"<+4'i+i'"-("+9V+i?(n+9')
herein q denotes the set of ^ quantities q^, ....qp, and q' denotes the set
of p quantities g/,
qp', and, for instance, u{n + q) denotes the quantity
. .
un
+ uq, namely
+ t(pnp + iti9,'+
Uiih+
+ q'Y
and T (ji
denotes tw
+ 2rnq + rg'^
...
-|-2T,,2n,n2 +...)
'*i'2p'.
namely
(tii),+
+2 S 2
T,,sn,.g'/
+ (tu^/--!-
...
s=l r=l
The
quantities
g,,
...,
qp, g/,
whatever
in the
...,
@ (m
most common
qp'
constitute, in
their
aggregate,
the
The quantities t;,,- are the periods of the Eiemann normal integrals of the first kind at
the second set of peBJod loops. It is clear however that any symmetrical matrix, cr, which
is such that for real values of k^, .... k^ the quadratic form irk^ has its imaginary part
positive,
the
may be equally used instead of t, to form a convergent series of the same form as
And it is worth while to make this remark in order to point out that the
series.
Riemann theta
symmetrical matrix
greater values of
is
still
greater.
We
shall afterwards
FUNDAMENTAL
175]
The
function 6(2*)
theory of
One
is
given by
elliptic
=hi; =iO;
249
is
elliptic functions.
IDENTITIES.
=0.0.
There are some general properties of the theta functions, immewhich it is desirable to
175.
for
purposes of reference.
is
stated
it
we may denote
integers;
4/3, ^a,
where
either
or
1.
by l^^,
or shortly, by
most common case
Further we use the abbreviation H^.w. or sometimes only !!,,
/Si,
it
...,
^0p, ^a^,
..., ^cip,
+ Ti,,
+ ri^p'mp',
wii'-l-
(i
= l,
2, ...,p),
Wr,,
is
are
hi
a;)
by the ith row of the matrix t. When ttij, ...,mp are integers, the quantity
Mi + Tim is the period to be associated with the argument Mj.
@ (- "
i/3,
ia)
= e-^- ( iA
;
(A).
4).
Thus @(w;
4)8,
i^oi) is
The behaviour
of the function
wherein r
is
(m,, ...,Ur+l
K+
r,u^+T^,r....,Up
T,,
is
@ (w) when
+ fi;
4/3,
Up-,
+ Tp, ,;
1, 2,
4/3,
4/3,
...,p,
= e"'"-(w; 4A i^)(u
4a) = 6"=^ ('v+K,,)-^3.
4a)
4^,
4a).
in the equation
(m
^ /3,
4 o),
(B)
FUNDAMENTAL
250
herein the quantities
IDENTITIES.
m/
rrii,
[175
are integers,
M+fl^
stands for
2 m/ (t,,
and
Wj'
+ Tr,
rwj, ),
J,
is
of
(m
J/3, \ol),
coefficients, of the
in regard to m,,
...,
second
Equation (B)
is
each of
if
yS',
a denotes a
+ Jfip.
^ a)
\^,
..;
= e-.'(+JP+ir+JT.'i @ (
+ ^/3', ^a + ^a'),
^/3
(C)
in
If,
@ (u - ^ ftp,
i;8, ^ a)
0,
= e i+i3+i-)
(tt
0)
= e"" 1"-+^"'
(m)
{u
this is
i^, ia)
it
+ Jfi^, .),
(D)
when each
Finally,
of m,
integers,
we have
the
equation
(w i^ + m,
;
Ja +m')
= e""" (m
(m;
^/8, ^a),
(E)
^/S, ^o),
^a does not
We
This process
is
absolutely convergent.
(A)
If in the general
term
g2iru(+io)+i,rT(+iay+^(+i)
we change the
27riM
signs of Wj,
...,
p, the exponent
becomes
176]
r= 1,
2, ..., jD
then, since
may
write m, for
consists of integers,
ft
-n-a,
that
is
251
mr=
becomes
irij3
is
...,
2irt(ni+}a)+iirT(m+Ja)+iri/3(m+iii).
B
,
of the
mp
replacing Ui,
original series,
obvious.
The exponent
(B)
wherein m-^nn' stands for a row, or column, oip quantities of which the general one
m^+Tr, 1 m^ +
is
is
+Tr,pmp',
equal to
2iriu {n
+7ri {ma
Replacing
e*"'"'"
By
(C)
by
and writing n
for
(B;
is
obtained.
- m'^) + 27rM)m.
we obtain
+^a)-7ria' (u+^rd)
this
is
(C).
It is obvious that equations (D) are only particular cases of equation (C),
equation (E)
is
and the
immediately obvious.
It follows fi'om the equation (A) that the number of odd theta functions contained in
the formula 0{u; ^8, ^a) is 2^-1 (2'' 1), and therefore that the number of even functions
2p-i(2p+1).
is 22J'-2P-'(2P-1), or
i/u
>
^P) yp>
^^^
or
either
1,
sets consist,
(i),
number,
say,
/( p -
1),
(ii),
integer,
together with,
integer.
^1^1 +
in
which
+J7pyp= an odd
Xiyi+
These
is
for
= (0,
(1,0), (0,0),
1),
+^p-iyp-i = an even
integer,
(^p.yp) = (l,
Thus
Hence the
!)
'
1 ).
252
177.
We
Riemann
e^, ..., Cp
surface,
a function of
(m +
=V
+6,. where
equation (B)
yfc)
(m),
...,
surface.
Dp'" are
{it)
is
the conditions
it satisfies
{Ur
Vi
By
x.
m denotes a
Riemann normal
the
[177
wherein h denotes a row, or column, of integers A;i, ..., k^ and h' denotes
a row or column * of integers A;/,
k^.
As a function of x, the function
. .
(v^- "* + e)
always
cannot, clearly,
finite
the function
i;/
the arguments Vr
we proceed
+ e)
(i;^- "*
infinite, for
fundamental theorem
first
become
by
fir
w^.
Ui,
are
zeros of the
p.
When
...,
+ e^
has always
is
"*
the arguments
Ui, ....
(w)
Up
can
The sum
is
there-
by the
p. 21.)
Now
the values of
a period-loop
values of
dx
a,,
by
differ
the loop
The
^^
at
2in,
or
0,
-^
are equal,
are equal
values of
equal to
hr,
according as s
=r
of the integral
-j
namely equal to (
1)
The notation
of the integral
= 1.
is,
hr
or not.
therefore,
{if-
"
-(-
this is
then the
e).
i(]
+ Ti/c',
...,
u +Tpk'.
sum
178]
178.
to
253
We
make some
surface in such a
way
@ {if- + e)
"*
may be
Riemann
regarded as
Then
log
{u)
is
the surface
is
still
^-ply
Fig. 2.
surface
'iiri
for
is
Zr.
@ (m) when
is
the
again destroyed
Any
closed circuit
same
as at the beginning.
From
the formulae
0(m,, ...,,
0(Mi
+ T,,
...,Ur+Tr,r,
-1-1,
....
...,Up)
-|-
T^,
p)
= %{u),
= C"^ "'+*%,
(m),
r'
g-ari (u,+1t,
can, at most, have, for opposite points of a^, hr, respectively, differences of
The
figure, Ur
are integers.
wherein gr and
which these increments occur are marked in the
vl'
" -H Cr
254
the increment
Ur
+ Tr^r,
taken round
mean
marked
is
(u)
all
the
Ur+
thus
br,
now
is
the
integral ^
hr.
log
(m)
du,,
zero.
two
sides of a, will be
left side
bution
The
which
is
^Tr,T
boundary in the figure. Let Ar denote the point where the loops
and that round Zr, meet together. The contribution to the integral
single
a,.,
[178
Since log
will
is
equal to grTr,
arises
of a^, from
'"
br
br is
left side
log
@ (m) du,
equal to
+ er + ^Tr, r + K]
This contri-
to A^.
Ar back
dv",
of the curve
*",
br,
from Ar back to
Ar
this is equal to
("r
where/,,,
is
when
equal to 1
of the integral
Zr,
= s, and
is
is
br.
p boundary
""
+{er+ K)fr,
8,
otherwise zero.
arises
by the
for the
^Tr, r) dvl'
may
^+
'dC
If
we
When we add
v/' ^'.
we
.... js,
we obtain
x,m
r = l\_
6,
'
Jbr
dv,
'
'
left side
ks
h^i
( {v:'"'+irr,r)dv'r,
T=lJbr
which, beside the constants of the surface, depends only on the place
we have the result
h,
+ giTt^s +
---+9prp.,
+ es=-vl"
'
...-vl'^'^''
+ k,
(s=
1, 2, ...,p).
179]
m^
179.
255
congruences
+>
=h;
(s
l,2, ...,p);
it is
..., Wj,, satisfies the conthese places, beside the fixed constants of the surface, depend only
on the place
to.
e,
Thus
if
we express the
= -(Vs
->rv,
-V,
{%
-v;
'-
to
any place m,
to
-v^p-'"--irg')
this,
by the fundamental
-v/'^),
^onding
the form
Zp, fnt>\
Zi, nil
2, ...,p).
'-A;-T,5f'),
where
@y(v,z, n -V,
(s=l,
);
p places,
m^,
result
rphis factor
It
does not
is possible, corre-
..., nip,
only on the position of m,, such that the zeros of the function,
^(yX^m
regarded as a function of
x,
yZ,,m\
^Zp,7np\
This
z^, ...,Zp.
is
a very funda-
mental resultf
m\.A,
V,
Tjip,
+V,
. . .
Ap
^
=K, = 1
J
r=lJ
z,
(V^
y^"- ""
..., nip,
j z,a
+^Tr,r)dv,
br
a being an arbitrary place. If, instead of to, we take another place /x, we
be required to determine places /tj, ..., fiphy the equations
shall, similarly,
t;r^'-h
+v7-'''
= K, = 2
f (t;r
T=1J
* If
iT,.,)d^r",
{s=\,%...,p);
br
vm.,
158).
Biemann, Ge$. Werie (1876), p. 125, ( 22). The places mi, .... mp are used by Clebsch
In Riemann's arrangement the existence of the
n. Gordan {Abel. Functionm, 1866), p. 195.
introduced.
solution of the inversion problem is not proved before the theta functions are
t
Cf.
CASE
256
[179
thus
i I vT"'dvr, = i/s,rVr,
+C''^= r=lJ
'=1
br
C"' +
wherein
/, r
when r =
and
s,
rrij,
which
/J,,
(i,
-Vs
places
..., nip,
-Vs
X,
The
we see by recalling
Thus (Chap. VIII.,
...,
Vs
are congruent to
otherwise zero, as
is
(S=l,%...,p),
^i,
is
+^"^+^"=0,
v:""''+
will also, in future,
'p. f'P
Zi.i
(s=i,2,...,p),
TO)
(mi, ...,mp,
fi).
If the places m,, ...,mp are not zeros of a (^-polynomial, this relation
determines
Ex.
/t,,
..., fip
p = l,
In case
The deductions
180.
the function
not
at
^2, , S'Dd
so.
may be
function
(if- ^ +
so far
a ^i-tuple zero at
log
@ (* +
"*
and
are
"^
of course
all
of the
The
otherwise obvious.
Z]i,
^^i-tuple zero at 0,
so that ^i
-I-
+Pkp-
e)
its
zeros of
when
this
a pa-tuple zero
The
surface
3.
The
we remember
when x is taken completely round
The mode of dissection of Fig. 3,
of a pair of loops (a,, br).
also be used in the previous case when the zeros of (v"-
e)
@ (if- + c)
is
may
x=z, as
e)
dissected into
dissected surface
that log
for
is
/i.
increases
To
by
iiri
-l-
first order.
integral
^.
log0(/''"
-I-
e) d'u^''",
Its
180]
value
is,
+ gj dv'r-K[
dv':
3,
PiV,
-f
"
257
^PkVs
(^^-
e,
J^,
,)
dvl'
(;,
1)
v^ ^'
.4],
Jo,
first
row
Oi,
6i,
Ci,
and
iTp^p) dv.
is
^,,
A,
J b.
b,
...,
Zn,
The
so on.
suffix
a, to
is
the
first
integral sign
in
Fig. 3.
the
left
cut
6,,
and so
on.
If,
the suffix
a,,
we put
as before,
ks,
= ^
(Vr
is
to
b,
k^ for
the
sum
+hrr,r)dv,
r=\JbT
we
the quantity
obtain, therefore, as the result of the integration, that
'is
By
the
left side of
a cnt o^
+ 5'iTj,i+
,
or 6,
is
+9p'rB,p+es
meant the
side
(u)
in
marked in the figure. The general question of the effect of variation
has been
functions
tlieta
of
the
transformation
most conveniently postponed until the
are
ib
considered.
1^
[180
258
equal to
is
and
this is
We
z,
_ ^^2-
_/i.^p,_/2>^''i+'-
;fc^_/i'^i_
^'>.+'.
Zlt,
-v[
^D
save that
thus obtain, of course, the same equations as before ( 179),
And
times.
times, ..., and Zt is repeated p*
is here repeated p,
x,m _^z,,m,_
_ ^zp, J"?
From
181.
_/"'"'''
the same as
'factor,' is
inference that
/i,
- Tsg),
(^- --;"'" -
(jogg jjot
the results of
the argument
-^'"^);
179, 180,
we can draw an
inference which
theta functions.
leads to most important developments in the theory of the
For, from
what
is
there obtained
follows that if
it
fs=v;'
for s
1, 2, ...,p,
this
is
'
+v7
be any places
z^,
...,
all
is
Zp
for
the function
Zp-i-
Puttmg
-vs
^p
-v'""^) has
function
off,
...,
^i, ...,
all
values
Whenyj, ...,^
form in question.
Nevertheless the
case which presents itself reminds us that in the investigation of the zeros
of
@ {if- "* -h e)
and
If,
all
it is
we have assumed
for
arguments rj, ...,rp. We assume* from the original definition of the theta
function, by means of a series, that this is not the case.
Further the function
each of the quantities
The
series is
{if- "* + e)
Ci, ..., Cp
is
and
if
2irtr3,
e**"'',
182]
the argument, of
259
finite
are increased
Cj, ..., gp
periods.
first
of
all,
gj, ..., gp
168) positions of
lying
,, ..., Zp
to
congruences
e.
= <'-"'' +
the places
+ <'"'^,
@(^-
_v^.-'".
values,
. .
And
good.
therefore,
_ T,^-*' '^-')
...,;,);
given
(s= 1,2,
Zi, ..., Zp
far
by any simultaneous
to values of
....
e,,
for
For such
identically.
the investigation so
Cp,
such
the
values,
vanishes identically.
function
an analytical function of the places "^ z^, ..., Zp^i, and vanishes identically
for all positions of each of these places within a certain continuum of two
is
dimensions,
must vanish
it
(^.
"
-I-
vanishes identically.
e)
The important
the function
It
is
convenient
now
desirable to
(I.)
If
fj,
to
number them.
...,^p be
is,
linearly
+Xpnp(x) (Chap.
\n,ix)+
more
II. 18,
Chap. VI.
then
101),
the function
all
-/'''^)
positions of x.
For then,
if
-I-
an argument
z,
=x + iy,
cannot vanish
for all
identically.
172
SUMMARY OF RESULT.
260
Zi,
pr1
determine
..., Zj+i,
= (?...,
Chap. VI.
fp) (see
[182
93, etc.,
and
(zi,
...,
Zp)
Then the
argument
_^,,m,^
^,m_^f,.._
(s
= l,2,...,p),
when x
is
at
thus
theta
at least equal to
1,
the theta
It follows
if
z^',
Zp'
x nor
be among
Zi
z^',
...
%(/"'regarded as a function oi
From
Zp,
and neither
z-i,
y'""''
-v'"
"^),
...,
Zp.
this Proposition
(v^'
z^
or
(ii)
when
Zi, ...,
Zp are zeros
of
<f>-polynoinial.
(II.)
and
let
Take t
function.
1 arbitrary places
ti>
wherein
exists, of order, Q,
> tg>
'^ij
I ^T+jg>
q= Q p+r+l,
and suppose z,, ..., z^ to be a set of places corethe poles of the rational function, of which, therefore, q are
sidual with
arbitrary.
Then the
@(ij'^'"'
function
+/<"^ _/."._
+ /''^' +
_ ^'Q, "V-QX
vanishes identically.
if we choose ^,+1, ..,
to such that {^u -, ^q) = (z^, .... z^), the
general argument of the theta function under consideration is congruent
For
to the
argument
^m,,
m_
^i]
7,
^ir+i-,, mr+i-,
^f,+i, th^+s-,
a particular case of
_
that
^f<j,
mp_.
occurring in
lower
is
182]
(III.)
. .
r^ that
,
.
261
@ (r) =
and,
0,
vanish
for
all
such that
_ y^-i- ^P-\
which
(cf.
and
may be remarked
it
that
if m', m,',
m^'
179)
mp)={m,
//, ....
m/)
therefore, also,
^m-,
m _ ^m,; m, _
r=
u'"'"
_ ^,/,
/n,,
_q
is
w'"
'"
"*'
_y^p-'.mp-i
is
@ (v"'' +
_yip-i,m'p.i^
x,m
z,,in,
j,''-'+r=v'''"'-v'-'"'-
Sp.j>.
v
now we have
is
Zi, ...,
which
is
Zp
may
@ (if'
'
since
Zp z,
@ (r)
vanishes,
and obtain
If the places
Zi, ...,Zp-i
r=
^w;'. '"
'"'
v'''-
_y!'p-i.mv-i^
-'
v"~'
'
+ />-" "^-'^0,
A PARTICULAR FORM
262
(if- +
function
'
vanishes identically
is
z-[,
[182
an
''"
@ (v^" u)
we can
function vanishes
When
(IV.)
is
+/'-'''^-'
vanishes
when u
conversely,
@ (/)
it
to
= M,
such that this
is
such that
is,
made.
the function
Then the
entirely arbitrary.
of sets of
Zp-i;
r)
;
number
infinite
z^, ...,
and
vanishes,
to.
(* '
+ r)
does not,
for the values of z considered, vanish identically for all positions of x, the
zeros of
'
the argument
This
for
cc
and
(III.)
but
it
is
of
(V.)
of
r.
is
sufficient
z,
(r)
but
z,
the positions of
z,
^,
^ considered, then
it
is
all
positions
regard to
possible to
find
x,
places
such that
^,, ....
Zp^ are
definite.
wherein
^i
. ,
2;^,
+ v^-^+r=v''
{v^- f + r)
is
(ii*. ^
of the
+ v^' f + r) now
;
clearly a zero
is
and vanishes'
also a zero; for, putting ffor a;, the function
(c^.^+^y^f+r)
( '
becomes
r), which also vanishes identically.
Putting, therefore,
f, z
for Zp_, and Zp respectively, the result enunciated is
obtained, the
identically;
r),
and fis
uniqueness
of the places
We
all
may
z^, ...,
state
positions of
x and
z,
and
(iP. ^ + f. f + r)
If
(t;*.
^^
+ r)
vanish for
182]
263
can be solved, and in the solution one of z,, ..., Zj^, may
be taken arbitrarUy,
and the^ others are thereby determined. Hence also we can
find places
Zi, -, z'p-i, other than 2,
z^^, such that
+/''-'''-'
z'^^
being arbitrary.
vj'''""
Q-q = V-r-\,
z,',
...,
putting
^-polynomial vanishing va
arbitrary place
+r=
'
The
(VI.)
'"
. .
. ,
- /"
'"'
v^t^'' "''^'
- v''
*"',
propositions
(''*"^'
z^,
v""-
^,, ...,
Q=^-l, 5=1,
Thus, when
z.
= 0,
-t-v"^'^ -l-r)
(III.)
(cf.
Prop. 1).
be identically zero
If
the places
and, of these places, q are arbitrary, the others being thereby determined.
These arbitrary
places,
+ v^-''' + r)
&(v'-' +v^''"+
say,
v'-'
wherein f
It
is
. .
v'"'''+
+v'^''"'
+ r = v''
(*" "'-1-1;*"^+
Thus the
-I-
*" ^
all
-I-
just written,
,,'-^
and
z^,
it
...,
Zq,
replacing
-|-
- /"'"",
v*"
also putting
^'
z-^
. . .
for
same
x,, ...,
a;,
/..^
fi,
+ i,^.-'"' +
fi, ..., ?,
-f-
r).
+ r).
..., f,,
fp in
the congruence
j,^.m,^^f,...,_^
as
by
'''
r),
z respectively for
+i,^.*'
1)"^
-t-
x the function
becomes
+ /-'.+
this is the
0(/.^ + *."=.+
Putting then
1)^" ""
@ (/' ^
. ,
becomes
'"
we have the
result stated.
264
Hence
also,
we can
tind places
..., f'p-i,
gi',
[182
other than
such
that
/'^'
+/"-' ^"-' = 0,
q of the places f/, ..., f'p_i being arbitrary. Therefore a </>-polynomial can
be chosen to vanish in f
fp-i and in q{=p 1 {Q q), when Q = p1)
,
. .
y^'^
for
+ v'''-''+
wherein
^1,
+i,^-''^-
identically,
g+1
zeros of
r-,
independent
linearly
</)-polynomials.
(VII.)
all
x^,
If the function
@ (/'
...,
Xg,
v""
''
+v'''"'^+r) be not
for
z,
proper positions of
x,
and be not
z,
then
we can
find
^ _ ^mp,
and can
(v''' '
the function
Zg,
+ /" ^ + r)
^'
places
TO
_ yfp-i. ""p-i
_ yfi. Wi _
^p_i, of
...,
that
This
is
@(y''='
We
s w"^' y^"'
_fj'-.'j'-i_
+ y^"^' +
if
we
notice that
+^^-^-r)={v''-' + v"-''+
+v'^-''
+ r).
v'^-^'
0,
and
repeated.
(VIII.)
fi, ..., fp_i
?;^
(cf.
?i,
179) in the
. .
..., fp_,,
?p_0
= (m^^
if ?,,...,
f ...,fp_i)
form
m^,..., i/).
f^, be arbitrary
= (m,=, m2^
..., mp).
?)""
_ t,^"
'"'
_yip-i,'>np-i
places, places
182]
m,, m^,
?i,
certainly vanishes.
It
+ v'"'' + r)
the
We assume*
function (v"' + v""
all
positions of x,
'
we can
x and
..., fj,_i,
is
may
positions of
finite
Proposition VII.
it
follows
such that
D^''"'' "''"'
this
@ (if' + r)
r)
It
z.
Then by
Xg, Zg.
...,
+v""^ +
''
z, x^, z^,
265
X, z, Xi, Zi.
that
may happen
all positions of
nip.
Then, by theorem
?j>-i
@ (r)
...,
we
r,
can, as
Hence
(IX.)
(mS
therefore the set
f,', ...,
we can
p1
...,
we can
^p-i,
?V-i.
f/, ...,
fi'.
fp_ f
^i, ...,
Now,
fV-i-
a very important
follows
-. ^'p~i)
{m,\ m,^
....
such that
m/);
....f'p-,,
Thus the
Oj,
places
..., |^p_i, fi
^i,
...,
fi',
f'p_i
arbitrarily,
Therefore,
^,', ...,
...,5'p_iis coresidual
of a set of 2p
corollary.
find places
is,
first
kind, the
(m=, Oi ,
in other words, if
Ci,
. .
Cp
. .
a^p-i)
= (wa', m^^
. mp%
(G)
2[^"'^
+v^''"]
2v'^''"-
v"'-''''
+ v"/-''+
+v^'-.''p
+ ^>.<'._
for s
by
where s =
* It
will
identically,
+9p-rs,p),
(G'),
1, 2, ...,p.
then aU
if
e{v<'-' + v^''''
+ 1 )th
u = r.
(),
+^'*' + r) vanishes
up to and
GEOMETRICAL INTERPRETATION
266
sets of places
'2:'p
-wii,
[182
For in equations
2^ values
hp
hi, ...,
and
and
g^,
possible
gp, h^,
...,
in
in
...,
1,
gr^,
g^
by an even integer,
...,p, lead to the same
hi,
711
or
either
of^Tj, ...,
is
ffp, fh, ...,hp
for the
i=\,
2,
(Chap. VIII.
..., nip.
158.)
We have seen ( 179) that the places mj, ..., nip depend only on the place
and on the mode of dissection of the Riemann surface. We are to see,
what follows, that the 2'p solutions of the equation (G) are to be associated,
an unique way, each with one of the 2'p essentially distinct theta functions
183.
Take a non-
interpreted geometrically.
where
^frn
general
. .
have
is
fi,
92, Ex.
The polynomial
ajp_2 as zeros,
we can form
(m=, a,,
...,
places nh,
..., fc^
....
m^
90)
will
If
<^'
its
yjr
n/ji
+ nB +
(^-polynomial.
those prescribed
a,
1)
(T
A.
of
will
it
of grade (
<fr,
ajp_j)
(Xi/r,-!- A<^)/A^',
= (*;,,
96)
repeated,
we have
k^),
(G),
it
is
consist of
(G")
possible to
zeros
of the polynomial
p-\
A and
A*
. .
modification
may
be
made
in
what
follows.
1^*]
{n
OF THESE PLACES.
l)a+n-3+fi
we form
267
(/(-polynomials
is immediately obvious algebraically.
the equation giving the values of x at the zeros of the general
(^-polynomial,
^101
+\i>pi
thep-1
^'"'""-h
where
flp_
stands for
We may give an
importance.
It will
be the
...,
of
rrip
= in^,.,
(H)
quantities such as
^.
!, , Op, y3 .... fip
-hz;'"''''""'
rrip
vi/,
+ ^l^s, 1+
OpT^.p,
-I-
being integers.
is
C+{:'^,y)i+ {x,
representing a plane quartic curve
(/j
is
of great
of the form
y\ + {x,y)3-{- {x,
= 3). Then
if
y)i
= 0,
A, B,
XI ' (X)nics
it will intersect it
can be drawn
again in 2 points A, B.
m itself.
In fact there are (Salmon, Higher Plane Curves, Dublin, 1879, p. 213) 28, =2^-1(2^-1),
double tangents
=^~^{7P+\), proper
36,
and
mj,
TOjj.
all
on the point
there
are therefore
in
"h-
the numbers of odd and even theta functions of half integer characteristics ( 176).
184.
m,,
...,
(X.)
mp
We
are one
equivalence (G).
We are now
same
2^-1
Let wi,', ...,m/ be any set, other than lUi, ....rrip, which
congruence (G). Then, by equations (G'), we have
2(v"''-'"'
\ g
/'"""')
-tft
0,
(s=l,
satisfies
the
2, ...,p),
268
and
[184
given by
where
tii,
,
Vs
y8.
+ aiT,, +
(3p
+apr,,p,
+Vs
we have
=iiip,.,
(s=l,
2, ...,p),
is
184]
269
(A), ( 175),
we have
(|.flp_ ,
odd,
of u, and
zero; since then 0(^11^,.) vanishes, there exist, by
is
...,
rii,
or
2(?;""'"'
+,,"p-'.'-
+
158)
(m\
so that,
by equation
+ ^'.'p)^
=0.^.a, =0.
we have
!=,
. .
= (mi=,
ri'p.,)
. .
mp"),
When
it is
to be expected that
case,
but exception
may
it
will
arise
{^ilfi, . + w)
when
is
= 0.
the fundamental
is
Riemann
generally the
surface
is
of
special character.
case, which,
may be
values of the argument, and special cases in which one or more even theta
functions do vanish for zero values of the argument.
Suppose then,
firstly,
of the argument.
zeros
(m^
nt'% ...,
w'p_i
\_,) = (m,^
m^-, ...,
nip'')
is
a half-period,
= - ili^-,
a',
say.
So
(^n^-,.')
therefore,
is
zero
all
the zeros
it is
z,, ...
Zp_,,
and therefore
(0)
+ v"'-''"'-')
would vanish
would vanish.
for all
270
185.
If,
for
it
</)-polynomial with
will still
periods,
[185
but there
will also
(2*'
1)
odd
there
half-
half-periods.
For
@{u +
to
2*1,
if Oi,
...,
ap,
...,
/3i,
m = 0, the
to Prop. XIV.,
= 0. By
it
(m
+ ^flp, .),
an argument which,
x and
Therefore,
z.
-"'""-
?;'"''"
in regard
is
postponed
@ {>/'' + ^il^^ a)
-^flp,.
even, and
convenience,
for
/8a is
by Prop.
vanishes
V., there is at
the equations
_^^P-.,P-..
hence there
..., z^p_,)
is
integers ^j,
places Wi,
then
it
is,
of
col be the
may
...
7!p_i to
Op, it
is possible to
number of
sets
obtainable.
As an example
surface
and
first
l),
(ii),
suppose
y2 = (x-ai)
= 28 = 2p "
(2"
-1
(x-a^);
^-polynomials, each of the form {x -a^)(a;~
We
a,),
xiii. p. 42.
185]
but, since, if e be
follows that
(c, c)
271
= (e,
^,
be replaced by
function
is
9 (u)
e,
and
vanishes
when u
">
1;^'
is
replaced by iQ^
function
e {li^-'-^Qp^ J
function
eiv^-'+ir^-'^-^Q^J vanished
By
and therefore e
when x
is
- JQ
at c ; since c
all positions
by Prop.
_^),
which
arbitrary the
is
of
s.
If the
VI., be possible,
in the equation
to choose both
Zj
and
As this
e{u + iQ^J, and
z^ arbitrarily.
coefficients,
Hence ^0^
^ is
is
its
first,
it follows,
but not
its
an even half-period.
by Prop. XIV.
second difierential
(See the tables for
is
p = 3,
being
y2=(a;-o,)
{x-a^p^i)
(i)
/2 + 2\
_,
(a;
- Op _ ,).
is zero,
As
there
no arbitrary
is
the equation
where n^,
...,
is
consequently
odd.
the
(ii)
polynomials such as {x a^
arbitrary.
the
(iii)
q = 2, Pa = l (mod. 2)
^a=0
(mod.
on.
And,
(x-aj,_^{x-cf(x-ef,
(or \iti\ch
finally,
{x-Cp-if, in which
all
of
c,, ...,
are arbitrary
On
in this
\
case q = ^-a~^ ^a=*-g (mod.
l?-t- 1
2).
*' "
{T->iT->
{T-MX*->
Now
in fact,
when ^= 1 (mod.
4)
Cp-i
2
is
2).
polynomials such as {x a^
and so
as^= 1
<'n
or 3 (mod.
4).
272
is
equal to
i/^22P +
while,
is
[185
2^2P+2coa^i^7r)
4)
Thus
all
or 22i-1-2p- or 2p-i(2p-1),
jr
And
for.
).
This
is
the
(2^
-1
).
there are
zero.
number
It is easy to
This number
is
equal to
which
number
is
when p
even.
is
^
P
the argument.
made
conveniently be
here.
TOi, ..., ip in
183,
we may
may
m taken at
polynomial ^z^-ajp+j, of grade 1; its remaining n^-2, =0, zeros, give no conditions for
the polynomial
of grade (?i-l)<r4-9i 3 + /i, =(2- 1) jd + 2-3-H, =^. Since tr+1, the
dimension of y, is /)-(- 1, the only possible form for yjf is that of an integral polynomial
i//>
in
of order p.
Whenp = 3,
This
is
c is arbitrary.
is
186.
Another matter* which connects the present theory with a subject afterwards
may be referred to here. Let ^Q be a half-period such that
the congruence
can be satisfied by oo' coresidual sets of places z-i, ..., 2p_j (as in Proposition VI.).
we have
{m\ z^, ..., 2\-x)=(toi^ ..., mp2),
so that (Prop. IX.)
z,, ...,
this polynomial
^C = t>'"''''"-J)^''-'*
Cf.
xiii. p.
35
set,
a 0-poIynomial
Then
denote
-/P-I.!-!^
Noether, Math. Annal. xvii. 263.
186]
273
then we have
so that
by
1^.
The
rational functions
-^/(f,,
2,',
^'l>\r
..., s'^.j.
we have
^'=4><t>',
and
4>'/<t>
= W<t>f,
and thus the function \/<^'/<^ may be regarded as a rational function if a proper sign
be always attached. The function has ^i, ..., 2p_i for poles and z^', ..., 2'p_, for zeros.
Conversely any rational function having 2j, ..., 2p_i for poles can be written in this form.
For if 2j", ..,, 2"p_i be the zeros of such a function, we have
v''"' ''
and
therefore,
by the
first
equation of this
+j;^'!'-> ^"-'
also
^Q = yi''v.m-l/>",m,_
_^-i,mp-,
^Q = j,"p.
= 0,
_^<P-',mp-i^
'n_i;fi."'i_
we can put
where (}>i, ..., (^, are particular polynomials such as (f>' or *, andX,Xj, ..., X, are constants.
In other words, corresponding to the oc ' sets of solutions of the original equation of this
, we have an equation of the form
\/* = X V^-t-Xj
-fX,
\/0"i-|-
V^,
wherein proper signs are to be attached to the ratios of any two of the square roots, and
any two of the q + 1 polynomials cj>, cf)^, ..., (^,, are such that their jiroduct is the square of
a (/)-polynomial. There are therefore ^giq + i) linearly independent quadratic relation.s
connecting the (|)-polynomials.
For example
(Cf.
which
112.)
= 3, the vanishing of an
* = {.r- c)^ such that
even theta
V*=-cvT-)-V^, =
where
(^^cjis,
Ex.
i.
arguments
-cV'<^i-t-V'(^3.
=(xf, =^/.
if
Ex. ii. Prove, for ^=4, that if two even theta functions vanish for zero values of the
even theta
arguments the sur&ce is necessarily hyperelliptic so that, then, eight other
thus
conditions
of
number,
The
2,
arguments.
functions also vanish for zero values of the
;
that it be hyperelliptic, is
necessary for the fundamental constants of the surface, in order
constants in the general
the same as the diflference, 9-7, between the number, Hp-Z, of
general hyperelliptic
the
in
constants
of
surface of deficiency 4, and the number, 2^0-1,
surface of deficiency
B.
4.
^^
274
that %{v'''
the
@ (r) = 0,
and such
positions of x and z,
If r denote
(XII.)
187.
[187
+ r)
Riemann
% (v"- + r)
n:::=iog
(t)^>+r- )"
I
^(^^+ r)l
@{v'-^+ rj
(if-'
single-valued on the
it is
(a.)
n"^'
+ r) (if' ^ + r)
Riemann
surface dissected
by the a and
period loops
it
(7)
it
a loop ai
hi it
is
it
+ r),
At
unaffected by a circuit of any one of the period loops.
loop
at
a
factor
unity
has clearly (Equation B, 175) the
;
-2i;"-^
which
is
also unity.
is
(''
27ri
j-j
i)
surface
(8)
it
is
independent of x
at
z,
namely
where
is
z^,
...,
to the value
by taking
"*>
"*p-'
nf''-^'
identically.
1.
(v''''
function
+ r)
@ (r) vanishes.
Hence we have
n::;=iog
Another particular
r
= ^njc,ie;
that kk'
is
fc.
^'
case,
0(/' _!,"'>_
_^Z!'-i.'ftp->_^.n*i>y
e(v^--'y^"'"'-
_^-.'n^->_^ft'i.)
of great importance,
denoting respectively
integers
made
is
obtained
by taking
kp, such
188]
275
and we hkve
to or less than
and
@(2,x,m_^m, _
_^KTOt^^)/ @(^,m_^ra, _
-i/^i-.m^^
,.-)
sets
= - (1;^'+" ""'+' +
..., 0*,
+i,i'p'.)^
/9,,
A,
yk+u
This function
is
become infinite or
e^, where
L, =-2iri(v''-^'+
Riemann
'"'
is zero.
is
at
z.
7p
...,
is
such
x.
6; it
-?;"* ""t)
'"'
-'oPk.'>"t),
it
has when
Therefore
m _ Dl> i _
(jf,
@(yZ,m_y^m, _
the places yt+i,
,yp being
i;(if 't
_^i.,m4
"li+i
yjt+u '"h+i
tji't+i.
tjrp. "Ip)"
yyp.mj,-^
and
ft,
...,
/Sjt,
> yp
not
zeros
of a
are
,
74+1,
7p
^-polynomial.
cf.
Prop.
XV.
Cor. iiL
(,,, 2
2.'''" ^'
4-
+?)*"*'
+ r)
182
X and
fi,
[188
INVESTIGATION
276
z,
x?
and
?p-i,
sets of places
_ ymp, m _ ^f,, m, _
_ ^&-i,m)i-i^
and
Now,
if
(^'^ +
with Zq, we obtain
the equation
in
I @/ (^^ +
^'
if"-"
i^>^
+ r-) = 0, we make
Xq
+ r) Xli (z,) = 0,
^'-'
1=1
wherein @/(m)
and
put
is
for
^{u),
D,i{x) for
ImD^Vi
quantities n^izg),
ever, the
",
a being arbitrary;
..., aj^-i,
Since,
Zq-i.
how-
..., 0.q{zq)
the
form
*'''
of
+vx:,.i,z,-i^^_
@ (m)
form
d'^i-
^1+
-\-ifi--^, H--'
@ (if)
and
@ (u)
vanish identically
when u =
r.
is
of the
and second
differ-
up
to
when u = r.
proceed
coefficients
then
order vanish
We
+r
when u
vanish identically
It follows
up
now
to
(y^''^'-|-
to
(u) and
Xi, Zi,
'<
a;,,
5'
positions of
x and
z.
(e +if'')@(e -If'')
firstly, it is
rational in
x and z
for,
considered as a function of
+ t.^'"-c)-T,V^,,/
_2,ri(t.^'"+/+i.^''-/)-xiT,,
x, it
has
188]
whose value
unity
is
considered as a function of
symmetrical in
has
2(p-l)
the expression
z, for
is
is
immediately seen to be
a;
zeros,
are independent of
poles,
277
Similarly as a function of z
z.
independent of
a;
it
has 2 (p
1) zeros
and
F(x)F(z), where F{x) denotes the definite rational function having the
proper zeros and poles, multiplied by a suitable constant factor, and F{z) is
the same rational function of z.
Putting, then, x to coincide with
z,
root,
we
infer
ii'(e)ni(x)
F{x)= t=i
/(/) n,(^)
t=i
where
= 2Tn DxVj'
D,i (x)
for
a arbitrary,
thence
is
we have
(r) = 0, 0/ (r) = 0.
pendently of X and
z.
positions of
We
x and
Therefore also
independently of x and z
all
Then the
and hence
r, for which
hand becomes infinitesimal, indethe left hand becomes infinitesimal
(v'- '+r)
z.
The theorem
function
@ (rf"''' +
+t;.-i. -i
from what we
identically,
As
and so
?;="'>'
^>
r) vanished
on.
that {if''-\-f)
e
^'
(/) = 0,
but
all
of
/ (/)
+ !;*.-.
^-"-'
e)
to
x and
z.
so
Let
all
coefficients of
^^-i, but such that the differential
'y^" '+... + !/''"- ''^"'-'+ e;
order do not vanish identically for it =
+ if"" '"' + e) does not vanish identically.
{v'^' '' +
so that the function
positions of
a;,,
() of the
first
n^ (!;/ t-i-/)
(!;> -^^-/)n'(i;''-'^+/)0(^>^'"'^-/)
nn {v='x''^+f)@(v''>^'''^-f)
'
278
[188
wherein
1, 2,
product
Firstly, this
be rational in
be entirely
will
it
x-^,
hr
As a
rational.
function of
Secondly,
m places
..., ,
z-^,
and
is
x-^,
the
(^,
as to
the zeros
also
z,
4-
e,
. . .
x^,
the zeros
...,
+ e) @
^r, Jm
2m
2(m 1)(2)
The
any one of
places depending
coincides with
x-^
otherwise zero at
have
function of
rational
when
(p
1)
1) are
factors 11'
zeros
also
then 2(^
(t/'*- ^i
(t;*>
2i
. . .
tF^^ '
- e)
poles of the
+/)
(j/^*
**/)
The
factors
On
of the
1)
the
/i=l
as a
the product
considering
is
x-^,
denominator
the
places
it is
4=2
and
2m
is
Xm, Zm-
Zi, ...,
a;,,
2(p m)
a;i,
2(^ m)
for zeros,
zeros depending on
e,
..., a;,,
2{p
1) poles
depending on/;
to
is
easy
is
depends on
of Xi,
it
The
wherein
The
now
z^, ...,
depends on
expression
ajj,
is
where
...
a;,,
zeros
and
poles.
regarded as a function of
same way
''
+ ...
<, is
is
z^,
regarded as a function
as,
e,f.
...,
It
z^-
...,
Xm)F{zy,
z^, ...,
is
its
Zm respectively
then
we
z^,
approach
obtain
+if'
1
^ + e) = 2;
.e, 2
{if'-"^
+ ...
+v=^'^-u ^-i
+ e) n,- (z^),
g:^- .-i
_.^,
j=i
\j i^' ''' +
^^'-''
'"-'
+ e)
n,-
(z^_o,
188]
where
so on,
and eventually,
tit
... t
279
z^
and
Similarly
;^;nl,/(y)nK..)
where
h,
Therefore, dividing
(-)'"
which is common
we have
by a
n'@2
to
1, 2, ...,
factor
(w^'.. ^*
+/) @2
(i;^/.
>''-/)
^1
m.
9H.
root,
RESULTING EXPRESSION
280
The remarkable
(XV.)
it is
investigation of Prop.
XIV.
is
due to Riemann
m 1, we
Using q instead of
are satisfied
by
function,
which
x*
which has
"
the equations
_ j)&-i, mpi
fi, ...,
^f
..., fp_i,
2(p
equations
^fii-i. "v-i
"!
1 .
if
Wi
yfi
sets of places
e = if^
are satisfied by
= D> "*
1) places ^,
^p_,,
f= D*' "
[188
i)'i
'"i
ifp-it mp-i
being supposed such that these equations have one and only one set of
solutions,
and has
(ii)
2{p-l-q)
places ^,+1,
a;,,
...,
= v""^'
"*
y"'"
'
>
"*'
j;&-i. p-i
'"1
_ ^p-i, mp-i
in the
"^i,
.,
form
a;,,
x, e)
-=-
(a;,
/),
The
Corollary i.
If we take
one ^-polynomial vanishes in
g
j}>np,
p-1
all
places
of them,
m _ |;fi, ), _
and define
by the equations
_ yip-i, mp-1
there will be no other set &, ..., fp_i, satisfying these equations, or 5=0.
If ?i. , ^p-i be the remaining zeros of the </)-polynomial which vanishes in
?i.
?p-i
we have
(Prop. IX.)
and therefore
Q = ^Wlp.
"
Similarly
^-polynomial,
.yf I
Wli
^fp-1 mp-i
1
we can put
f= vyWp, ?
/yt,
Ml,
9J^P-1
Wtp-1
OF RATIONAL FUNCTIONS
188]
Then the
fi
where
(a;),
. .
coefficients of
Hence
t^,
given by
fi, ...,
281
<!>(, e)
-=-
and
Thus the
/).
<I> (a;,
given by
<j>p(a;)
first
kind.
Jij,
...,
nj,_i
we can
2@/(ifi)<^i(a;),
Jfl being an odd half-period.
If another
(^-polynomial
Ui', ...,
n'p_i,
?ii,
..., 7ip^i,
and therefore (Prop. VI.) the function @{ir''' + ^il) would vanish
cally; in that case (Prop. XIV. p. 276) the coefficients @i' (^D.) would
We
the
7>i'
. .
V^' '",...,
if
Vp'" be any
kind, expressible
first
"
Vp
""
by
in
terms
linearly
of
the
+ \,pK"''
= Xi,,U^-^
second order at
where
i^. (),
V'-'",
and
'",...,
of
integrals
Wj, ...,
Ui,
+Xi,pUp,
(i
...,
= l,2,
...,^),
Up given by
(i=l,
2, ...,p),
coiTCsponding to F*'
....-^pix) are the (^-polynomials
a set of simultaneous half-periods of the integi-als
^n
Vp
independent integrals
Riemann normal
vt"'=\.Vf"'+
Ff'
identi-
vanish.
kind;
first
there
such that
"*.
denotes
If
in
stand for
i(A,
+ ^VT,M+
+ kp'-'i,p)>
(i
l, 2,
is
...,p),
OF SPECIAL KIND.
282
and
&),_ g, w'r, .
[188
= 2X;,
Wj, s
= 2Xi,
eu'i, s
+ 2\j,
Wj,
+2X^,p<B;,,,
+ 2\i, p
(i,s=l,2,
...,p),
Ti, e
where, in the
then
^n
first
2\i_
equation,
Alto;,
1+
2 (b's, s
we
are to take
+kj,coi^p
according as
or
co'p^ g,
+ kp'co'i^p,
+ kiii)'i^i+
(i
may be
double tangent
=s
or i=^s,
is
l, 2, ...,p).
is
that whose
and
its
equation
may
wx (in)
Corollary
= 0.
If the equations
ii.
= ^Wp, m
'^i
m,
^fa
l2
-yfp-li lj>-l
...,
can be
^p-i,
and
satisfied,
otherwise vanishing in
(a;)
'i,j(e)n^(a;0
! to
given by
= O.
JKr.
In the case of a plane quintic curve having two double points, this gives iis the
equation of the straight lines joining these double points to an arbitrary point Xi, of the
curve.
Corollary
iii.
We
of which the multiplicity {q) is greater than the excess of the order of the
function over the deficiency of the surface, say, q =
+ r + \, can be
Qp
^Q for zeros,
and
fi, ...,
If
Jq for poles,
l/(e)ni(a;)-l0/(/)n,(^).
where (cf
93,
Chap. VI.)
y"^''"
'"_/>'
'""_..,
esu"^'
es
v^""''
'"
^^ ^-"'P' _ ^h' "h _
_^^fi-T>'"-T_j,fi. '"B-T+l
_^{ii>P-i
_^iq, mp-i_
189]
189.
slightly
^ (m
it
is
283
convenient to introduce a
so far considered
we denote by
wherein the summation extends to all positive and negative integer values of
the p integers ii, ..., rip, a is any symmetrical matrix whatever of ^ rows and
columns, h is any matrix whatever of p rows and columns, in general not
symmetrical, b
is
^(u;q,
We make
q')
= e^-'(U;
q,
^).
Appendix ii.).
elements,
and
h
denote
any matrix
letter
of
column)
row
(or
If u denote a
p
shall
is
row
letter
we
generally
write
then
hu
a
of p rows and columns,
=
matrix
obtained
from
hvu, where h is the
huv for hu.v; and we have huv
h by transposition of rows and columns. Further if k be any matrix of ^ rows
and columns, hu.kv= hkvu = khuv. For the present every matrix denoted by
a single letter is a square matrix of p rows and columns.
consistent use of the notation of matrices (see
Now
let w,
elements Pi,
row
letter,
ft)',
...,
r),
i{
Pp, P/,
so that
fip
in a similar
way we
+ Xlp
will
ft)',
7),
n', is
is
flp,
,,
= 2i7P + 27;'P';
u.
the following,
^(w + fip
which
be denoted by
= 2a)P + 2ft)'P'
The equation we
6>,
will
then M
2,
g)
= effi-(+ini-)-"^^+2'"(^9'--P''e-2-P<''^(M,P+9),
(L),
for
0(m).
*
The
Leipzig 1880.
AbeUchen Functionen von dr.,' FanM,
Schottky, AbrUs einer Theone der
(xvn.), p. 115.
AnnaU
Math.
Cayley.
by
suggested
introduction of the matrix notation is
284;
[189
In order that this equation may hold it is suflBcient that the terms on the
two sides of the equation, which contain the same values of the summation
this will be so if
letters n^, ...,np, should be equal
;
first
we
require
= 6.
h(u + D,p)+lq +inq = hu+l{^-\-P') + m,{P-q),
6
and
a (
np)=
190.
it will
be convenient at
first to
how
neglect the fact that a and 6 are symmetrical matrices, in order to see
far it is necessary.
The second
= ^nri, hw =
h, b.
^b,
whereby
<o,
as'
= vi
2a} = l,2co' = r,
In particular when h
The
2aunp +
that
loops.
aD,^p
is
(2aflp
-Hp- 2hP') u + (afip - ^Hp) Q,p - iriPP' - bP"+ 2 (h^p - -n-iP - iF) q' = 0;
may be
we must have,
Hp=2anp-2hP',
and
(aOp
from the
first
of these,
{iriP
6P') P'
190]
285
to
equations
Hj,= 2anp-2hP',
leading to
= 2aaj, V = 2aGj'-2A,
J7
and
17
tj'
h.
These
equations, with
hnj.
= -7riP + bP',
or
ha>
are
= ^Tri,
hco'
= ^b,
all
are satisfied
When
they
we have
^ (m +
flp, ^)
(L),
where
\p (u)
Ex.
where
^Hpiu+^np)- iriPP'.
is
is
given by
a certain constant.
&>,
m',
rj, 1/'
there
must therefore be
to',
The equations
For
\ (,HpQ.^
Fpfic - ZT^np =
this relation includes all the
{rtP
+ Tj'F) (toQ +
w'Q')
27ri
{Pq - P'Q)
for it gives
k-^^
(PQ^
- P'Q)'
286
CASE
IS
ODD OR EVEN.
fjco)
PQ +
(cor,'
[190
v'to)
= \in{PQ'-FQ),
and expressing that
P, Q, P',
Q',
we
(2S'P
Ex.
new
Kiemann equations
by
+ 2v'Q) {2wF + ^PfjQ) - (2wP + 2rjQ) (2'P' + 2^'Q')= ^wi (PQ' - P'Q).
If
i.
we
in the
^ function linear
functions of any
Ex.
v,
Prove that
ii.
Xp (m + n) + X (m) -
provided
M+P^N+Q.
The equation (L) is simplified when P, P' both consist of integers. For
M, M' be rows of integers, it is easy (putting a new summation letter,
m, for n + M', in the exponent of the general term of ^ (m q+ M,c[ M'),)
if
-\-
to verify that
^ (t
Therefore,
+ i/,
q, q').
if
^ (m + fim
and
5'
5)
g)_
in particular
^ (m) is written
for
^ (m
0, 0).
The reader
will
= 0, = 0, = 2in,
\p (m) = - 27riP' (w + JP + \tF)
i?
t)
- iriPF.
One
^(m; q
where
M stands
191.
...,
we have
q),
Mp, Mi,
...,
Mp.
^ (u), we may
When
for
+ M) = e^^i'%{u;
in,
^ (u)
^ (- M +
n,
q)
e> (-)+2'ri(79--m'<,)
^ (_ u,
q)
192]
hence when
S- (u,
q) is odd, or
is
= \_ (m), we
even, since X, (- m)
therefore,
287
^^
have
(L),
^ (m + D,m,
= ^ (m - flm, 9) e^i-in)+4'n(9'-'9),
= ^ (u, q) e^-mM+^2"!" in) + "<"9'
q),
'?)
while
also,
^(w + fi^,
Thus the
q)
= 'ii{u,
q)e>^m+^i^-m'q}_
expression
Xam (m
must be an
m'g' mm')
This
is
be integral for
all
Hence the necessary and sufficient condition is that q and q' consist
integers.
In that case we prove as before that S^ {u, q) is odd
according as
4!qq' is
an odd or even
integer.
192.
which b =
we
linearly
2mVi
so that
is
as in 139,
is
And
From
u'-
denote
,...,Up
i{ Ui
first
138,
=/ii,iMi
+hi^pUp
{1
= 1,2,
...,p),
and
^ (u"' *,
where u =
iTTT,
any p
of halfor even
q)
q),
"-
the formula
m.'
we
infer,
that
fi(M
+ n)-fi(M) = 2(7;i,x7ni-l-
+Vi,p'mp
v'i.i'm,'+
+ 2w,, r,
...,Up
+'/'i,pV);
288
Thus
[192
be the argument
if Us
,f-"'_,t='-,_,
X,
where
kind,
Ui'
..., Wp'
are any
linearly
in the definition of
^ (m)
first
be the same as
that previously used (Chap. VII. 138) in the definition of the integral
Li'", so that the matrices
i;,
will
tj'
it
fi()+ir.
regarded as a function of
And
193.
in fact,
x, are zero.
is
equation
= i^' " + S
Vr^
[{Xr,
X)
- (Xr,
tJr
a)]
dt
r=\
wherein
Vr, i
x^, ...,
Vr, i
(Chap. VII.
For we have just remarked that the two sides of this equation regarded as
of x have the same periods
the left-hand side is only infinite
functions
at the places
Ci,
x^, ...,
a?],
...,
a^
if
used in forming
Cp
...,
Xp,
it
in
it
i/
(Chap. VII.
138),
we
replace
c,,
...,
Cp
by
vi,ri-xt
vp^ii-xp
-2(aj,iWi
+ai^pUp
),
and becomes infinite only at the places ar, ...,Xp. Hence the difference
of the two sides of the equation is a rational function with only
p poles,
,
a;,,
...,Xp,
III. 37,
zero.
194.
It will be seen
in
the
hyperelliptic
functions only.
* See Clcbsch
Thomae, CrelU,
Thomae,
lines, see
^^*]
289
( 196).
of 193.
The
+inp,
{U-' '
case.
^ (=,- _./_
^1,
a;,,
. ,
Detennining
or Xp.
?n,,
. . ,
m^
this,
where m
by equation (L)
m:. '"
Therefore
( 1
^ ^ ^Ot='-
is
".
tti
side to be ^P-,a
+ ^k
equal to
is
a.
- u^'^^p;
\^, ^a)
equal to
^(W*'"- U'"'^(lt/','_yz,,m,
we may write
-M^i), mj.)
is
"
_M>4).mp)
_Mp,mp)'
L,
^(F-s)'
therefore equal to
(j^, m_jpi,m, _
(tr^-
^ (/'."'_>'">_
the expression
,
/^(tt*. "'_/'.."
_yip, mp^/
'i,{V-r)'
is
coincides with
so that
+ M"^''"p =|np,.,
w'"'-"'-!-
and
when x
t;^!
v^' '^)
"i
yXp,
j/ip) /
i;'"' "*'
@ (^, m _ ^, m, _
(v^>
yCp, J^)
_ j^p, "ip)
equal to
-2aU(r-s) + 2ar{r-s),
or
-2a(?7-F)(^-).
that
is
B.
2ait^'
''
(*"
*'
+?(^'^),
19
'
DEDUCTION OF A FORMULA
290
[194
which denotes
r=l
same
i,
187,
a,
here used,
is
the
as that used in 138, Chap. VII., and denoting the canonical integral
r=l=l
From
195.
'''"
r=i
^^
we have
the formula
^ (m*'
"*
^(Ul"'
M*^.. '.
-TT.
- M^' "^)'
since
^r. Mr
T>r. fr
r.
Mr
,-X,
1=1
we
obtain
where
JJ
[/
= M^'
tt^;,
'
m,
+ M^. '"P,
"'
+ M^p. "V,
and therefore
r=l
Hence, differentiating,
S -f
[(:r,.
X)
{xr, /.)]
[^)
^, (tiM.
U),
where
r.()
= g|log^(.);
but, from
dC7-i
where
dx^,
.... diTp
= i)^"'"'.d^,+
+Du^-'^.dxp,
_,
ef/i""'"''
a;,, ...,
Xp,
we
obtain
dxr
rf
thus
r=l
which
is
etc
196]
From
196.
291
the equation
+-^^"^=1?*.
^r..+
differentiating in regard to x,
write in
the form
r=l
r=l
where J7=tt^""''
Thus,
if
^u^,^^^
we take
Z7o
= tt^"' +
+jimp.j>.
determined from
just as m^,
a;
..., nij,
(m,
the arguments *
"
/i,,
be =
will
?7o
exact value of
"'
J?'^'"
it*' "*
we
m,,
..., rrip),
l^,.
Dp
for the
[/
+ Ft "" = S
= {x,
...,iip)
7-
- '
'
=l
?r(M^'-
/ir()
it"^"'
r=l
= - 2 Mr (a;)
?r (l'""
"'++
U'l" "P).
r=l
If in this equation
F^''^'
where
we put x
+iC"^ = - r2=
we
at vi
derive
Mr(m)?rK-"" +
(M),
+M-p."'^),
z^axe arbitrary.
Zj, ...,
m, by
..., Ci^p,
Ci, 1,
so that
(Ci.mi, ...,mp)
we
obtain
p equations
j/"^-'
of the form
+F''"''-''
=-
lMr(Ci)rrO/"'*''
x, x^, ...,
;,!,...,
(a;,
a;,-,
I,
a;;,
...,
J'^i."''<.'
Ci
w^^i'"^-
^ determined
Xi^p)
= {Ci,
_ w^i.
a;,,
i.
+m'^''''^).
by the congruence
...,Xp);
"^m
...
u^'-
"'
''"
is
equal to
...+i?''^-'"'*-''= S/ir(C/)fr(f^<J
'-'t
no+M^'"'-
(m,Ci,i, ...,Ci^p),
1.=
+W
^-...-
192
);
292
[196
therefore
^iiilQ
where
/jLs
(Cr),
-U"
+U
'
...
-u'^"'
'"'")=.
2 Vr,i[F^""'''-'+...+F C,
v^,
is
when
In particular,
already denoted
a>i
(r,
s)th element
/j.^
and
J'
Vf " =
is
are those
{Oi{x)dtx,
Ja
X,
Xj
;^Xo^^iV---V
197.
7j
functions of
^*.
^i. P
Take a
places.
we have*
variable place
z,
such that
z,
= (Z,
Xi,
...
Xp).
5'i(M.
'-M^..'ni_
_jj^p,7iV)+
" - t(^
fi (m".
U^p."V)
j-Vs,
[{zs, z)
- {z
a)\
-^
dt
wherein
to z
Vg,
is
formed with
and denoting
.^fPij(f^)
ff
{u)
z^,
by
...
Zp,
we
have,
by
differentiating in regard
fi^ j (m).
[/^j(^)-/^j(^.)^'-
-'^^(^i')-
-M^^)^dz
-H^^p)-jl
;Aif+j_to,.)-(,.,,)]j_5^^K.,
.
dzi
dzg
dz,
dt
where
[/"= m^.
w^
m^..""!
In this equation a
is
u^""^,
arbitrary.
?7
Let
it
"*
now be put
w^.""?).
to coincide with z
hence
This form
is
p. 488.
197]
293
Therefore
1=1 i=i
dz,
J
i=l
=1
(^., z)
= D; 2s
]
where
Z)z'
means a
are functions of
,
M, (i)
t=l-
'
^)
(^.
= l i=l
i=l
i c, (&)
Zr + s=i
[(z., ^)
differentiation taking
dt
- {z,
a)]
at
no account of the
WkR.-ll-'fiz, a;
Lt "
are
c is
...,
a;
an arbitrary place
= - W{z;
is
(k, z)
- {k,
when the
algebraically calculable
integrals
k,
and
...,
^^)
- 22Sa,.
k,Zi, ...,Zp)-2
iv.
, /u.^ (^-)
'
[
125)
p
2,ar,stJ;iz)H'r(k).
r=l=l
If
now
{k, Zi, ..., Zp)
{z,
^-l,
..., kp),
so that
U = W'' iC"
'^'
i)Xp,mp
= u'''^ u'"'^'
= *
""
u'''
""i
M^' "^
_ |t*p, mpj
and
{x, Zi, ..., Zp)
(x,
k ...,kp)=(k,Xi,
..., Xp),
is
-EliPij(U).fii(k)fij(z)=W(z;
i
k,z,
^p)
Tf(ifc;
i %
z,h,...,kp)
+ 2i
iar,stir{z)t^B{kl
r=l 5 = 1
iv.
125.
ar,sH',{z) f^s{k),
r-l-l
=
by Ex.
Zp
dk'
a)
138),
U=i
in
z,
= ^z'\ S
= D'^
dt
EXAMPLES.
294
By
[197
the congruences
M^i> ^1
the places
z^, ...,Zp
+ W^J" *P =
-I-
{z;
and therefore the function
In fact we have
Xp, k, z only.
x\
"/^(^p.
Q),
x,iCi, ....Xp, z,
can be expressed by
z,x-^,
x,
x-^,
...,
...,x^=0.
interest of the formula lies in the fact that the left-hand side is a
The
u'<
way
particular
Up.
..., .v^, i,
is
to
particular case,
-U-"
"'')
17
we determine
K^.
'"
jj^ii
so that Xi,
...,
we can
pl
infer
of this form, in
x^', ...,
which
w W w {x,)=D,D,R^-
x^, ...,
by the congruences
y,,_i
.r^p_i
(N)
_^Xp,mp_ _r^^Xp,in_yX,',m,_
ii_
_yX',-i, mf-i_yX,
Wp-i
...,
Ap_i,
X,.
a'
where r=l,
And we
Here the right-hand side does not depend upon the place
2, ...,(/>- 1).
can obtain p such sets of equations.
198.
Ex.
ii.
i.
iii.
is
Prove that
a rational function of
Ex.
formula
1.
is
final
x.
Prove that
.r, a^j,
..., Xj,.
if
{x, Zi, ..., Zp)
= {z,
= {a,
then
ylr{x,
Deduce the
first
The function
a;
z,
formula of
= r*' "
-|-
r^"
final
"
-(-
formula of
.^p'""':
196.
198]
Prove that
.v. iv.
if
ei=it;''"'+
where
a^,
295
+r^;.'""'',
...,
''
'
then
where
7\a
= =^
^^
y\
that
125
iv.
it
follows therefore
by that example,
rid
I'i
Qyd\\ +
is
a perfect differential
it is
e ( F'^- -
of the
Of.
+ Q^dVj,
built
F*'
'"'
7^..
'"p).
of algebraical integrals.
theta series, see Clebsch and Gordan, Abelsche Functionen (Leipzig, 1866), pp. 190
Ex.
Prove that
v.
195.
if
{m\
.r,,
,,
Zp)^{c?, Wj^
")
then
-r*'""'^)=i(r^i'''+
5^1oge(r-'''"'-r^i'"''-
Ex.
vi.
+r^"'-''>).
Prove that
i=l
Ex.
vii.
If
T{x, a
x^,
...,
Xf)
= \^{x,a;
z,
x^,
...,
x)-F^' ^^x,
prove that
-*"'"')
log^(M^'"'-w^-'"'-
= A+AiUi'''+
where A, A^,
Ex.
where
viii.
...,
A^ase independent
of
+ApVp'''+
dxT{x, a
j,-.
Prove that
is
as in 193.
j,-,,
...,
Xp),
[199
CHAPTER
XI.
We
a special
one, characterised by the existence of a rational function of the second
In virtue of this circumstance we are able to associate the theory
order.
199.
= 4 (a: - a,)
y"-
We
Tp
... (.^
is
ftp) (a;
Cj)
- Cj,+i)-
when
2,
may
is
We may
argument.
Riemann theta
hyperelliptic case.
present volume.
*
memoirs
of Boseuhain,
Gopel, and
Weierstrass, referred to in 173, Chap. X., which deal with the hyperelliptic ease, and general
memoirs on the theta functions, the reader may consult, Prym, Zur Theorie der Functionen
in
einer
zweibltittrigen
Fldclie
1885);
1866)
(Ziirich,
Schottky,
162
Prym, Neiie
Abriss einer
bei
Neumann,
Anwenduny der
werden (Halle, 1876) Brioschi, Ann. d. Mat. t. x. (1880), and t. xiv. (1886) Thomae, Crelle, lxxi.
Krause, Die Transformation der hyperellip. Funct. erster Ordnuag (Leipzig, 1886)
(1870), p. 201
Forsyth, " Memoir on the theta functions," Phil. Trans., 1882 Forsyth, " On Abel's theorem,"
;
Cayley,
"Memoir on
the
and
Crelle,
Bd. 83, 84, 85, 87, 88; Bolza, Gottinger Nachrichten 1894, p. 268. The addition equation is
considered in a dissertation by Hancock, Berlin, 1894 (Bernstein). For further references see the
later chapters of this volume which deal with theta functions.
THE ZEROS.
200]
297
200.
Throughout this chapter we suppose the relative positions of the
branch places and period loops to be as in the annexed figure (4), the branch
place a being at infinity.
Fig. 4.
(*.)"--
In the general case, in considering the zeros of the function ^ (m^' "* e),
we were led to associate with the place m, other p places w>i, ..., nip, such
that ^(m*shall
has mj,
"*)
always take
shewn that
b,h
are the
1+
=?ni(i,,
wherein
???,
. .
+mp(Or,p+in^(o'r^i
rUp, ,',
,
.
h remaining
any branch
which
'''
m'"'
^ (m*. *) =
mp
can be
It
+1
{r=l,
half-periods,
iil''
*',
2, ...,p),
p odd
6^,
we
if,
branch
2p-|-l
we
this fact
shewn that
h' is
. . .
at infinity.
+ mpa'r,p,
p+\
and
is
0,
place,
u'''^' = ^D,.
denote by putting
shall often
In this case we
179).
is
6,6'
Ur
X.
if b, h'
Up
...,
h',
for
have, necessarily,
m,m^, ...,mp,
the places b,bi,...,bp. In particular it can be shewn that, when for b the
branch place a is taken, and the branch places are situated as in the figure
" "i, ..., n"' "p is an odd half-period. We have therefore the
(4), each of
statement, which
is
x^,
Xp as
zeros.
its
It
is
"
assumed
that
the /miction
be
seen
to
be
justified.
m*-'"'
is
a half-period,
an odd half-period.
(ii)
In regard to
is
it
Riemann
surface
is
sufficient to
prove
(i)
...,
w"'"" is
is as follows:
Let
upon
THE ZEROS.
298
[200
the figure, whereby to define the left-hand side, and the right-hand side
finally let a
so that
nrir
from right to
then
either
is
left,
Thus,
Ur
the
for
"^iWr,
and Ur'
of these results
pass from
Cj
+ mp'w'r^p.
+ m/coV, 1-1-
WlpWr, p
W
first
H-
we have
6,6'
h'
if this
to a^,
'
= Wr,
1
and
Br, 2.
will
it
is
we should have
be sufficient to prove
exactly similar.
Now we
can
Fig. 5.
Ur
On
we can
pass from
Z(0 r,l
Ur
to a^
Ci
But, in
fact,
as the integral
(Uf.
two
)
sheets,
=-
M^
i^' "
+ {Ur
is
may
which we
result
={Ur
Ur
to
c,
Thus we have a
of the form
-'
|
we have
,
and (m,
= w.
y has
THE HALF-PERIODS.
201]
299
Therefore,
= r,l,
Mr
">,
w"'
"
"^
we
which
if
this
2+
(o'r,
&)',-,
p)
+ (<'., +
',,
p),
+ (&>r, r, i+l -
<r, p),
+ lj,<r, p + nii'<'r, +
1+
+ Wl'pQ)'r, p
We have stated
i.
a>'r, i)
^;r.
(<,,
equal to
(eo'r,
and
+ mpiiip =
-f-
that
if 6
1.
*%
(a
Oj),
(ci,
6-2),
*''
c),
(cj,
(cj,
c),
vi:.
The reader
ii.
201.
will find it
and
Consider
fifth
now the
2p-|-l
where
on,
b, b',
and
than
b" are
finally
half-periods
a,
w*'
"
and
-I-
(^
any three
half-periods
From
a.
''' ",
we can form
vf>- "-
1-
these
wherein
j
b, b'
** wherein
we can form
is
[a
half-
-f-
* " -f
any of
half-periods
by adding any
w". ",
and
so
of the
The number
*['T)
equal to
or to
300
[201
we may say
that this
number
zero half-period.
And
tt*
'
",
we may put
it*"
and
here, since
p, or
is
having a
we
this
"
-I-
is
may
M*" *
-I-
-f-
It*""
"^
s 0,
sums
of
cannot be greater than 2p. Now this conequivalent with the statement that there exists a rational function
different periods,
less,
gruence
'''<
be omitted.
'
gruent.
((,*
for
(Chap. VIII.
Since a
I 158).
is
be expressed
+ y(a;,
1)
and the number of its zeros is the greater of the integers 2r, 2p+l +s. Thus
the function under consideration would necessarily be expressible in the
form (x, l)r. But such a function, if zero at a branch place, would be
zero to the second order.
Thus no such function exists.
On
is
zero to the
and
is
infinite at
order
first
order at each
a to the
(2^-1- l)th
w"""4202.
With the
-h it"p'"-|-
wherein
+M'=p."+ w.
7ill(Br,l+
we may
"'"-!-
associate the
+mp(Or,p+
TOi'toV,
is
= 0.
expressed by
+Wp'a)V,p,
symbol
(fci
A:^
Ki
ic^j
. .
.
fcp
. ,
tCp
sum
k,
equal to
of two or
202]
by
Thus
2.
call
Tf
it
when ;;=
for instance,
.
2,
we should
301
write (^^]
UU + f^^)
VOly'
C^^]
VIOJ-
we
this symbol the characteristic-symbol,
we have therefore proved,
in the previous article, that each
of the 2^ - 1 possible characteristic-symbols
other than that one which has all its elements zero
can be obtained as the sum
of not more than p chosen from 2p + 1 fundamental characteristic-symbols,
these
I.-
true for
is
We
cases*.
all
may
Thus
p=
(n)
respectively
2,
OOOOQ
by the numbers
1
The
02.
5:
|j, which
= 10
is
equal to (,,)
(,
pairs, represent
1, 3, 0, 2,
4 in
written.
Further illustration
In case
p = S,
the
is
2''p
0,
1,
2,
3,
4,
5,
6.
All
may be
may be mentioned
is in
all
The theorem
proof,
is
further
302
203.
h^m,
we
write
= niiBr, +
m'
\mg=M, + \ks,
and each of
kj
kg,
is
^ m/
either
.)
of
m^ft)^,
which an element
+ m/wV, +
1,
we have
= ?>(";
is
given by
+ rriprn'r^ p,
or
^(-|-in^,
m',
[203
M + \k,
M' + ^k')^,
where
X=
[27;
{M + ^k) +
27?'
{M' +\k')] [m
and therefore
^ (m
The
ifc,
1 A;')
e-^-"-'"^'
(u
* fi,,
,)-
represented by
We
chapter
last
+?''
the
+'"), or by
^H =MI"'" M""'
and,
when there
_Mp-l.'p-l^
which vanish
the argument, that every even half-period can be represented in the form
|n' S
M>'.
""i
-f
+M'7,',p;
form
J
M^P-I
for
'
'^-1,
and
fej,
n-^,
..., 6^.
.... Tip,
in case
^=
3 there
is
one such
function ( 185, Chap. X.), and the corresponding even half-period \i"
such that we can put
is
204]
TABLE.
wherein
is
x-^
= M^i. ,
now, as in
200,
Ur"
we
303
Since
easily find
= (a>r, +
3
<'7-, 1
to'r, 2
<>'j-,
w"^'
s).
"'
O),,
O),.^ j
a)V, o,
and therefore
^ fl' =
is
o'r, s)-
for
zero
values of the
j)
= 4, the
argument are
for
periods given by
where
b is in turn
204.
to
The
verify the
second and
equation (iQ)r=TOi<o,,
I.
FuDction
fifth
columns.
The
set of
>
is
denoted by putting
in the case
p = 2.
JS2
=J
'
\12/
304
//.
Function
Ten even
[204
p = 2.
205]
305
Next we give the corresponding results for jd = 3. Each half-period can be formed
of not more than 3 of the seven integrals '- " ( 202)
the proper integrals
are indicated by the suffix letters employed to represent the function. We may also
associate the branch places with the numbers 0, 1, 2, 3, 4, 5, 6, say, in accordance with the
scheme
205.
sum
as a
1.
2)
3.
C.
Ci,
Cj,
C3
1,
3,
5,
0,
2,
4,
6;
then the functions 5i(m), 53(2*), 3^(u) will be odd, and the functions 5(m), S^W, 3^{'u), \(k)
will be even
and every function will have a suffix formed of 1 or 2 or 3 of these numbers.
There is however another way in which the 64 characteristics can be associated with the
;
combinations of seven numbers, and one which has the advantage that all the seven
niunbers and their 21 combinations of two are associated with odd functions, while all
the even functions except that in which the associated half-period
such a scheme
is possible.
It will be seen in
One way
1,
is
2,
is
many ways
4,
5,
6,
By
column.
/.
p=
Z.
306
BY ACTUAL CALCULATION
Table
I.
(continued.)
[205
205]
OP THE FUNCTIONS.
JTS6
even characteristics
for
307
p=
3.
308
METHODS OF NOTATION.
Table II. (continued).
[205
206]
309
We can equally well choose seven fundamental odd characteristic-symbols, associated with
the integrals denoted by any one of the following sets
:
cci,
CiC
c^c
C3C
The
c c^,
c C3,
ca.fi^,
ca^a^,
c aya^,
c^c^c^
c-fl^^,
Ci^ai,
CifliOj,
c C2C3
C1C2,
c^c^,
c^c^i
C2C3,
c^^a^,
c/x^a^,
c^a^a^,
c c^Ci
C3C1,
C3C2,
Cja^a^,
c^a^i,
c,a^a^,
c c^c^
c^a^^,
c^a^a,,
ca./i^
a,,
a-fl.^,
a^a^,
CiO^a^,
0=2,
a2<^3,
a^Ui,
c-fi^a^,
<hH'^\^
H"-^\-:
c3'*i
"3,
''s'^i;
''a'^zi
Ciajaj,
C2a,a2,
(^^o-x'^i-,
ca-fl^
general theorem
is it
is possible,
",
fax, eXfi
ca/3,
f)
= 0.
The proof
is
given in
a subsequent chapter.
Consider
206.
now what
6,
. .
6^
^(m|''>'"+
^(m),
where
i^p),
and u
given by
is
Ur
= u;'
'
+ C'^,
(r=l,
2, ...,j9),
The
or,
zeros of
^ (m)
z^, ...,
Zp
by*
a,
^,
with any
..., aip
x^, ...,Xp
'Xj,
be at
the places
a,
z-^^,
z^,
...,Zp cannot
and Chap.
98,
The two
^ (m),
denoted by x and
x.
..
Xp
III.)
same
value,
be coresidual
when the
places
z^, ...,Zp
regarded as a function of
OC3,
is,
a;i.
310
The
zeros
+ **>")
of ^(zt|M''""+
are
to
[206
be determined by the
congruence
by
or,
a*-')
= (6i,
...,
h,
X2, , p);
207.
sum
the
Now,
in fact, if the
sum
of the characteristics
r,, ...,
._ ^(tt; gi)^(w;
a periodic function of
^p...,
qn differs from
is
6, ^2. >
2,
ga)
u.
where
m denotes
a row of integers,
we have
2r', Sg 2r,
and
if
Sg'
side
is
equal to
1.
208.
/{u)
is
..., tCp.
(wltt''''')
x^, ...,Xp.
By what has been proved
numerator and denominator it has, as a function
of Xi, the zero b, of the second order, and is infinite at a, that is, at infinity,
Thus it is equal to
also to the second order.
(b x^), where
does not
As the function is symmetrical in a;,, x^, ...,Xp, it must
depend on x^.
therefore be equal to K{b-Xi) ... (b-Xp), where
is an absolute constant.
is
^ib-xMb-<^^)...(b-Xp) =
^
vK
^-i!il^)
^ (m)
may be
the function on the two sides of any period loop have a quotient which
constant along that loop, and equal to + 1.
is
209]
311
Brioschi .
and
even functions.
of the functions
V'^=sn
u''.
) /
Of these
6.
The
q^ {u).
...,q
q^^
{u) are
\/l-a;=cnu, \/l-^^^=dn
m,
by
{u),
{u)
u,
of the integral
dx
V4^-(1-.2;)(1-F.r)'
'
209.
+''* '')y-HM)
^_ S(m|m''-'"+
^(it|M''.')
wherein
bi, ..., b^
the cases
A;
^(w;m*'1'')
<^ + 1. By
and if ^1,
Zp denote the zeros of ^ (u m*"- " +
numerator, as here written, consist of the places
. .
We
Zif
..
Zpj
x^
. . .
Xp
, Xp.
sets
of
consider only
rational in Xi,
is
+'':.)
Thus the
a.
p + k
Xp.
It
z-i,
Now
any
u,
a.'^,
...,
Xp.
Let
X,
...,
x,
so chosen
that the
The denominator,
61, ..., bk,
and
as here
also vanishes
Xp, Xp.
ji
A 1, it
is
By
t.
is
t.
( 27).
x. (1880),
t.
xiv. (1886).
of
Werke
DEDUCTION OF RELATIONS
312
(wherein 2
A;
denominator)
(k-1).
should be equal to
[209
is
ju.
+2
coefficients in
2X+2p+ 1 >2(fj,+k)
+ 2p + 1 < 2^ +
= p-i{k + l). In
2\
\ = ^{k-S),
ij.
By
is
\= ^k 1, fj,=pl ^k,
when k
odd,
and
and then
+ fi+2=p.
+ v{x bi) ...{x-
'K
depending on x,
y, to
we
Xp
...,
ay
a^rVr
wherein
<^
denotes
(x)
bt)
not
factor
(a;
yr
y,-
M-l
/x
bk),
the
for r
first
>
1,
is
suffixes.
is
is
a;,,
...,Xp
by
A(jEi,
we
..., Xp),
that
Ifryr
^(| ^'"
+M^t')^*-'(M)
is
A;
is
210.
first,
second,
^-l
f-
'''
"
'''\
A{x^,...,Xp)
is
1,
X = ^(k
odd,
By means
...
\ = ^k
even,
and when k
_yr_
<f>(Xr)'---'(f>{Xry
in which the
when
OC^_yr
^(mIm**'")
^{u\u>>"'')
where
^ Q HM'
of the
and here
1 ^k
fi = p ^{k+l).
fj,=p
S),
algebraic
expression which
^ (m|m*''')/^(m), we
are
now
we have
already
able to deduce an
^(m1w*""'+
since
it
+M**'")/a-(M);
and taking
by
..., b/c
m*"""-!-
is
1, 2,
+ m''''"
...,p,
k,
the
identically zero,
it
follows
that,
in
the
^^^J
hyperellipticcase, if
u denote
^n of
^-
''^'"'"^^*'^*
^r*'^'''
all the
a?i,
...,
^...
^^^-P*
simplest case
^^^re R{x)
is
= {x-x;){x-x:)...{x-xp),
^ (m|m.. + *..
)/^
^
where
an
is
4,,,
functions.
absolute
K = ^1,
b,
?ft>,
3qb,qb,qb,
2.
is
an absolute constant.
_Jh^
9b,. b,
'^nqb.qb,
=p - 2,
and
and
and,
b,)
there
if q,^
{xr)
(2p +
are
,,. ^^
l)
such
yr
= 1 i<^r
,,
- b,) {x, -
constant;
0,
(u)
Vr
V=i (Xr
,_
all
where
1 ratios
Xp, by the
The
gig
is
E'
'
(x,.)
^,^ _^
qb,.bb,
-ii3fe?63
B,^qb,q^,,qb,'
by those
More
generally, the
the
places x...,Xp,
and since
tions;
^''"'''
Of such
sideration
may
be mentioned at once.
^2iqbM
under con-
We clearly have
^12?6,9/'
^31?6,?6,
'
and therefore
6,-65
9bJ {u)
"23
It is
proved below
Other relations
the 5^8 of single
Werke
i.
^i
will
-"31
( 213) that
be given
and double
p. 336).
(u)
J^^
A\
A\i^{b^~b^)
suffixes, for
= 2, ^ = 3. A
any value of p,
is
{b^~bi)
{bi~bi).
lii.
CASE
314
[211
IS
211.
i.
^=(^-^)
where
by
= (^- J;)
iJ (^)
...,
Xp, a, as poles,
(*-"^)!aT^^)'
replaced
(|-^p). and, in the summation, x, y^ are to be
(|-^i)
X, y.
Prove that
if
then
where
is
an absolute constant.
Bx.
R (br)
(6-^p)'
(b~x){b-Xi)
a^'{u)
elliptic case,
ii.
place b^
is
put in place of
6,
and
+ M'"^-")3''-t(tt)
(!'''"+
5(m|m*''")
S{u\u'''''')
prove that
^^(^^)|^)-''^/"''"-^><'<"-'^|
^.=^^(^)
*^.
^A{x,
where
B is
of the(/) + l) quantities
j;,
jj, ...,Xp,
(f)
...,
(x^)
M=i>-i (* +
Ex.
iii.
is
even,
..,,
Xp),
= {xr-b{)
Xi,
first
row,
X=i(i-2), ^l.=p-\k,
x^,,
and,
when k
is
odd,
X=^(^-l),
is
a constant
!)
!^-!
T;""
multiple of
^/^(^')
This formula
^(^>)
is
true
k(^)-'^
5 (f = (| )
V.
is
when k = %
a')
(^ -
.J;!)
properly obtained.
is
'
put at
(6.-^.)
is
f (^^_J(^^_fc^) ^,^-)
(? - J'p).
Then the
(6,-^,) V(\-^)(6,-^.)
V(&.-^)(6x-^0
'a;.
- ^-fc'
when k=l.
wherein
^;)<"^
iii.
^^^^
Ex.
X+Xia; + X2^2+
of 130,
Chap VII
'
takes
is
^\^^^^y;ipi-2^
^y
?1 i"tg>--l polynomial, i^K
z and"^f
of order ^^ + 1
ea.;h of them, and
d!^*
F {x,
195), with
(cf. p.
X=X,
z)
=/ {x,
-^='
z)
is
symmetric
in
the two
the conditions
satisfies
of the form
which
.),
variables x,
where
-Er. vii.
is
315
vi.
'^^
+ {x-zf^ {x,
z),
X2^ + 3=0,
and
Vr {x, z) is
In case
{x, z) IS
^ = 2,
given by
P{=',^)
Ex.
viii.
in each*.
variables
and write
+x,,,,.,..r+x,,,,.r,
given by
z) is
are to be replaced by
z^, z^
:;;,
1, z, 1
respectively.
the same as that which in the ordinary symbolical notation for binary forms
we introduce homogeneous
/(.)=x.r+x,.r.,+
is
of order ^
z,
= {x-ad{x-a^){z-c){z-c,){z-c^) + {z-a,){z-a^){x-c){x~c,)[x-c,).
This
x,
and f{x)={x-a,){x-a^){x-c){x-c,){x-c,),
prove that a form of
ii.
viii.
is
aT\
for F{x,
z),
prove that
if gj, e^, x,
x\,
...,
x,,
be any values of x,
r=o[G'(x,)f^
G'{x,)G'{x.)
lO'{e,)f^[G'(e,)Y-^0'{e,)0'{e,)'
Ex.
X.
Hence
it
foUowst,
ip{p + 1)
y^h.&ii
(|-a;p),
etc.,
...,
Xp.
(^-.^i)...
(i-Xp), that
E(e,)Ii(e^[^^
(6,, XI (e,) ^z
^^^
is
yr
_^^^ ^^^_^^^
T
^ ^^^ J
f{e,)R{e, )
^^^
f{e,)R{e,)
(e,-e,yRie,)
_ ^^^, ^ ^^^^
f{ee,)
+ (^^-e,)"^
equal to
* It follows that the hyperelliptic canonical integral of the third kind obtained on page 195
integral,
is
F (x, z) satisfying
the conditions
316
z)
2)
p+1
from the
obtained in Ex.
[211
quantities x; x^,
...,
x^,
and
viii.
Hx. xi. It follows therefore by Ex. iv. that when b^, b^ are any branch places of the
surface associated with the equation y''-f{x)=0, there exists an equation of the form
^
where
u = ic''
C
"
-^Hb,)R{b,):&2
Wlu)
{b,-b,r
g.^^^^Q,^^^)
is
''" "'
+it'P'
"".
The importance
{^-Xp),
and
"' +
For instance the solution
is
+uT"'' =
given by the 2^
({=1,2,
Ui-
equations
...,
And
it is
denote the
ii^,
Up.
...,
finite
...,p).
For
.4^.
b.
As already remarked
201)
we have
Cp}
(b,b ...,bp)
{a,
and therefore
Now we
easily find
of 1p integers, P^,
...,
V,
by the formulae of
..., Pp,
-..,
190,
bp).
Chap. X. that
if
P be a set
Pp, P/,
^'(u + ^np)
^(w;^P)^
hence,
^-^^^^Aib-b.)...ib-bp),
and, writing
V,
V in place of
l{u + J'^)
thus,
we have
=^(^-W)...{b-bp');
by multiplication
e-^pp = A^(b-b,)...(b-
bp) (b
- h')
...
(6
bp').
213]
IN
317
and hence
^th^) ^ ^ {b-x,){b-x,)^_{b-x^)
- a,)
where/(a;) denotes {x
according as
The
m**'
" is
. . .
reader should deduce this result from the equation ( 171, Chap. IX.)
rin
f(t/,,
...,
When
tt
V(U
^(C/i,
Z7-*
-vl
(/p, f,,-),,)
U-f6*'y^)-(^_^(j)
v'i- ("^~-^(-^i))
= M*' "+''"'+
S^u\^)^
+ M^>"p, we
{b-x){b-x,)
(6-.rp)
4 xJg^PP
where/f(|) = (^-:i7)(|-^-,)
If
. . .
211)
_y^ _i_T
(.r,.)
'
(|-:rp).
we put
particular
in
P (a;) = (a; aj
r P
i.
L=o x^-b
f (J)
(Z-^(a^))
a;.
9^ ()
i-^'-^i^p) )
t/p.
...,
(.t Op)
turn at
in
and Q (x)
the
places
...
Oj,
(x
...,
a^,
Cp), and
write
use the
equation
(x-Xp)
P(x)
(^x-xC)
we can
where
is
6;
and
...
...,
is
We
- a:.)
(ttf
(g,-
- Xp)
(a;-ai)P'(ai)
"'
"^T
y( M|'H.'' )_
Another form
yi
from
a;^
is
given below
( 216),
where
Xi is also given.
can also obtain the constant factor in the algebraic expres.sion of the
Let
ft, ,
h^
Zj, ..., Zp
+M^.'^+*"''=^'>' +
so that
+'""'';
vanishes
then z ... z a are the zeros of a rational function which
Such a function can be expressed in the form
+ C^r-feOKl)"-'
{x-x^)'
{x-x{)
.y
Werke
i.
p. 328.
318
where
{x,
l)-i
is
[213
the p equations
-yi + K-6,)(^i,
thus the function
l)''-'
= 0,
and
therefore, putting 62
Mn
fo'' -^i
(62-V~^
(62-^1)
Now we
(i"=l, 2, ...,p);
is
"'
WJ
'l*fi.(^i-^)(^i-&2) ^'
'
-Vu^- "p by
u,
and m^"
+m^'
"'
""
by i>,
the results
^ '('^k'"") ^
where
m*"-'
''=^Qp
+ ( &2-^.)
jy
S2W52(m|m'"")
which, by the formulae of
62-^1)
i.h- h)
190,
is
the
IT,
y.-
'^L^.=lK-6l)(^i-62)^('^.)J
'
S{u\u''^- + u'''''')S{u) _
t is
hence we have
s^(.|.^^-'').jW
where
5^W^) ^ ^
(ft.-^p)
same
.JJ-
as
P
y,
of this
constant factor.
Ex.
214.
Beside
coefficients
of
differential coefficients of
theta functions
^(m)
We
have
algebraically;
as
arguments
was to be expected.
in regard to the
,,
the
second
...,Up are
!-''
+2
i^t, i
[(a;*,
*=i
* Cf, also
Thomae,
X)
- (a;*,
/i)]
^
at
^^*3
319
...,
a^i,
then
,a;p) = xP-'-hixP-^-^+h^xP-'-^-
p fundamental
integrals
"^--^^
Thus,
/i,
a, Oi, ...,
we take
..
the integrals
'~dt'
FT^^O
J?" (a;)
if
m^-",
...,
+/-.y-ift,
it is
where
f ",
Xp, r together,
Xp).
ap for m,
wij, ...,
m^
fi,
we
have, writing
( 200),
*.=
[xtx
F' {xk)
Xifj.]'
Jf
k=\
is
{Xi)
Xk
equal to
?.{m^.
which
is
Now
+ m=^...+
-t-M
^,
{t)
denote
{t
p)
'
^y) !l:i^*
a;.
x){tXi)...{t Xp),
'^^
R'ix,)
R'ix)
R'{xp)
then also
^iXp-i-i
( a^i;
a;,,
F(^)
Now
is
....a^p)
Xp-t(a;i; ,'.
"^
ypy,^-f-i(^p;
a^^,
..,a;p_0
F'(Xp)
-,
a;i>)-Xj'-<(^i;
3^i. ^^a.
_
-Jp-.-^{.x.,...,Xp).
.^p)
equal to
[x{''-xV\^ + h)+Xr'\xh+k,)-
+{-\)^'xkp_^,]
320
[214
{x,-x) [4"'"'
- x{''~\ +
Xp,
x^
+ (-)P-^-V.--i]
or to
x)
{Xi
i|^j,_j_j
(x^
x^, ...
Xp).
Hence
X^ i(^u
^2. .
a^.
"
^ yy-i(a-i
^p )
a^i,
R'(x,)
x^, ...,
'
Xp) + (x^-x)xp-i-i{^i;
{x,-x)F'{x,)
^11 ^2) )
Xpiy^ii
^p)
While,
'
~F'{x,)
%p 1
x-x,
y^i
^3
,'Vp)
^2,
>
^p)
F'{x,)
also,
Xy-i \X
',
Xj,
..
Xp)
f,
"
R'{x)
\^k
TCpi
Xi,
Xp)
1^
X-Xt'
F'ixn)
t=i
Thus
fp-i(x,
...,Xp)=
Xi,
,
-''
J-=l
(,**)
Xk
+/j,_i_i
(a^i,
...,Xp).
+ M==.'''. +
...+M=i>."p)
+ ii''' + i^''+...
\-Ll'"''-\fp-i{x,Xi,...,Xp)
equal to
^i{u^- -
+ w^..
In this equation
...
tlie
left-hand side
x.
is
...
symmetrical in
Hence the
left-hand side
hand
constant by
(7 we
'^i'
>
^y)
UpXp-j (^f
From
/x.
^,
i^i
a^-i)
'^
2R'{Xp)
2iJ'(a;)
215.
and the
a constant
is
have
yXp-i(^i
is
X,,..., Xp).
'-fa
It
+ M^,'^)_xr""'-
-i?""^
obtained,
If
it is
- ykXp-ii^k
-,
X, Xi,
2R'(x,)
.^^Xp)
ytXp-<-i
('^t
a^i.
2F'{xi)
'Bp)
215]
321
when expanded
where
is
when
limit
in powers of t
~ {\+AP +
...),
Li
are cancelled
by terms
x^,
...,
Xp
we
infer that
the
difference
_ yXp-ii^>
T^. c
^1.
2R'
has a limit independent of aii,
...,
,Xp)
(x)
u''"''>
+ ...+u't>'"'p,
It
i.
Thus the
final
formula of
p=l
we
where
an absolute constant.
f,
This
The
is
{u''"'''>=0.
Thus
identification can be
"O+if^ll
is
equal to
W- =,
-A
of the facts
- >'
(u), (ii)
^'^^"y
f,(..-'o=f.(o.+'-)=-4'"=-r^"B.
final
the results
argument u
that
200) that
+ m'^''"' +
infer
214, respectively,
I> is
Ex. ii.
formula of
In case
m.
(i)
The Weierstrass
%*"' = w + ' - u, so
fi"'^
'^^^^
^>"
''^
EXAMPLES.
322
chosen as in
places
138,
Cj, a,, c
Ex.
i.,
are chosen
(iii)
dCu = -
by Weierstraas
when m=0,
[215
therefore
(iv)
d ('"'''')=
-f.
^^^ branch
(in
is 0.
The
D is zero.
Ex.
For ^ = 2 we have
iii.
y{x-Xi-x^)
i{x x^{x-x^
-f2(!(^'''
_ y^{x^-x-x.i)
i{Xi x){x^x^
y^{x^-x-x{}
^
2(^2~^')(^2~^l)
'
+ M'^'>'''+!t^^'''^) = i=''''+Z^"''+Z^'''
y
Vl
2
2(.r-.-ri)(.r-.i7.^)
(j?i
- a.-)
Vi.
(.^1
- 2:2)
+ ^^
^
and
-Clin""
''>
<"*)
= Z?" >+Zf
''^,
where with a suitable determination of the matrix a which occurs in the definition of the
integrals L"'
**
(m),
we may take
( 138,
Ex.
Z^-=/j
i.
Chap. VII.)
|x,x^.
-f. ('""'' +
Ex.
iv.
We have
+ M^'P) = i^">'''
+
"
210) obtained 22"-
Z is
= *'' +
(''"^L^
t=ija^ y
5(ht''"''+
where
= ^'I
+ Z'^'P
P
by
^-,t3i"Op;
Hence we have
<ri(k*""+
+^**-)-f,()=l3,,E=i%^--:^?H)2|^.
is
a,
obtain
so that
-f,(|,^'') = Z^..''.+
+Zi.- .
...,:.,)
{b^x^),
we
216]
By means
of the formula
(i(u+^ap.P')=Vi,iPi +
which
is
obtained
Ex.
is
from the formulae of 190, we can infer that the formula just
in accordance with the final formula of 214.
We have seen
V.
( 185,
="'"'
may be shewn
Hence
-fM^'^P.
-I-
possible sets of
it
by taking
321) that
(p.
To
+v'i.pPp' + (i{u\hOp.P'),
+Vi,Pp+v'i.iPi
easily obtained
It
323
if
for .,,...,
...,
a^,
.ij,
the
half-period
we have
fa)=?;.
0,2.,,
CiWl,r,
...,
fiC^^'+l'-O^-Xf
or
f.K,,
0,p, r) ='?., r,
*'"'
and
Thus each
i;j, ,.,
rfi,
=-Zj-'"-.
Or fi(j*"-'^)
periods.
Ex.
vi.
The formula
of the text
(p.
321)
is
equivalent to
where
For example when
216.
^=2
It is easy to prove, as
remarked in Ex.
iii.
215, that if
the integrals
138, Chap. VII.) be determined so that
i.,
then
p, fixPo
fkScPdx
Therefore, if
?r ()
be denoted by
p.,
{u).
we have
aui
212
324
and
<Pp,
where
iW-^ Mp+r ^i
arguments
[216
u,
jr^^
Whence,
if
^k f" ^
Xpi \^*
F'
*=i
_,
>
^p)
^i>
'
...,
Xp satisfying the
{x^)
xlF{x)
.
k=x{3:-ook)F'{x^y
but we have
a;'-'
^j,, i
>=i
(m)
2 4'Vp,i(")
_ *
t=i
Thus
!'
=!
a;,,
F{x) = xP-
xP-^
j>,,
- xP-^^p_ j,-,(u}-
p (u)
^^_
(u)
= 0.
^'
= (-)''~'Fp,i)-r+l(M).
^^t.ytXp-iCa^t; x
9"i
we
putting
because
F (x^) = 0.
for
i,
F'
(xt)
...,x)
'
infer that
Thus,
if
we obtain
yk=ir{xt).
BY THE p-FUNCTIONS.
216]
325
We
p
have previously
x^, ...,Xp
such equations as
y (^
|Mi, )
(aj-x-,)
^^^^- =
fii
Ex.
is
(aj
- Xj,) _ {ai-x,)...(ai-Xp)
of the form
iPp.
W-
&P. 1^'
( **)
-iPp.^
W-
For ^ = 1 we have
i.
This
...
commonly written
is
in the
form
6i -" Cq
sn^ (m
^.r.
ii.
For ^ = 2 we
V ^1 Cj)
have
..a,,
a\
iii.
W"
Pi.
2 (),
etc.,
^ij
itiq^^, \i^q^.
= V -/ {a^), we have
= S>22 () P12 (') - P12 () F22 (') + (1 + 2) [(?12 (m) - P12 (')] + 12 [P22 () - F22 (')]
;!;. iv.
Prove that
+"
^=3-^;+^'3i7:+
.Er. V.
If,
with
P {x)
-1
-,
37/
^.r. vi.
to denote {x-a^)
prove that
notation,
"^3F
shew that
-/^^lthen
jg
3Fr
sir/
9Mp'
if
-/:;^i,
(<(-:gi)
(gj-^p)
^'(<)
[216
EXAMPLES.
326
The arguments Fj, ..., Fp are those used by Weierstrass {Math. Werke, Bd. I. Berlin,
results with those here
1894, p. 297). The result of Ex. iv. is necessary to compare his
obtained.
The equation yr = ']r{Xr) is given by Weierstrass. The relation of Ex. vi.
Hancock {Eine Form des Additionstheorem u. s. w. Diss. Berlin, 1894,
is given by
Bernstein).
Ex.
vii.
fi
(tt"^'
"
vu,ii{^ic,
^)-
{^-k,
J*)]-^
where
that the function
is
Ex.
F^
""
viii.
denote
in Ex.
i.
.v.
Here
c is
138,
if"' ",
u'''",
obtained in
{x
- .Vp).
138,
and
when the
place a is at infinity,
px,^^
VV-n r^"*-
7:
Hence, when X2,,+ i = 4, shew that the equation obtained in
from the equation (Chap. X. 196)
/^^"'" +
+i^J>.'"P=- I u,(OT)fr(!t^"'"' +
215
(p.
321)
is
deducible
+U'P'"''').
We
ix.
if
we take
places x^,
..,, fj,,
such that
M = a^i'''i+
+u^f
=M'^""'+
+ M^P' "P
+U^'"'^I\
-M-t)=M^""'
'^P
these 3p places will be the zeros of a rational function which has a^, ..., a,, as poles, each
to the third order.
This function is expressible in the form {My + NP)!?^, where
denotes {x-a-^
{x-a^\ Jf is an
integral polynomial in
of order p.
integral polynomial in
a;
of order
p \, and
is
an
217]
by
EXAMPLES.
Chap. VIII., where /=i^- " = iX,,,^ (^
154,
(^^-'+
and taking
X2p + i
= 4, we
327
^.^
Writing
Z in the form
)P
)^+ix+
-iA.
N^P-Ul^Q^{x-x,)
where Q={x-c){x-Ci)
(^-Cp),
/).-^,ari
where
= \/(ai - ^,)
i?i
{x-x){x-z,)
we
- Xp), ?i=v/K-^i)
A we eventually have*
(a^
t-i
Ex.
(^-a
{x-z^){.v-C,)
-^
Denoting
^o;. xi.
by
2, ...,p),
K-Zp), ^i = V'(ai-f,)
P(ai)
(a.--f);
V-QiOi)
X.
= 1,
(2
()
-f
(m + jj).
the equation
Cv,
gives
-Fp.r
(r)
where
denotes
/>,
VC";
W+
'-'i
Pp,
(i
W=
ir^\,
Ci|>I''Vi-i>(jl'''] vju
2, ..., p),
1=1
.J^p)-
It has been
shewn that pi
is
a single valued
function of u and it may be denoted by pi (u). Similarly stj is a single valued function
of u + v, being equal to p({ uv). The equation here obtained enables us therefore to
express pi{u+v) in terms of pi(u), pi{v), and the difiFerential coefficients of these; for
we have obtained sufficient equations to express Pp_ ^ (w), Pp, r (*) in terms of the functions
Pi{u), Pi{v).
more elementary
in a
way.
We
217.
last
chapter
i 3
if,
integrals
if
''
( 192,
a,
occuning in the
we have
in the hyperellip-
tic case
T/
where f{x,z)=
x^z^{2\^
+ \^^,{x + z)].
This equation
is,
however, in-
i=0
*
i.
side, is given
by Forsyth, Phil.
RATIONAL EXPRESSION
328
[217
J^ y
we
k=i
take
where
0,- ^
Cjc^t;
then
a+^C,
138) this
is
^z,'c
(p-
f{x, z)
is
194)
(p.
where
now
"m^.'
so that,
zY tl^Ci^ ^,x^~'^z^-^,
^(t() ( 189)
Since
22(7^, ^wj'
ic''-
+ *. =
(?i,it(i
m^.
+ u"' "% we
(7j,^ttp,
^p{p+l)
have
equations of the
form
22g,(.)4-^4-^=^^;^^'.
where
Hence,
if ei, e^
^.+ m^..
^,
+m^j>.p,
,-
= 0,
1,
...,
and s = 0,
p,
1,
...,
p.
is
'y~{^y;^
4Cr (Xr)
(r (aig)
being correspond-
Gm=i^-ed{^-e,)(^-x)(^-x,)...{^-x^), R{^)H^-x)(^-x,)...{^-x^).
Suppose
now
that
f{x,z)=f(x,z) + 4!{x-zy'Zl,Aijxi''^xi~\
i
f(x, z)
is
22
and by Ex.
[^,
x.
(.)
^^r = R
-A. J
211 this
iE(.)ij(..)[!^,^_,,_
viii.
(e.)
211
where
'
then we obtain
R (..) 22 ^If-Jp;:'^
equal to
is
yr
) {e,
4,{e,-e,yR{e,)
f(.e,)Rie ,)
(e,
- e,y R {e,j
Ae_e,)
4
(e.
e,y
OF THE FUNCTIONS
217]
329
g).
and therefore
E ie.)
[I
(,.,^)J:,^)^,(,^)
f{e,)R{e,)
4
This
is
xi.
^jiPi,j(u)h
b,
(e^
in it the matrix
211 that
/(e
f{e,)R{e,)
- e,y R (e,)
(e,
- e^Y R (e,)
is
-^^^^-^^^ +
places,
^p{p +
solving
as a linear function of
e\~ ei
eO=
we have
b,
b^,
is
also inde-
are in
turn
ip (i? +
Bi, Ca
being
any
quantities whatever.
1?
By
X
(t()
p+
e,)
^^^
By
(e.
arbitrary.
be any branch
if h^, b^
putting
b,
at a, that
also at a, this
is
^(m|m'')
i.
dividing by
(first
/^Voo~"
Ex.
at infinity
When ^ = 1,
'
p-1
~"^
6^
),
and putting
216)
-Py,i()-
&P,p(^'')-
the expression
f{x,z),
S^'z'[2X2i
+ X2i + i(.r+i)], =
-2g^-g^{x-\-z)-\-Axz{x + z),
and
if
s^^Az^-g^z-g^.
Therefore,
by the formula
this is
^x.
a well-known
ii.
i.,
result.
When ^=2, we
easily find
R (ei) R{ei) _
G' {x^) G' {x,)
* It is given
{x- eiHx-ei)
[X - Xr) {x - X,)
1
(.IV
- ^,7
p. 268.
330
and thus the expression
^l,
is
(m)
+ (1+2)
6^1, 2
()
+ l2 ^2. 2
[217
p.
equal to
_ (^2-ei)(^2-g2
{X^ - X)
-Xi)
^yy-i
say,
Ex.
i.,
we
?1,
()
where M
i-.
+ (ei + 2) Pi,
m"^'
+ m'^^
'
()
+ Pl2 ()
(1
of highest degree in
a, that is at infinity,
+ 12 g'^ 2 ('<) = -
the result
-f'(i^f )2 - ^1^2
+62) +
2,
(!
^1%=
^2-2 ()
61)
is
(a,
'_
'
as in 138, Ex.
'
e,)
"2)
Ex.
From the
iv.
g> (t0
where
[, 2 are
formula,
+ g>.2
"'
forp=2
W K+
2)
i.,
that
'_
'
i-2
Pj, 2
f'2g2^-
'
-txt^q 2 /("i'
+ ^-ZTT
?i2^+r7T3^2+i2
4(a,-a2)'
where
(I
"''.
Prove, for jo =
iii.
Pll ()
"'
z)
^i)
perfectly general.
is
-f (^.
A{x-Xif
(j,-2
K+
'
<2)
,
l^ll^'
- (. + 2) 12,
+ P22 W- "l"2=^^
^^^^^
ita^-t?
iii.)
'
that
"i~"2
+ M)
- ^
(m)
m,
+ ^,2 (m"^' +
"
'') j?22
() -
x,
P22
22p.()e,
62
218.
ej,
+ ) P12 () = 0.
tjj
&i)
A(ei)/-(62)22jy,(j_^)j^,(j^^|_4(5__j^)2 + ^r..
the case
(^'
/,
prove that
52^^^)
J.
are constants.
We
We
have in case
p=l
such a formula
we
p=2,
+ u')'^(u-u' )
,.
^(u + u'\u^-<')'^(u-u')
218]
by
TWO.
IS
331
^ (u
Though we
m. ) h- ?^ (m).
way, it will
be found that a certain interest attaches to the mode of
proof employed here.
x^ so that
i'
(a,
Xi,
"^2, X,',
Zi
are positions of
Zi, z^
V=
a^i
'^lyi
+ M^>."a +
zj^/. i
z^)
for
(ot^,
+ m''/, 2)
a/)
yi
F(xy 'P{xy
y-.
P{x,)'
F{x,)
.yi
yi
P(V)'
F{x,')
x^y^
yi
P{xiy P(x,'y
P (x) denotes
wherein
to those of 209
from
^ (m) ^'
A
Now,
if
is
(')
therefore find,
_^ {xj
x^f
By
vanishes.
considerations analogous
y^',
(a;/
an absolute constant.
if 7)1=1/
VV =
and,
we
^(u + u')^(u-u')
where
Oj) (x a^),
(x
(>;ii72
JP
(x,), etc.,
+Vi'lif (^1
but, putting y2
^''^
'""^
^'''^
= 4P (a;)
find
- ^i)'' ('^i' - -^V)^ - [('7i')i' + W2') (^'i' - ^'i) ('^2' - ('Ji'Jz' +'72'7i') (-^i' -
J5"
we
(:r),
-^'2)
-^'i)
= 4 (x - a,
) (a;
''"'
- Oj)
""'^
'^
~
(x - c)
(a;
- Ci)
(.E
^"^
'"^'
(-''2'
- -''1)]^
this leads to
^''''
" '^'^^
'
- C2), we have
(a-p)afi
16
rQx^Px-QxlPxi Qx^Px^'-Qx^'Fx^
X^-X^
X.^-Xi
'
~iXj-X^){Xi'-X2')\_
_ qx^Px^-qx;Pxi qx^Px^-Qx^Px{\
X2-X2
Xi'-Xj
J'
'
332
and
+
may
as
[218
(a,-^,)K-^2)= V^P>JK)?i'.
^'i) (2
- ^2')
CEj
{"-2-^1)
and
1
1 (liZViV^ +
^V^'i-W ~l-2
2-2-2 I'MPi^2)PW)f(^^)
1 ^ ( +
')
gi2
?l'?2"
(), j,,
thus
e have
3 (-m')
is,
- P'
The product
x-^, x.^.
(a,)
F"(i) = (i-2)=--f"(2)
\/
Now,
(aj).
thus we obtain
^ (u + m') ^ (m -
2t') a^''
putting
li!
in
which case
If in this formula
we
qi,
= [^ (0)]^
= 0,
being determined by
vanish.
we
formulae
Jl2(+" + a'"='"),
=?i2(i,),
=^
52(m|m''"
'
^H-)
190,
+ """)
gl2M)'
y
.2
Cl/.^
B (m+^'|m- ''+m"
'')3
(m-m'Im"- "+''"
")
,qiq^ _
and therefore
S2.5(tt + ')5(M-U')
^2(^)^2(V)
=1^' - 2
'2^
2-2.2-2
- g.'g2
Hg2^gl ^
q^W _ gV
of
219]
where B
(u) denotes
B (u '
"
still
^'
+ m''^'
).
IS
TWO.
result of Ex.
333
iv.
217, to give
[<Pn ()
Further
we
find,
^,
X,(M + M')
that
where
i)
m + i Q^ ,,
j;'
= ' + ^ Q,,^
iPi.
i (")
=-
,,
or
1,
= 2Xv() + 2Xi,(?0,
It should
g^
log
5 (
be noticed that
I i, ^n')
hence
form
a,-a^
where
\-'{ii)a"-{v)
(m)
o-
denotes
o-
~ ^" ^"^ +
^('*i(ii))' ^d
formula
Weierstrass's corresponding
termined so that
^" ^"^
(u)/u
for
when u =
0.
^i,
^=
;(")=-
1,
the
Hx.
5^ log
function
(See
^r
^-2
2/1
5_r
^'i
B',
is
de-
^-p{x{)P{xyP{x')P{x^)
C2,
etc.,
= {a^-a^),
and
_I
^2-^1'
A can be
yi'yz'
I
a-{u)
In
~\
y?
"I
(.r2-2)('^'2-3)J
yiy2
(m).
o-
212, 213.)
Prove that
a,),
j3= (a3 -
^^ (">
^'^ ^")
y
'
where
y2Z=2j^' (A A' + AA') + 5^' (^2'P3' + ^3^2')- ^^ (^^3 + 7^^' + AA' + A'A)
-4'fi {y^P3+yaP,' + P^P,' + P,'P,).
In this form
219.
334
This
is
[219
Thus we take
u, u'.
the function
Put
^ (m), ^ (tt
and
^(m + ?*'!''"
the zeros of
a,, x^,
A,
Aj/P
of x, is
=
I
i^^i
O'l),
i'
T/j
(a;2
ch),
i^2
- tti).
{x' - Oi),
?i' (i^/
wherein
77,
= yi/P (a;,),
etc.
Oj^, a.^^
a^i
^2
*/'.
^i'. 1
a;,'^
a;/,
the zeros of
x^
.r, Xj',
for its
other zeros
A=
to be zeros
and
fflj)
{x^
a^)
and
^(m-
as a function of
m'),
way
as poles,
P {x^) = {x^
where
(a-j),
Vi
7)^'
respectively are a, x^
m"' - )
")
such a function of
TJjXi,
%,
!,
r)^Xs,
Tj^,
X^,
'y/ajo',
v,
a^a'.
^i is
(h", Oj"
a;,,
as poles
are
and
A/P(x^), where
hence we find
=
{xi
wherein
C is an
proceed to put the right-hand side into a particular form; for this purpose
- Xi)
(Oi
X2) by Pi
and
'
absolute constant
We
write
pij
by
03, a^,
a^ in
c,
Cj,
Cj,
which
refer to the
we
branch
p^,
forjOipyjrj.y,
yi
2/2
jr'i.y,
p'i.y;
alsolet
^2
\^2 ~ ^3)
(aj
- ^j)
(12
- ^,')
(aj
- x^)
P(x) = (.r-a,)(a;-a2),
219]
335
it
we
find that
equal to
Og
^2
~ ^2f
\*^2
\'^2
~ "^1
- (*'2 - 3)
(^i'
- *V) =
- O3)
(a;/
- 3:2') =
(372
we are able
to give the
^'*2
^2
W-
^2) (^2'
W(:r2'
'*1
+ 72
{x{
'^2/
Xi
- O''
- "a) -
- 03) -
^2) (^1'
- x^)
^'2)
('V
"3).
- x^)
(.r/
03),
**'2/
V'*'!
W(:i;i'
(-^i'
**'!
'^S
U2
etc.
- ^2)
'*2
W-
^2)
(^"l'
- "2)
{^1
- "4)
i^2
- 3) ^T^'2
6)
+ (^1' - 02)
A =
(';i'72
'?l''/2')
":i
thus
- ^2) (W - 2) (^2' - 3)
Now we have, by
*'2
{-^l'
- "^s)
(^2
- 4)
(^2'
(^1
(^1'
A,
.;,
(Xi
- ai)
x^) (aV
(a;,'
+vi (^V - i)
- ^1)
(^2'
- 6)}
expansion,
- ajj)
i?i' i?2
AA there will
(i)
(.^l'
-^i) i-^2
- i) C^'i' - ^1)
(^1'
" ^1)
-
(^1'
''2) (-^1
(^2'--''l),
'^2)
'2).
("J,
- %) y (^1 - Xj)
K+
^2
- 2i),
(x^
- X,)
X,) {X,
- x^)
{,/ 2
[x^'
- a,)
{x^
(^1'
- '2)'
Consider
now the
expression
PnPiP^ +Pi2P2P'23p\i
336
V{x(- a,)
we
this expression, as
ment
''''''\^!''7''-'\w (^,'
by
see
of the Investigation,
(a^i'
+< - 2ai)
'?l
(^1
- "i)
(^2
utilising
[219
- %) -
'?2
(^2
- 2)
^1
- 3)
equal to
is
K denotes
where
V\ bi' ^ (^1'
-'?2
["/a'^
Comparing
- "i)
(< -
this
i) (^1'
^2)^
+ li
^ ('^2'
- '^i)^ + ')i'^
W-
- i)
(^2
- i)
W-
^2')^]
(^1' -^2')^]-
we
obtain the
result
PiPiP'iPP^^i
1
(^1
~ ^V
\'^\
~" *^2
V*^!
~ -^l)
("^l
~ -^2) (-^2
""
^V
('^2
~ ^2)
= {<^2-ai){"'2-a'-^{Pi3P<LPi+P'nPiPi)+Pi2P2P'iiP'it+P\iPiP22Pi:,Now we
shall therefore
put^j = J/jy,-,
jo,,,
= xy,,jy,-, y;
(p.
we
Pi Pi
334) the quotient
5 {u^- u'\u'''''')B(u-u' )
is
= 3,
m^
(?12?2?'23?'45
^/ -f
(at)
+ 9''l2?2'?23?45)when i=l
or
2,
4, 5.
Hence we have
^
.
^(m + m'!m..)^(w-m')
^(M)^2(it')
where
value of
220.
is
e is
The equation
just obtained
^ (m + m' w""") ^ g
^ (m + u')
giagag/
4-
The
221]
and limiting
be,
respectively,
337
a^, a^, a^ to
functions;
defining* the ten even functions ^i{u),'is^{a),...,'^^{'u) to he respectively
identical with the functions '^(u), ^(m), ..., %a,(u), and the six odd functimis
c,,
Cj,
c,
^02 (),
. .
\r35 Jo2^
01
J 12 "T
V
J52QV
'2
f^s
here
denotes ^(m),
This equation
"^
J 02^01^12/
J"5
CV
'^02 '^ 02
"^~
'J'o4
2^'
'J
^ CV
1
-^ai
J24^ 14^3
2
24
().
is
'^cc,(u),
^ 04^ 24^14^3
"i"
-L "2V 2Ci.' 2
~r
Jfri
04
'
an absolute constant.
may be
p=
,
1 in
,,
Sn (U
^
the form
+ U)=
^
16
-=-^(m
+ m'|w''"'');
the function
for
and
if this
^(u + u'\u'''<'')-i-'^{u
+ u') we
an expression
obtain
for
The formulae
developed at length.
are
results
Prove,
ii
jj_,.(m)
branch
etc.,
that
190)
set of formulae.
of the form
In case
(u),
wherein
places.
may
b,
^ (it|M*' ) 4- ^ (u),
6i, h denote any
"2,
the fifteen quotients are connected by thirteen algebraic
In virtue of the algebraic expression of these fifteen quotients,
they may be studied independently of the theta functions. We therefore
give below some examples of the equations connecting them.
^.,
".
+ *!!.
relations.
B.
k',
or zero,
and
positive, or zero,
than 2. In
and the elements
less
or zero.
22
338
Ex.
i.
There
importance
[221
is
in itself, in
historical interest.
_ B{u\u'''')
Let
_ B{u\u''^
'" + %""")
.9
(w|m''""+"
be three functions whose suffixes, together, involve all the five finite branch places. Then
these three functions satisfy a biquadratic relation, which, if the functions be regarded as
Cartesian coordinates in a space of three dimensions, represents a quartic surface with
sixteen nodal points.
In
fact, if jOa
5,
^'
"]
we have
+ 4(a;l-6^)(.r^-6^)(:^:^-e^)(a:^-eg)(J;^-e3)-2yly2
^W-x^f
_-t (g^-i^)(J-^2-6^)(^l-el)(a:^-e^)(.r^-e3)
where
4(^1-^2)^
it is
there
is (
211, Ex.
vii.)
yj,
{x,
x)
= iy"^, = 2/
be denoted by
if this
'
(jr),
say,
and f g-
,/, (.r,
= '^l^
j)~|
thus
where f{x^, x^) is a certain symmetrical expression of frequent occurrence (cf. 217), the
same whatever branch places 6,, b.^ may be, and A, , C are such that (a;,, x^ vanishes
V'
when for x^, x^ are put any one of the four pairs of values (fij, 62), {e^, 3), (3, e,), {e^, e^)
;
PX.cr'^Pl.a+l'Pl.c+'Vl+P'
where
X,
/i, v,
Similarly
pI <=.=^>a
But
also
it
o.-^I^'pI,
c+^'pI+p'-
thus we have
^^, etc.,
{Crelle,
xxxv. 1847,
p. 291).
This proof
is
y^__
^,
by Gopel
WHEN p
221]
Ex:
EQUAL TO TWO.
IS
Prove that
ii.
+F,
/_
=(i-C2)(Ci-a,),
(i-Ci)(a,-c-)
(Ci-ai)(ei-c)
i.
in detail.
Viv, and,
(v)
it
i.
332.)
p.
These
is
as follows
if
<= 1
Werke, vol.
=2e2iri'i(n+4')+i'n-(+9')=+2>ri9(M+9')
^=\{"), y='9oiW.
a:
{u\u'>-'')-^B{u\ are
ii.
iv.
=S
?6,6.i' 1b,h,^
(Weierstrass, Math.
so that
q^,,
?6' ?6,6,>
Ex.
'
c,)
iii.
(j3)
Pc
(c-aj) (c-
^,
^.
Ex.
339
linear function
^=^i{'), l^^^i-"),
o and
if a, b,
cx+dy-az ht
c,
d denote the
be denoted by
series,
(c,
values of x, y,
d,
-a, ~b),
that
ei{V) = {c,d,-a,-h),
e,HV) = {a,b,c,d)
e^^ {V)
= {b,-a,-d,c),
z, t
etc.,
e^^V)=={a,b,-c,-d).
Relations of this character are actually obtained by Gopel, in this way. It will be
the purpose of introducing the subject of a subsequent chapter, if the
method of obtaining one of these relations be explained here. The general term of the
suflScient, for
series
e^
F)
is
_
where q'--i(l,
0),
q=i (1,
0),
and
220)
7riD( + i)')+}iTT(+')2+27ri?(n+j')
namely
is
_grf[i(i+i)+jj] + i['-(n,+J)'+2T,j(n,+J)(nj)+Tn52)+i.r(n,+i).
e2^
V)
is
niL, where
is
equal to
JriK(/ + m/) + m, + m, + l;
222
340
[221
there are therefore four kinds of terms in the product according to the evenness or
oddness of the two integers n^+nii, ii^+m^. Consider only one kind, namely when
+ m^
jij
+ mj,
if
now we put
n^
thus, to
?ij, 7i2)
71;+ Mil, W2 + 2
2 such that
^/^ ) )
iirr,,
is
is
equal to
'"ij
then
5^0; ^ T^")
>
^''
that
is,
N^, M^,
iVj,
^^^
^ven
of 5oj
M^
therefore, as
{v),
and
Jf,2+2TrT,2lf Jfj+Tjjifa*
,
V that
is,
of ^5
(ij),
and e^''<^'+^'= -
1, it
follows that
which are of the kind under consideration consist of all the terms of
ay. It can similarly be seen that the three other sorts of
terms, when 9ij + )?ij is even and n.^ + m^ odd, when n^-\-m^ is odd and i2+2 '^^ '*'" ^^en,
are, in their aggregate the terms of the sum hx + dz-ct.
We can
^5
1')
5qj {v), or
also, in
eo3e23e3(r)e,4(F)+ee2eo2(F)es(K)=e,2eo,e,(F) 634(10,
M), e^^ = 2 (ad + 6c), 62^ = 2 {ah -
e2
0(13
denoting 603
= a2-62-c2 +
rf2,
0,2^
cd), Q^i
(0), etc.
= 2 (a6 + cd),
= 02- 62 + c2-rf2,
(6,
form
c, b,a)
-c){a,h,c,d)
-a, -d,
c) (a, 6,
-c, -d).
p.
215.
t.
IX.
is
(1893); Schottky, Crelle, cv. pp. 233, 269; Wirtinger, Untersuchungen iiber Tketa-
The
obvious,
rationalised form of the equation, from which the presence of the sixteen nodes is
is obtained in chapter XV. of the present volume.
WHEN p
221]
Ex.
EQUAL TO TWO.
IS
341
Obtain the following relations, connecting the ratios of the values of the even
when p = 2. They may be obtained from
V.
the relations
212)
(6
by substituting
- a;,)
{b
- X2) = Ve'-'-P^/'
3*
S*
3"
Ci
S*C3
and
-.3*
3*
aiCj
(! - aj)
Caj
(c-C2)fe-2)K-c)-(l-l)
(Ci-C2)(c2-a2)(a2-Ci).(ai-c)
") H- 5^ (u)
ajC^
ajCi
3*
-.3*
ca2
ca,
-a^)
(c,
u'''
-.3*
S*
a|Ci
{c2-c){c-ai){ai-c^).
x^.
(ai-a^){a2-Ci){Ci-ai).{c^-c)
(b) S^ (u
(c
- cj (Cj - tti)
(oti
- c)
(c,^
a^)
(c-Cl)(0l-O2)K-c)-(2-l)
:
Infer that
We have proved
in fact, as follows
+ Voi - ^2
^';. ('')
a.,
(m)
3ca,3ca,3a, (m) 3a,c, (u) +3a,c,Sca,3a, (u) 3a,c, (u) =3c3c,3c, {u) 3a,a, (u).
Obtain formulae to express the ratios of the differential coefficients of the odd
theta functions for zero values of the arguments.
Ex.
vi.
Ex.
vii.
3()
Prove that
wherein
are
any two
finite
t is
one way
=y.,
".
+ =="."=,
= A<>S(m;
m + m''"
"=,
and
j;
= M^" +
*'
'"''',
we
find,
Writing
as follows:
is
P), that
Aiog^("l"'':r+"'"''^=Z;-^'+C2(''-^^"'')-<r2(''-"^"'').
Bin)
*2
and, by the formula expressmg fi(M
'
.!.
'-
'
b,-b^r
-5
z,
- ^2
L(^i
Si
- *i)
(^1
- ^2)
(^2
- *')
(^2
'"
it^."
is
ji'^i
"1
equal to
"I
- ^-2)
where s
'
One
z,
"-H^'
when 6
h,
a2, is expressible
('')
3 ^04 () - ^4 (")
^'
^""^
by Weierstrass's
342
[221
+ 5j
where
^4, 3^
arguments
Thus we
denote
v^,
obtain, eventually,
^5 ()
5' (M)
= g-
S^ (u)
.9
(u)
may
write
(v)
(u + v)
= 3{u)+
(v^
if,
\-
herein, the
v^
^j 3
(u).
first differential
coefficients of the
where
5o4 (v)
5^ (0),
.94 (0),
(i))
5'4 ()
5(j),
- 3o4 ()
3'i
W = f M 5o
^4
()
'4^0
+^|^502
^4-^0
W M
(),
3 = 5(0), 3'=3'{0).
=0
r5^S(w|tt""" + M'"=')1
J"=
L3%
and the
or 3'q4=0,
In the cases
argument
here
result of the general formula agrees with the formula of this example.
p>2 we have even theta functions vanishing for zero values of the
we have one of the differential coefficients of an odd function vanishing
argument.
Beside the
Note.
references
given
in
"On
this
chapter there
is
a paper
by Bolza,
and second derivatives of hyperelliptic (r-functions " (see Acta Math. xx. (Feb. 1896), p. 1
"Zur Lehre von den hyperelliptischen Integralen, von Paul Epstein"), which was overlooked tiU the chapter was
completed. The fundamental formula of Klein, utilised by Bolza, is developed, in
what appeared to be its proper place, in chapter XIV. of the present volume. See also
American Journal,
xvii.
11
(1895),
the
first
Wiltheiss, Crelle, xcix. p. 247, Math. Annal. xxxi. p. 417; Brioschi, Rend. d. Ace. dei
Lincei,
Crelle,
Frobenius,
To the note on
Functnen. in einer
To the note on
chapter
viii.,
p.
zweibliitt. Fliiche
(Zurich, 1866), p. 12
functions of three variables are given by Cayley and Borchardt, Crelle, Lxxxvii. (1878).
*
223]
CHAPTER
A
222.
XII.
theory
is
resumed from
this point of
But there
it is
the
is
historically a
shewn that on a
development*
surface which
is
in
many
Riemann
The suggestion
is
importance
for
any Riemann
surface.
Let
a, yS, 7,
S be
r-4"^
* Beferred to
?-A'
344
wherein
f=
fi
(a^
is
/a
and
real,
be
less
We
In
than unity.
the roots of
introduce
now the
all
assumed that B,
this chapter it is
unity.
are
+ fi)/{y^ + B),
shall
J5,
[223
/x,
is
not equal to
(6),
Fig. 6.
and a point C" be taken between A, B in such a way that 1 > AC'/CB >fi,
but otherwise arbitrarily then the locus of a point P such that APjPB
= AC'/C'B is a circle. Take now a point C also between A and B, such that
;
CB/AC= fj,C'B/AC,
for
which
will
lie
a,nd
mark the
P'B/AP'^CB/AC
circle
since
which
P'B/AP'
points A,
B cut
the
first circle,
which we
is
is less
If
now any
circle
circle
through the
C,
in the points
g-,
the argument of
-P
^-B
C in
PP, about
AP^
sum
circle
about
in a counter
r-P
r-^~
fi-P
r,-^~'^
K-B
K-A-
C into a
2-4]
345
transformed point
is
the
?'
'^Ti
^^^'ein OiSi
'
- A7i =
(i
1,
= 1,
2,
. .
. ,
p),
or also by
assume that
by
by
S,
let
2p
circles
all
obtained, each is
the circles be denoted
^i-S.
formation
a^i,
the circle
(7/ will
%Gj.
within Gj will be
a region, %^jS,
bounded by the
circle
Ijdng within
they cannot
^,(7^
circles lying
circles lying in
lie
without this
circle,
namely
in
'itiS,
because
the picture of a space, S, whose only boundaries are the 2p fundamental circles Gi, O/, ..., Cp, Gp. Proceeding in the manner thus indicated
we shall obtain by induction the result enunciated in the following statement,
^iS
is
wherein
%, and
into Gi-. Let all possible multiples of powers o/^i, S-j ,...,&,,,% be formed,
the corresponding regions, obtained by applying to S the transformations
and
*
ci.
(1887),
p. 227,
and
346
[224
224]
3-
D
y DD
D
DD
DP
347
348
[224
^=
^?^?^?.
formed by
+ n + rj
and let ^
m^k is formed by
factors,
+?>,
=h
be
the
ofh+1,
product
not
1,
the inverse of
Then
factors.
'^t,
so that
the region
mS
m^S.
Thus, the region ^j/S entirely surrounds the space '^^jS, and the latter
surrounds ^i^/>S, or ^i^j'^^/S; but ^j/S
reader
may
is
surrounded by
'iti'itr^S
or S.
drawn
<f>,
in place of circles,
is
The
(7),
and
taken.
consisting
substitutions
. . .
% gives
a single covering of the whole plane, every point being as nearly reached
as we desire, by taking a sufficient number of factors, and no point being
rise to
There are in
225.
number
of
K = a^y
...
is,
lying within aS, in fact lying within a^S, nay more, lying within affyS, and
number
of
There is an
which the corresponding
them
for
agree to a suflScient
number
none of them
lie
of the left-hand
The most
are obtained from iS with a finite number of substitutions.
important of these singular points are those for which the con-esponding
226]
by
349
we have
226.
If ^,
wherein aS
and
if this
<f>
^7 = 1 = AD BC,
is
given by
5"
notation of matrices
(a'
I
and
a'h'
/3'
S'
!
\
/3
h\
We
^ B
C d\
),
suppose
all
possible substitu-
f'
factor, ^i or
%%,
1)
%'itj-\ ^r'^j,
whose products
so on.
{2piy
substitutions
series
2 (mod 7)"*,
wherein
jfc
is
r = +i8)/(7?+ 5)
substitution to
and the
its
will
set of
contain
Let now
%^
350
then by the formula just set down 7n+i = 7jit+ S7i, where, if
f = (aif+ A)/(7i?+8i). be put in the form Q;' - Bi)j{J;' - Ai)
factors;
or
^i,
[226
= Pii^-Bi)l(^-Ai), we
have
ft,
t,
^i
7i.
respectively equal to
Aip'i
Bi-Ai
BiPi
Bip
Bi-Ai
Pi _
Bi-Ai'
Pi)
AiBi(pi
AjPi
pi
Bi-Ai~
'
'
'
'i
'
less or
may
pt
^p or one of ^f^,
...,
^p^
@n^S
If
now
lies entirely
^r
is
..., ^j,,
or
... ,'drp
),
lies
Now,
of
of n
...,
fii^,
i/r(^^~^,
...
according as ^j
^~
is
/>,
where ^,
%S ( 224)
lies
SJyn
...,
S/7
is
one of the
or coincides with
lie
lie
one of
is
one of
is
when
Cp
can only
one
= %<f>-^tfr-'...,
can only
is
accordingly either
is
if
^f
or
^1, ...,S-j,;
/li
8/7,
'^i
Bi-Ai
and
@=
^,, ...,^~\
it;
one of &,,
Pi
the modulus of
we have
these formulae
be either
2p fundamental substitutions
the region
By
-Pi- Bi-Ai
of ^1,
p,-
= i,
must be one
factors,
and not
n+1
factors
in
7n
not of infinitely small modulus; if, however, the modulus of pi be
greater than unity, and r = i, ^, must be ^j, namely one of ^,
Sj,, and
in that case the modulus of .4 j + S/7 is not infinitely small.
Thus, according
is
as
I
Pi
1,
we may put
I
where
X,
is
Bi
+ S/7 > \,
I
Ai
+ S/7 > X,
is
distance of Bi, Ai, respectively, from the nearest point of the circle within
which
8/7
lies.
227]
351
we have
or
mod
(7^+1/7^^)
<"
where o- is a positive finite quantity, for which an arbitrary lower limit may
be assigned independent of the substitutions of which is compounded, and
independent o{ n, provided the moduli fi^, ..., fj,p be supposed sufficiently small,
and the p pairs of circles be sufficiently distant from one another.
Ex.
and the
is
is
as great as possible
least of the
tive,
circles are
We
(7'
quantities
the series
2 mod 7"*
7o-[2;,
then
if 7
be the
than
+ MiTl) + M^ll)V
],
and therefore certainly convergent if <7* > 2^ 1, which, as shewn above, may
be supposed, fi^, ..., fip being sufiiciently small.
227.
following inference:
Let
a-,,
...,
a-j,
be
compound
the
fundamental substitutions
%,
...,
let
w(;-||) +)-'
converges absolutely and uniformly * for all positions of f other than (i) the
singular points of the group, and the points f = S/7, namely the points
derivable from f = 00 by the substitutions of the group, including the point
5"= 00
itself, (ii)
continuous function of if for all the values of ^ other than the excepted ones.
The function will have poles at the poles of f{^) and the points thence
derived by the substitutions of the group it may have essential singularities
;
/((a?+/3)/(7r +
* In regard to
f; for the convergence
))-
f.
352
Denote
this function
identical substitution
associated with ^
or,
the function
is
by
all
;
by F(^)
denote
^^o denotes
all
recognising this
ao,
[227
we immediately
fact,
find
So)*
when the
variable
we introduce a
whereof the
full justification
representation
is
Let a Riemann surface be taken, on which the 2p period loops are cut let
the circumference of the circle Ci of the f plane be associated with one side
;
of the circle
(7/
two sides of
this curve
P' of the
circle Ci,
and
let
the
the Riemann surface. Then the function F(X)jG{^), which has the
same value at any two near points on opposite sides of the barrier, and
has the same value at any point Q of the circle Ci as at the corresponding
point Q' of the circle Ci, will correspond to a function uniform on the
undissected Riemann surface.
In this representation the whole of the
Riemann surface corresponds to the region 8 any region ^jiS corresponds to
(oj) of
F{^/G{^) be
thus
if
function on
are
Riemann
the
It will
the values
Ui, Ui
for instance, of
2mi^r, Ui
+ 2a>'i^r,
an integral of the
first
Ui+2a)i^r+^^'i,r may be
+ 4o),-_ ,
kind,
repre-
Riemann
surface.
These remarks
will
in
353
the analogue of f by
228]
The
fj
Bi)
function
has essential singularities at the singular points of the group, and has poles
at the places f=a, f=oo and at the analogues of these places.
Let the
points 00 a be joined by an arbitrary barrier lying in S, and the analogues of
,
this barrier
this
uniformly convergent
Slogl:^, = n^-^
is
competent
when f
to represent a function of
,say,
log(f
a),
oo
is
and
its
and at f =
00
like
logfp))
as
it is infinite
at
^=a
n^-f-n^^ =2iog%^::^ =
a,
siog|^^^^=^
+ 2iog^'^^*-2iog|^^*.
<x>
where
5'i,,
is
side
is
independent of
f,
being equal to
11
and
^'J; in order
to prove this in another way, and obtain at the same time a result which
will subsequently be useful, we introduce an abbreviated notation; denote
simply by the letter r; then if j be in turn every subwhose product symbol has not a positive or negative
n at its right-hand end, all the substitutions of the
substitution
power of the
h being in turn equal to all positive and negative
symbol
jw*
group have the
the substitution
hence
2[log(fi-a)-log(?i-a)],
is
equal to
21og^><^-^)^.
23
354
M = ?i-^
where iV = ?i",
independent of ^
and
but, in fact, ^^
if
we
is
[228
, and f
is
A^; thus
n^^'j is
where ^i
all
right-hand end,
we have
If for abbreviation
Ex.
we put
prove that
C, c
a, 00
being an arbitrary
o"
'*>*'
11
a
',
transformation applied to
nt,6 = 21og
r,
p$,
where
^~^n
ijoint.
229.
ri,
the variables,
all
is
we have
.f-^r'(a)/?-^rH6).
also
^'Ml^ll^,
Thus we have
where
a
a, 6
ft
_,
27rt
hence also
^"
where
27ri''?.
r,
= i~^,
-2,rtr
^Ui-?/
?.
230]
355
= st,
-.
that
where
^p',
is,
f(,
for,
^,i
= ^s^(, we
have
so that
=V,
Vst
and in
formula
particular,
when
+Vt
st is
we
see
by the
itself,
'^J' ...,
we obtain
=\iVi
. . .
Vr
d''
+\V,,
functions of
v\'^, ..., Vp
"
.
230.
''"
2iri
But
is
'""^
'
not an uniform function of f for let ^ describe the circumference of the circle (7 in a counter clockwise direction then, by the factor
group.
it is
^ Bn,
v,l
" increases
by unity
(?,
constituted by
for,
of
when
the
the* form
^^/S, contains a point ^j(5), the product representing the substitution J has
a positive power of ^ as its left-hand factor, and in that case the region
Similarly
clockwise direction,
7;^"
increases
",
'^j
by
if
But
unity.
circles,
if
no increase
(7'
in a
S by
P of G'
to
"
the corresponding point P' of C. Then v^ is greater by unity on one side
denote any one of
if
Further
of this barrier than on the other side.
the substitutions ^ .... ^p, we have
C
*
Where
-^n
=\ -\ = \
-"f^.f
\^
is
232
356
where ^
is
j,
lt
C',
barriers,
all
and
T^m,
indepen-
Ji-
connecting the
4'
4'
is also
paii-s
of circles
The behaviour
the difference
Vn= T, m
V^
[230
"
...,
is
therefore
upon a Riemann
surface, the correspondence of the pair of circumferences C', (7,/ and the two
sides of the barrier P'P, to the two sides of the period loops (6). (), on the
Riemann normal
integrals
Riemann
is
considering the
(a,i)
Riemann
surface.
On
Riemann
the
itself.
231.
We may
t,i,
formula
'=2li^'"g
let
power of
thus we
may
write ^^
* Barriers
= ^^''^t.
or, for
232]
357
is
= s^. 2
T, ,
and
therefore, if
log
m = n,
when
^'""~f" / ^-"'~;f'
be not equal to
n,
^^r'(^m) - Bj ^];\A,n)
27ri t
while
separating
for
- A
which k
the identical
is
substitution,
+ J_ 2'
log
27ri *
A*"(-Sn)-i?n ^;^(^)-^ \
/
W'(^) - B ^r\An) - A J
T
T,
where
= 2^
lo? Ce''"^
log
(Me ;
s denotes
2^.
'
^"
2' loff
log l
1^^^^
~ ^* ^^"^
_ ^^ ^^^^
/
/
== 1,
is
not included
-^"-^(^n)
^^ _ ^^ ^^^^^^
",
are drawn.
Ex.
232.
i.
If v "
as in the case of
= tt + iw
a Riemann
surface,
by taking the
integral
judw round
the
",
prove,
closed
curves each formed by the circumferences of a pair of circles and the two sides of the
barrier joining them, that the imaginary partof
JV^, ...,
Np
i\''i2r
being any
, by contour integration.
real quantities
?jj''
"
+ 2#,jy2''i2+
+ iV^/' ".
is positive,
result T^ =T-,^
Ex.
ii.
has analogous properties to Riemann's normal elementary integral of the second kind.
Ex.
iii.
Prove that
rf'.^=(y*a+8i)^rf-^
358
Ex.
iv.
[232
yrZ +
KY
prove that
* (2,
U -*
(2,
^),
* (Cp,
1
0,
(Cp, I),
1
* (Cp,
1
f l),
,
...,
Cp,
f be any arbitrary
(Cp, fp)
1
unchanged by the substitutions of the group, and has simple poles at z, Cj, ... Cp, and
and a simple zero at ^, and its analogues. Thus the function is similar to
the function y^r{x,a; , q ..., c^) of 122, and every function which is unchanged by the
substitutions of the group can be expressed by means of it.
is
their analogues,
dz
di
(of Ex.
Ex.
V.
Ex.
vi.
iii.)
Prove that
In casep=l, we have
r=-.log(,0,
where
{a,.
prove,
and
-(f')-^B/!^>(^)
and shew that G
Ex.
vii.
(f) is
Taking two
a>,
6 (v'' ", 5
i, ^).
'
= t- log
^=
y.,
and
let
233]
359
denote Weierstrass's
f) (m)
if a, c
(a)
is
first order,
(j3)
is real
circle,
The
Q can be
Ex. viii. Prove that any two uniform ftinotions of f having no discontinuities except
which are unaltered by the substitutions of the group, are connected by an algebraic
and that, if these two be properly chosen, any other uniform function
relation (cf. 235)
of f having no discontinuities except poles, which is unaltered by the substitutions of the
group, can be expressed rationally in terms of them. The development of the theory on
these lines is identical with the theory of rational functions on a Riemann surface, but
Thus for instance we have a
is simpler on account of the absence of branch places.
theory of fundamental integral functions, an integral fimction being one which is only
And we can form a function such
infinite in the poles of an arbitrarily chosen function x.
as 5 {x, z) ( 124, Chap. VII.) ; but the essential part of that function is much more
poles,
233.
The preceding
07
(f, y),
which
is
{?,
y/zi,
C;}
let
n,'
,
log
(^-r^yK^t) ^ 7
that, in the
^''^
(^.-
- 7) ia - ?)
'
(^
- 7i) (0 - f.)
360
and
its inverse,
only one
is
[233
to be taken,
is
Thus we have*
excluded.
_w[-(.-0(-r)-"<:;i'=.f-7)n-|E|^;.
-(f-7)n'ir,7/7<,fi|.
i
where
11'
ratio.
{?,
7/71,
fij
denotes a cross
^(?,7). =(r-7)n'{r,7/7i.
W;
it
it
it depends only on
function of f and y, single-valued on the whole ^-p\a,ne
of ?"=<;
vanishes only at
it
the factor a (u
a)
It should
(f, 7) does not depend upon the choice we make in the product between
any substitution and its inverse this follows by applying the substitution
5T
S r^
to every
element of any
We
234.
enquire
now
factor.
It will
!)",
7) under
7"?'+^n
=^(?. 7)
where (
(f,
be proved that
signs to be explained.
This result can be obtained, save for a sign, from the definition of
as a limit, from the function 11
ct (f, 7),
a pair of
*
circles,
ci.
(1887), p.
242
(at
Vi'", ...,Vp
See
234]
which
are definite
t)"'
so therefore
t,
is also
361
given in
e'^'^'f'
231,
and
equal to
is
Bn-{B,,\ IB^-(A^),
fiie^^'W
."^n
\i>n.)s I
-^
(-4,,);
not definite.
is
But
in fact, as it arises in
when
y' ~i~
w -An
T / >,
An/
)
f
is
has reached f from ^by a path which does not cross the barriers. Thus k
perfectly definite when the barriers are drawn, and the sign of the
quantity
Bn - {B\
An-{Bn)s
-"", ^^eit'nll'
is
Bn
{An)s
An
(An)s
perfectly definite
(- 1)~\
PF'.
The annexed
In the
first
Fig. 8.
is
ic
greater than
2Tr.
When
/c
is
thus
formula pi = (1 +?)/(+
S),
also
is
definite;
it
may be
= /iei^''
or
4 = - mI e^"-
We
shall
We
have
^{^,y)
?-7
fi-7
'
7.-? "?-?'
362
where
denotes in turn
all
X= 1,
for
2, 3, ..., 00
?"
~ 'y
and
h,
k take
^"^+^
~y
7^
denotes
positive
all
~ ?
A, s,
substitutions
all
where
has the
?^~ ?
A ?ji*im*~"7
yrJ'ifn''
- 7 jj ?^-? jj yr^^n
fn^'+l - ? X 7'' - ?
?n^
7
j-r
{"nhiks
= ^~7 jj
?-7
Therefore
thus
g^ (?. 'y)
[234
?''>^+'
A.
{ An)n>'s
7''ot'^
rr
Cn^+l
S'n^+1
fn
bn*m* + l
~ ?
g"
Wwit +
~b
the transformation of the second part of the product being precisely as in the
first
part,
?"- 7
'
?n
-7
'
-^n- f
7
(-4n)ns 7
TT (Jjnjnhs
A.s
^n^ K-y
7-
A)^s
(-
(D)nfts
y - An'
'
^n->^
^
f
^n
^n\->^
7 ^n-t-'t-* ?n ~ ?-^*-1m-*
7 S'n-tm-'' ? ?n-*ir-i>ii- A
, *
_.
ft,
ft.
(^n)7iA
'
(-B)A,
n 7 (-Bn)-l-I
i7 (-4)-*s-l
since h
and
7)
U-(-4n)n-'^s-l,
may
'
5'n (5nXi-*s-l
replace
khyh,
second product,
ft,
ji
, ft
7 ~(-Bn)*8-l
7 ~ (-^n)?!**-!
(-4n)n>s-l
? (Bn)nAs-l
?
'
-^ S log ^ZM^)
y^
^_
y-An^
n'^s.
7_-
n^s-\ or be
^-(BnUs y - {AnUs
V^n)
the identical substitution,
n ^-(BnU-r y-(A.,U ~i
235]
363
therefore
^ (?, 7)
__
- ^
?>i
.,
-4n jj
- (if^u-,
(-4)g
?:-
(^)8
(4u,-,
-4
and hence
or
ST (?.
now from
"'
'^''^" ^
the
values ot a,
immediately find
7f + 2
thus, as ^*
r-^
7)
= (-
[p;* (f -
Ifnfj^ e*"",
^)
- pi (f - 5)]/(5 - ^)
the
we have
hence, finally
where (-l/'e'^Vne*"
(-1)*7, (-
1)*''8
The
235.
is
immediately from
convergent,
may be regarded
''^'^
fundamental.
as
is
expressible
by
la-
(f, 7),
in the form
'^^(.,7)..(r,c)'
by the
definition Vn"*
= -^.Y\T''
and
(cf.
Ex.
value
ties
is
ii,
232).
Further,
if
except poles,
it is
364
a Riemann surface,
[235
;Sf,
Vi'''
+?;/'"*
is
a constant for
all
= m; +
...,
nip'
values of
+mp'Ti^p,
w,'t;,i+
And
f.
Ex.
In the
elliptic case
tively
the
for
(f;
/3i,
may be
...,p),
1,
it
...,
is
easy to
a^ in this
absorbed.
v^
integrals
Thence
replacing some of
(i
v^,
we
immediately find
(-yi
-yt
7-yi
_B-A
_-^^
p2*
fi_ \1 -Spaces
-()+
2p^cos 27r
2;r (m
y)+p^
( C-^i_
f
\i-~P)
y-Qi
'*
^niniriu-v)
smTTMsmjri'
and hence
,^
5 .4
^""'^
which*, putting
e'"''=p', is
2i
sinTrfM-?))
2p'co.s27r(? i')-|-a2'
sinTTMsinwv,"
(l-p*)^
equal to
(^r^J'^e-2'^<-"')V[2<o(M-J));
where m
is
2o,,
26,r]
- sin ,r
sin
,71-,
12
TO
n**
(l-p'')^'
may be
We
The
excess of the
number
f,
vol.
i.
p. 400.
236]
= dvn^jd^,
function ^n{K),
number
which
nowhere
is
2in
we
If
365
infinite, in
is
I rr^(ir)_?i^(?)i^>.
a counter-clockwise direction.
of
in
is
or 2p
denominator
we may
df,
say that
"*
dv,[
has 2p
oo in
virtue of the
The theory
there
will,
@ (yf-
function
"^
vfi
in
..., trip
"!
Riemann
we take
and
surface
in the
i)ip,m-^
^re
the places
fi,
...,
And
f^.
in the region S, which, repeated, constitute the zero of a differential dv^satisfy the equations typified
^,
^,
and
by
The
vary
when the
positions
of mj, ...,mp
may
barriers
changed.
But
it
is
by
;>
the
barriers,
(c^-,),
will
when ?
is
of which
(c,)
/S
so
EXPRESSION OF
366
m (f,
BY THETA FUNCTIONS.
7)
any
(ci),
. .
unnecessary
(c^-i) are
. ,
required
is
vi^,
in
fact
the
number
and poles of
But
S which
zeros
[236
Vp^, whereof w
...,
among
the logarithmic
is
...,
Vp^
is
unaltered
circles,
barriers
Vi'^ , ...,
is
the region S, save for an arbitrary constant multiplier, provided the sign of
if
7 be any
\J
-yz-fJ t-
is
And,
this being
We
237.
proceed
now
'
where
s'v
'
= s/v,
''
+ Sp'vp
7.
and zeros
poles
Til,
np^i
00
7,
while
ii,
For
..., ?ip_,,
nr (f,
is
two sides of
in the region S.
and
''
and
(t)^'^
this equation
+ ^fls,^.)
vanishes to the
yfr{^) vanishes to
the
first
first
order at
order at f = 7, and is
Thus the quotient of the two sides of the
7) vanishes to the
first
equation has no infinities within the region S. Further the square of this
quotient is uniform within the region S, independently of the barriers; for
this statement holds of each of the factors
And, if f be replaced by f, the square of the quotient of the two sides of the
equation becomes (cf 175, Chap. X.) multiplied by the factor
which
is
equal to unity.
Nowf
At the analogues
U+iK
of
and
a function of
is
f = cc neither sr (f,
7)
nor
1/
(f )
becomes
infinite.
If
238]
367
(^, 7).
infinities,
From
it
we
(ii)
^ (f
1)',
The constant
of the form (
!)*'.
chosen so that
A = l.
For
expression
1=1
= d(u)/dui.
when f
near to 7,
is
the value (? 7)
238.
which
Then
(cf.
A {^y) + ...,
equal to
4-
The developments
will
discussed, render
The
is
it
unnecessary
for
fip
/j.^
wherein
The
function
is
X (?,
^^
S,
and
denotes a
M)
= (7n? +
S)
\ (r,
'^^^'^^^
fin)
(jnfl
MX
(?, ^),
!..^-H/3.-?(7.M+"S7)
__V
~
\ (?,
m),
Sfr-/3i-/^(-7? + ai)
-i
c,
Qr-m
X(r,M)
to that
-X(/.,
denoted by
Thus
i.
f).
and no other
/*,
00
as
we assume,
368
order at f =
first
number
of
00
The
poles in S, which
its
[238
(f, 7).
is
number
excess of the
given by
-.if
n=iJ
rfr.
27ri
where the dash denotes a differentiation in regard to f and the integrals are
taken counter-clockwise round the circles C,', ..., Cp, namely by
,
equal to p.
these by
/i,,
>
..., fip.
such region
the result
is
found to be
1.
/(r)=[^(r,/^)p^|.
where v
is
Then /(f)
zeros of dv.
v{'^
let ?,,..., ^^^^ denote the
Vp"*
uniform within the region S, and is unaltered by
is
fi^^,
It has poles
...
Hp-,
"
Riemann
Thus we have,
+ W^ '"' + "''' +
precisely as for
surface,
{fJi.=
\,2,...,p),
or ( 179, p. 256),
(M
and therefore,
such that
have
if
j?i,,
r..
..., nip
(/''_*'_
182,
p.
fi,
( 236),
^ = Xp we
265).
(/^,^
When
...,r^-2)H(M,=, ...,^/),
...,/x/)s(<, ...,,V).
Now
= i (^i + -t/T^, +
1
+kpTi^p),
(t=l,
2,
...,J3),
being integers.
zeros are
/i-i,
..., fip.
It is
W-'t'vH-^
\
n)
^)
(f,
or (f,
=(^
l)-'.+'-6-^-(''''+4^.)
'
238]
369
wherein
ki
for
A;'/""
AVt,-,
denotes
v4"''+
+ V^'". and 11 denotes the p
+ V'^i.p. ^^' within S, no zeros or poles, and is
quantities
such that,
a fundamental substitution,
i?'(r)/i?'(o=(-i)^-+*"-*
175, Chap. X.)
(cf.
thus, also, Qn
+ hn kn
X
(f,
is
an even integer,
^)
^ (?,
^)
= 4e -
= 2S,
say,
F{^
is
a constant
and we have
(/ " - \P),
'^'*^'''^'
"
...,k^ are
A;/,
wherein
is
is
all
even
put kr ='iHr
then, as
we have
independent of
f,
is
also
be expressed by putting f =
(?,
^)
^ (r,
/i
A^)
independent of
thence we obtain,
=@
(/''- i5'
n.
The value
of
%,
fi.
B may
finally,
first
3T (f, /i),
the
fundamental
will
circles,
and
The
function
unchanged, if the
Tn,m, and therefore the integers g^, ..., gp, will remain
by unity.
increased
be
will
integer
the
But
barriers remain unaltered.
unaltered,
on the other hand, the coefficients a, ^, y, S remaining
equation
the
of
side
left-hand
one of the barriers be drawn differently, the
by
increased
be
will
hp
remains unaltered; on the right-hand one of A ...,
of
each
therefore
and
increased by unity,
an integer, say, for example,
If,
T, '
...,
Tr
also increased
by unity.
Putting u
for
v^''^
-^g- y>-,
'
24
and
370
theta series
Ex.
an even multiple of
is
( ^)
(f,
Ex.
occur,
and
In case
ii.
{f , fj
is
unchanged.
refers to
all
when any
^ -
denotes |
fi, /ij}
term
must not
by
+ 17m/,
n,.
iri,
i.
ii,
is
2-7ri
which
[238
substitution
is
^
C'
'^
-S-^ amn(u-v)
2t
sin^7r(M-i;)
i^'
sinjrMsiuTrt)'
"
''' '^i
l-2p*cos27r(M-j;)+p2*'
''
and hence
,
,^
2isin7rMsin7rt>
fi
^^''^'~{B-A)siDw{u-v)\_
When A=0
this
S 4(
*",-.,
n-
(m y) "]
!;)
nu sin irv
(^-^)
0-3 [2<i)
J"
(u 1))]
<r[2o.(M-j))]
770-3(0)
Thus
+ p') sin^
becomes*
4i'm sin
quantity.
l)'''pi' (1
l-2p'cos2jr(M + p2'
2<i),
'
2aT as periods,
being an arbitrary
>
Ex.)
( 235,
'^
<^3(0)
5(0)
'
(i)
When ^ = 1 we
similarly find
^{C, )
= ^/"5'^sinOT .^sCM^irZ+i)]
{B-A)n^^{a)
o-[2a.(M-t>)]
-2,.(-)
^
and hence
=f (f. f-)'^{<L,
;/1
2e
'
e(0)
=l+2?cos(2ffK)+22^co3(4ffM) + 2j9cos(6,rw) +
where g = e"".
* Cf.
'
i.
p. 422.
238]
Ex.
a- (f,
371
Denoting
iii.
identical sub-
^,
y.)
^a:. IV.
If 2, s be two single-valued functions of
f, without essential singularities
which are unaltered by the substitutions of the group, the algebraic
relation connecting
z and s may be associated with a Riemann
surface, whereon ( is an infinitely valued
function
and if ^, s be properly chosen, any single-valued function of without
essential
f
singularities, which is unaltered by the substitutions of the
group, is a rational function on
the Riemann surface. But if
;
where
C'=-j-^, etc.,
we immediately
therefore, as
{f,
z},
singularities,
and
is
-{z, C}
unaltered
is
{^J
by the
V^ -^]
z}
J'
if
be chosen so that F^
yjf
(f),
the function
Ctf
-^
dy
\/ rn-,
fi
p.
i? is
(fj /*)>
Ex.
Cf.
336
in
be rational functions.
P= --rlog-j-,
is
Riemann
Thus
may
surface,' the
for instance
take for
we may
the function
considered in
viii.
is
where
di
/dz
= A /j^
Therefore, if
Idz
/ -r;
s),
+pf.,[z.,jP..,f]r=o,
take for
= 2I{z,
we have
and
{C,
P = - -^log
= {(,z};
for
-^ + R. Y=0,
238 f.
232.
Riemann,
Ges.
Werke
f.
242
372
[238
Ex. V. If, as in Ex. iv., we suppose a Eiemann surface constructed such that to
every point f of the f plane there corresponds a place (2, s) of the Riemann surface, and
in particular to the jjoint f=^ there corresponds the place {x, y), and if R, S be functions
of ^ defined
by the expansions
^l0g^(f,|)=-^4^+i^+(3-^)iJ+
^^=l-i.S(f-f)^+
prove that
S are
and that R,
rational functions of
x and
y.
given any
last two examples suggest a problem of capital importance
Riemann surface, to find a function f, which will effect a conformal representation of the
surface to such a f-region as that here discussed. This problem may be regarded as that
Ex.
The
vi.
I {z,
s).
consult
Schottky, Crelle, Lxxxiii. (1877), p. 336, and Crelle, ci. (1887), p. 268, and Poinoard,
Acta Mathematica, iv. (1884), p. 224, and Bulletin de la Soc. Math, de France, t. xi. (18 May,
In the
1883), p. 112.
elliptic case,
where
*/
-jT
taking
= K2i-'^f^J)'=^'
and f
a/
and t as
There
239.
it is
dV
ri^V
is
may be referred to in
one another, which are all
Take p
conclusion.
periods,
C.
We
obtain
shall
circles C/,
at right angles.
The
C^',
...,
circle
case referred to
Cp are
all interior
-4,
It can
to the circle C.
be
C as well as in regard to the circle (7, and further that if &> denote
the process of inversion in regard to the circle C and < that of inversion in
the circle
regard to
<B^
(7,
= S-,;;^<B.
called an,
W,
Hence
if
is (ua), so
that w^<b
= '^~^,
(7
points of intersection of 0,
be called
a, h, it
may be shewn
without
much
difficulty that
Pn.
or
be
r,
v";" *"
=i+
Qn,
f"' '
B,
(,
1, 2,
. .
p n + r),
;
THE HYPERELLIFl'IC
239]
where P,
y,
Qn,
CASE.
373
expressed by
Denote them,
"r.
K, e
by
and a
respectively,
Xr (?)
and x
Each
(f).
them has a
of
and
single
therefore,
we
_ x(^)-x (a,)
^^^^>-x(0-x{Ky
,f..
But
it
n^'\ +
is
n^-\+
+ n^-\
y.xkryiO,
Thus the
Riemann
a;(f), etc..
special case
surface
"
= ^ + Qn,
etc.,
cor-
It is manifest
consult
ill.
p.
p.
619.
The
may
of automorphic functions
(cf.
From the
is
z of
Ex.
vi.
238.
[240
CHAPTER
On Radical
240.
Thk
reader
is
XIII.
Functions.
already familiar
wth
Such a property
instance we have*
V(l
Thus
for
u.
sn m) {1 ksn u)
l-22"sin^+?'"'
if
1
sn
ii
represent
sn' u may
is
be
possessed by
242]
375
been led to the consideration of functions of the form V(c ajj)... (c Xp),
which are expressible by theta functions with arguments u, =*>" +
it^'"p.
arguments
u, but,
only
single-valued
surface
is
functions
the
of
way, also of every one of the places x^, ...,Xp. In fact the square root Vc x
is a single-valued function of the place x because, c being a branch place,
X c
vanishes to the second order at the place, and the point at infinity
ratio,
c
a;
is
a;
The
More generally
241.
Z he a
it
may be
Riemann
mth
order, the
root,
\/Z,
And
it
is
For
kind.
roots of
ex-
Let
Vi'
II^'
be
the normal
..., Vp'
we have* equations
m(t?"'-|-
wherein
rtth
first
mth
way by obtaining an
of the
is
surface, that if
and poles
is
values
have a
will
t'i'"
"')
= ^i + ^'I'Ti,
+lcp'Ti^p.
-f
{i
= l,
2, ...,p),
i.
Hence
the expression
+nj
min"-" +
]-2Hii'v',-''-
-2^apV^'
wherein a
for
where
is
a certain constant.
period loop
factor
242.
a,,
m-
the dissected
it
defines
e'
with a factor
\/Z on
More accurately
',
VZ
is affected,
at the
i.
'.
We
Chap. VIII.
155.
376
Riemann
They
period loop.
\/Z by means
[242
a factor
each
for
such functions as
of theta functions
arising from the fact that the theory of the theta functions
To explain
mode
^^
'^{u-E; Qj'^iu-F;
'
ft),
^ (u,
this
q)
22
factors,
189) given by
ga2+2A(7i+9'|+J(n+5')2+2iri5(n+g')_
...,
E, F,
...
denote any
arguments.
where M,
M'
(u
+ flj,
190)
;
q)
= ^M m+2ni(M,r-M'q) ^ (
^iriM iq'
namely,
r'
Q'
/i/r
- R - ...)-2iriM' {q + r +
_Q_ii;
is
_...),
is
-E-F~...) + 2Trm{q'^r +
-\.,(e+/+
_Q'_ii'_...)
-2iriM'{q
Thus
^^^
H ) {u) = e\ wheie L
-\^{u-E)-X{u-F)-
Xj,(-e) + X.(i6-/)+
+r+
-Q-R-,,,)_
if
^Ei + Fi+
ei+fi+
and
li
+n
qi+ri'+
where Ki,
of
K(
-{Qi
+Ri +...)=^^Ki,
+ Ri' +
(Qi'
m is an
...)
(i
= l,
2, ...,p),
= -Ki',
M, M', that
fl,
w*i<
i+
+u'^'
"p,
244]
where
a,, ...,
a^ are
<Jr
(u)
each
is
the
mth
mp
the places m^
A. I,
each
377
..,
Jip,
i4i
. . .
zip , ...
u""^'
+u'v-^''
+ 2t^"^^+
+M^'
^+
m n^ K-,
(A).
'
of p,
and
also the
number
of
m times
a multiple of p. It is easy however to remove this restriction by supposing a certain number of the places a;,, ..., Xp, z^, ..., Zp to
repeated poles
is
rational function
X,
. .
.
Xp, Zi,
...,
Zp
is
characterised by the
inverse
have no
infinities
and
its
The
functions
new
property being that the values of the function at the two sides of any period
loop have a ratio, constant along that loop, which
is
For
a root of imity.
these fimctions there holds a theorem, expressed by the equations (A) above,
of the
(Chap. IX.
it is
may be
may be
interpretation
The most important of the radical functions are those which are
244.
square roots of rational functions, and in view of the general theory developed
ourselves to these functions.
in the next chapter it will be sufficient to confine
378
[244
denoted by
When
all
or ^"',
<l>i''',
p fundamental
when
its
^r{p
An
integral polynomial
(^-polynomials, 0,,
..., <^p,
fall
therefore, in general, at
by X"'
havef already been concerned with such polynomials, X^, of the
?(/) 1) distinct places, the polynomial will be denoted
in ^1,
be
will
no condition.
or F""*
first
we
degree
..., (^j,.
It
to be
is
VX''''
which have already! occurred in the theory of the theta functions. These
places are such that if a non-adjoint polynomial, A, of grade /i, be taken to
vanish to the second order at m, there is an adjoint polynomial, i/r, of grade
(
l)o-4-
Ji
-I- /x.,
...,
^1,
...,
*'^ will||
m, and
let its
other zeros be
contain o(jj
Let
zeros.
order.
X'''
i/f
is
nfi
zeros of
other than m.
1)
cr -f-?i
Since
3>''*
now IT
3-1- /x
yjr
(jan
..., nip,
yfr,
each repeated,
So chosen
it
so that
follows
may
be
denoted by X'^'. Thus the places m^, ..., rup arise as the remaining zeros of a
form X"i (with 3 (p l), =p+ 2p S, zeros, each of the second order), whose
other 2p 3 separate zeros are zeros of an arbitrary (f)-polynomial, <f>o, which
is
the case.
it
now
?ii,
follows, since
..., ?ip_i
m,
Vi
ft),
Vi"
X'^'/'^o^'X'^', that,
upon the
we have
mi
'-
Up
-Vi"'^'
"'
=-^iki +
kiTi,i+
+ kp'Ti^p),
* Chap. VI. 110 ft., and the referencses there given, and Klein, Math. Annal. xxxvi.
% Chap. X. 179.
t Chap. X. 188, p. 281.
Chap. X. 183, Chap. VI. 92, Ex. ix.
Chap. VI. 111.
IT Chap. X. 183.
11
it
..., m^,,
p. aS.
245]
379
factor, is the
n mi,7>, +
+11
p-l,p-l
+11
+in(K-.i\
ip, i
>
may
'^X^^'/(f>o'^X^^>
+ ftnt'
i>
'
^
!
be regarded as a single-
valued function on the dissected Riemann surface, whose values on the two
sides of any period loop have a ratio constant along that loop. These constant
ratios are equal to
e"*'-'
When
respectively.
and
for
e""'*"-
the places
first
vip are
...,
?)ii,
^ ]
Also,
if
order in
^ A'p
^'l
>
2^P
Y^'>
.4,,
y(3)/_yi3) is
Vi
we
may happen
....
+\''ri,p)>
-'ri(V-r/.i
+ V'-..p)
^\'
...,
|Xp'.
odd characteristic
characteristic with a
A^^,
F'^'
surface.
..., -^X^,
'^1^^1.1
is
Riemann
And, as in
+ ll!i'"
are integers, .4
Xj,'
Xi, ...,
=$('^ +
+Vi
II'^'"
where
is
it
..., nip'
Let
245
XU
X"*+''
(2v
ir,
now
^r,
/'I'f
A
^J'l
' * ' i
are Ci,C2,
Or
places.
...,
be
...,
any polynomial
each repeated
A^ and X'" be as
^"^2J^~*3
Then
mi, ..., wip.
whose
let ^o
zeros
consist
of
be as before, vanishing
the function
sets of
380
[245
is a rational function of order 2 (2i/+ l)(p 1) + 2, whose zeros are vi,Zi,z, ...,
and whose poles consist of the places 7)^,, ..., nip, and the zeros of <!><''', each
Hence
repeated.
as before <^v'Z'^''+"/^"'''^-^'"
...,
Vi
is
=i
+Vi
Vi
^^
(g'i
5'i
Ti,
+
(i =
->-qpTi,p),
2,...,p):
l,
we had used any other polynomial ^"'', the characterby Abel's theorem, only be affected by the addition of integers.
Suppose now that F'^^+'i is another polynomial, and take a polynomial ^i^'
and
instead of $'"',
if,
istic could,
then
if
of
V JT '*+''/* '''''^^^'''
<^o
zeros of VF(-''+^',
Vi
where M^,
and
d^, d^,
Mp,
M(
...
...,
we have when
Mp
of '^''^\
= |(gt + 9i Ti,
+ Mi + M^Ti,
+Vi
+i'i
...,
only by integers,
x^,
...
x<i,
denote the
the equation
+qpTi,p)
1 -I-
+ Mp'Tp^i,
-I-
Z^ is
equation
we
Hence
infer,
-f-,
when
there eocists
VX'^''+'', VF(2'^+i'
a form
<!) !'*+''+'>
v'jif (-"+1)
where
is
<I)<^+"+''
aiid
=^ "+''+".
same value of
fi,
whose
characteristics, save for integers, are the same, are expressible in the
(j,(M+.'+i)|VZ'^''+",
we
all
a rational function.
Nl"'')
form
are
therefore expressible
where v Fj
constants.
'''^
,
The
assignation of
2^ (p
1)
the remaining
zeros.
of zeros determined
When
by the
/*
of a polynomial
*
..., Xj^i^j-n,
there
may be a
-f-
X,,,
1) (^
...,
are
zeros of
any form
V F'^+"
number
Chap. VIII.
Xa^c^-i)
1) p,
others.
1, = (2/i
158.
V.X'^^i
having
OF RADICAL FUNCTIONS.
246]
381
VZi^ =
Quiy
$11)
It can
jjy
When
integers.
VZ, we may
if X'^**'
is
and
odd,
<!>(".
which
2/x,
all
order, and <I>'^' be any polynomial of degree fi, the quotient v'X'^''7<I>''^' may
be interpreted as a single-valued function on the dissected surface, and the
may be
form VX'^'"
VX"'''' are
$1*^+1)
Also
VX,
if
differing
is
sum
a rational function
which
v'Z'-''*.
sum
and
if
we determine (^
of each pair
1) pairs of
odd
characteristics,
is,
...
VX^) = x/xfFf'-fX,Vx^'-f-
As a matter
-HX^.iVxpT^i-
written
as the
sum
of
two odd
characteristics
in
/>
= 3.
We may
suppose
the equation expressed homogeneously by the coordinates x^, x.^, x, and take
the fundamental (^-polynomials to be (t>i=Xi, <j)=X2, ^3 = 373. There are
Vxi^ =
where
\i
Vx^i
-I-
Vx7' + X, V5V^'
X,
and
VX]'^', VXa'^',
-I-
\,
...,
Vx/*,
are special forms of
the assigned characteristic. The points of contact of all cubics X'^' of given
odd characteristic are obtainable by drawing variable conies through the
points
of
contact
characteristic.
Let
of
D.
Or a
characteristic differing
integers.
382
[246
are
xj)
xj) = 0,
Hence all
= 0,
of
contact cubics of the same characteristic as 'JD are included in the formula
or
X<='=Z +
where
f2P
+ i)P=,
P = X^x^ + X^j + ^3X3, Xj, Xj, Xj being constants the conic through the
D which passes through the points of contact of X'^' is
;
points of contact of
given by
n = 2'/DX"\
or il
= 2PD +
fig]
equally be written
4>X^'>B
If then
X=x Y=x
by
Z=X3, T=-'^XJI),
T given by
Y, Z,
__
_
VX = Vi) (XiX + X, F + X3^ - T),
!=
we have
4X0 (X,
r, Z)
2TD (X,
where Xo (X, F, Z)
respectively
X,
D (X
F,
Z)
Y,
z) =
+ a (X,
n^ (X, y, z),
F,
Z)
= 0,
is
F, Z, etc.
is
by the
2TD (X,
F,
Z)
+ n, (X, Y,Z) =
4X0 (X,
the curve
lies also
also
n^ (X, F,
Z)
Y,
Z) +
m, (X
F, Z)
+ X (X,
F,
Z) =
0,
be written
{T- Py D (X
where
Z) D (X, Y, Z)
T'D (X,
which can
F,
F,
Z) + {T-P)D, (X,
F,
Z) + X (X,
X('i
F,
Z) = 0,
= DP^ + n,P + X,
as above.
curve,
(ii)
that
the cubic
F=
surface
at
the
point
247]
X Q =Y = Z
X = (i=Y= Z
the point
D {X,
the plane
is
the 27
to
^=0
Y,
in the 27
that the
Z) = 0,
(iii)
straight
383
the
of
lines
surface
cubic
may be
surface.
of the bitangents
on a cubic surface.
metrical treatises.
Frahm, Math. Annal. Vil. and Toeplitz, Math. Annal. xi.; Salmon, Higher
Plane Curves (1879), p. 231, note; Klein, Math. Annal. xxxvi. p. 51.
also
247.
We
with one of the odd characteristics the association depends upon the mode
We have stated moreover
of dissection of the fundamental Riemann surface.
;
there are 8 36
.
= 288
ways
in
which
is
to be
proved
later, that
all
fundamental odd
fundamental
characteristics
can
be
denoted by the
These
characteristics.
what
follows
we
shall,
for the sake of
and
in
numbers 1, 2, 3, 4, 5, 6, 7,
definiteness, suppose them to be either the characteristics so denoted in the
table given 205, or one of the seven sets whose letter notation is given at
sented by combinations of one, two, or three of seven
characteristic, which, save for integers, has all its elements zero
may
say, the
sum
of these characteristics
is
products
by
Vititt^a, VjiaMsi
Vttjitij
this
Mj,
we
will
be
m,_j
= 0.
The three
multipliers be absorbed in
the
as
will
is
or,
is zero.
%, we have
123.
(
Hence
if
suitable numerical
V((.t(i2,
-I-
UiUu
UsUjsY
= 4Mji(2W,j'J(i4.
384
[247
If therefore
/=
JtjMji
+ MaWjj - WiMj3
<f>
W2M12
IWfhi
M3W13
>
we have
(/- 0)
Now if/ ^
(/+<!>)
were divisible by
= 0,
Ui^
Mj,
^"aWij (U3U13
and/+^
iUiU^u^iVri^
f<f>
',
divisible
would make
Hence, without
perfectly general.
- U^Uii).
/= 0,
when the
we may take
this is impossible
loss of generality,
quartic
2X1(2^12,
2
/+ = - (W3W13 - thUu),
<^
X being
ffjUjj
Therefore,
M4,
when
tti, ii^,
Mi4
as the right-hand
a quadric function
is
of the coordinates this will lead to a sextic equation in X, having the roots
X=
0,
\=
if
different
We
(m?) M17).
Mis),
double tangents
= M3t<i3 \j (MsMsi
M4M34
-1-
M3M,2
M1M23) + Xi'U^Ua,
-I-
M1M23
M2M31)
+ XiUsU^,
= Mjltaa X3 (M1M23 +
"aMsi
M3M12)
+ X^lh W31,
UiU^ =
lliUa
we
(m^, M14),
- Xa (M3M12
(B)
and hence
M4
(^'
+ ^) =
or on X23
+ r^ =
M23
(X2U3
common
Mi, M4
may
either lies on
be written in the
X3
form vAUiU^i
+ '^Buu^ \- ^Cu-^u^^ = 0,
Mi,
equation VmjMjs
thus
u^
may be
-t-
Mj,
u, are
'^u^u^i
Mi
and X2M3
is
that
-)--?
"^u^Uyi
Ui,u,,U23 intersect,
if
Hence
impossible.
M4
must pass
that
X3
= 0,
expressed linearly by
4
it follows
else,
,, u^,
= a, Ml + ajMs -H a^Us =
aiWi
1 /,
t-I X2M3
W2\
+^
247]
385
then the
or
W4
so that,
if ki
r-
+Y
AiM^
z(,
+ \3M31
2u.^
aJijU^j
We
eta
(I3
- hiu = ?^ + ^1 _
where
we
infer,
else
(2
h.,u,.,
+ a,//,),
therefore, as
u^s,
Atj
are not
fundamental
ft 3
and hence,
M12
M31,
the
A;2
^3,
k,
and
say,
ft 2 Cfcg
+ 2/iiai + aihi- =
ft 2 '^S
or
A,
=
,
Thus
Oi
a^
(! M,
as
or
'
+^
+ +
a,
rt,,
M2
+ flsMs) = 0.
(C)
a-.
thus
tint
U'lt
A^
A.3
fCl
we have
\
B.
\3
Oi
A.3
^1
^1
^2
as
25
386
and therefore, as Xi
'.
-,
^3=
and similarly
-,
[247
X,
we have, by
-Uu
But
M5
6,
we have
(rt.Mo
+ 62 Mj + 631(3,
Me
1*7
d-^^l-i
tangents
u^, u^,
u-,
written in a form
hi, h, h^
I,
m,
ft]
'/u^iiis
l-r-
+-
Oi
Ci
M23,
U31,
+ :r=0,
O2
terms of
in
Aka^
From
C3
t>3
Aka+
j4A;a3
+
+
02
0,
0.
D03
pk', pi',
then,
for ii-ajp,
if
1)3,
I,
m, n are determinable;
we put
for u-sijp,
a;
for
and
is
undeter-
cii '\fk',
ft for
a,
(C
23>
%i.
"12
fta),
(/3i,
ft, ft),
niay be expressed in
(ii)
^,4 = ^23
3
Oj
where
(i)
terms of
Bnd^
+ Gmc + Dnd = 0,
Blb
k',
CWc,
with suitable
exist,
(X3
the expressions
Mo, u,;
k,,
+ Blh^ +
a^
c.^
A
B C- D =
+p
+ + -T 0,
(^3
As the
tti
A
+ Br + G + DJ- = 0,
ttj
di, d^, dj in
in place of k.
and
c,, Cj, c,
11,
equations to determine
in
+ djMa + rfsMj,
also three other equations such as (C), differing from (C) in the
place of
+ 03^3),
we put
if
+k
\-a^u^-\-a.^v3,
^,4
= w^/p V^',
v.n
03
tti
Djj
= u^lp
= %!
Va^Ui
+ aiUi,
Qj
^k', v^
O-l
W34=
!
v^jp
'Jk'.
*'l
- + a,Mi +
03
OsWs,
247]
387
It will be recalled that in the course of the analysis the absolute values,
ratios of the coefficients in u^, u^, Ua, u^, u^, u^, m,,
have
definitely fixed.
values of
VM1W23
+ \U2V31 + Vm3?;i2 = 0,
lines as bitangents.
It
by
45, 46, 47, 5G, 57, 67.
If the characteristics
is
it
terms of
of
1, 2, 3, 4, 5, 6,
1, 4, 5, 2, 3, 6,
clear that as
u^,
M4, M5.
n-i,
original memoir,
Va;f
+ VjfT? +
'^/z^
= 0,
and the 28 double tangents are given by the following scheme, where the
a;
(2)
3/
= 0,
+ y+z = 0,
characteristic
(3)^=0,
(5)
A,x
(7)
(14)
+ y + z = 0,
(24)
is
prefixed to each
(23)^ = 0,
+ A,y + AaZ = 0,
(6)
=
G,x + C,y + Gaz 0,
ri+z + x
= 0,
= 0,
(25)
-|-
(16) I-
+ 5^2/ + B,z = 0,
(26)
(17)
+ C^.2/ +
= 0,
(27)
J+
(15)
l--\-A,y + A,z
Cs^
(31)
+ A,z + A,x=^,
C^3^
+ ^^^ = 0.
,,
= 0,
+ B^y + B,z =
B,x
(34)
(37)
0,
+ x + y = 0,
r
(35) j-^+
(36) |-
(12)^=0,
A,x + ^=2/ =
+ A* + B,y = 0,
^ + ^i* + G,y =
0,
CrelU
cvii
(1890),
252
388
^_^
rfi7^
'
_2/
('^5)
^^^^
A
A,
"
n^
A 4 \'^
(l-A\Ay
l-B,B,'^l-B,B,'^l-B,B,
rf;
'^
^- = a,
4-
l~A,A,'l-A,A,l-A,A,
'
6',C,
- Q6\ +
4
A,
a--A.AA'^
^3^0 ^ (1 - ^.^^)
(A
^'
_A
^^
1/
- C,C, ~
fc
oTcT^c;^) +
^"^^^
Here the
six quantities x, y,
z,
xlo
xl]
^+
g;
[247
r),
f,
c, (1
C3 (1
= ^c,c,)
= o.
C3^
= 0.
+ ?+ + y +
'?
- O3C0 +
"ils
C-,^
0,2/
Conversely,
number,
of
is,
6,
and, by the
first
^,
f in terms of the
the last of the equations (D) will determine Cj, Cj, C^
arbitrary lines x, y,
z,
is
all
then
it
/;,
algebraically that the 28 lines here given are double tangents of the quartic
curve Va;^
248.
Vt/7;
Before
+^
zil,
= 0.
the
connection between the two representations of the tangents which have been
given.
= 4ZA
246, 247)
(a;?
+ 3/7,-^0' =
4?^ a'y.
and
x^, X3)
{x-,,
(cf
p.
= ^,
382)
replacing
the
= z^-x^- yn,
fourth
Xo (ar,,
coordinate
x^, x^)
by
T+
= xyrj
u,
where
249]
is
x, y,
z or
x-,, x_,
x,,
389
the equation of the
cubic surface
iT+ uy D + (T+ u) n + Z =
0,
becomes
T'^
+ T {z^-
x^
- 2/7? +
2u^)
+ M^f +
(z^
It
or
{T+iiy^ + {T+
which
will
u) {z^
r,)
Q.
Write now
w=u-x-
= u-x-z,
^,
=u x,
v'
= u+y,
w'
v',
=u+
w' be arbitrary
t;
Z,
and
be
(T + u)(T +
from the vertex
where
X=
v)
{T + w)
Y= Z can
-{T +
(T + v'){T +
u'}
w')
+ V(u - i/) (u - w') + ^/{u - v) {u' P P'=u + v + 'w u' v' w'; we have in fact
V(P ^FHm^-
[7)
x=u u',
= v'
u,
(i)
= u' v,
ri
'w'
n,
w) =
0,
^=u' w,
^,=-(x + y + z + v + 0.=^P-P'-
Now
(i)
otherwise established.
Ex.
249.
quartic
is
Prove that when the sum of the characteristics of three bitangents of the
lie upon a conic.
i.
By enumerating
the constants
we
x.
odd
ii.
We
characteristic.
is
even, denoted,
=X
V^iMzMj + Xi
'\/ziiU^2"'13
+ \ V?'2"23"21+^3 Vtt3%l32>
* Any cubic surface can be brought into this form, Salmon, Solid Geometry (1882), 533.
t See Frost, Solid Geometry (1886), 537. The three last equations (D) of 247 are deducible
from the equations occurring in Frost. The three equations correspond to the three roots of the
cubic equation used by Frost.
390
where
X, Xj, Xj,
denoted by the
!r.
^3 are constants,
and
suffixes, as in 247.
iii.
ut, Uij
[249
Kx.
when p = 3.
iv.
known
zeros,
250.
v'^l'Pt,
I'P',
3.
is
of importance
Using the places tn^, ..., trip, associated as in Chap. X. with an arbitrary
and supposing them, each repeated, to be the remaining zeros of a
form A'''', which vani.shes to the second order in each of the places Ai, ...,-42j,_3
in which an arbitrary (^-polynomial, (^0. which vanishes in m, further vanishes,
place m,
as in 244, let
equations
all
of
. .
^ojj_3 , Ci
Cp
...,
and z
arbitrary place; and, as follows from these equations, the places x^,
may be
log-!^
;=::r
= constant + n.
is
Hence
arbitrary.
as the characteristic of
is
.
an
x^ps
..., Cp
+n
Cj,
'
+11-.'
+ 7n[<TiVi
wherein the place a
determinate, respectively,
VF^,
if
it follows, if
we speak
V^<''
+
+
+ n
+ a-p'Vp
],
of
the characteristic
i^Pi,
,ipp,
hPi-
iPp)
Math. Annal.
Functionen."
X'2p^3 , cXi
Zum
, .
ctp ,
250]
wherein
equal to
is
+n:;,\+-''[-ft>r'''+
AeK"<i,+
is
391
is (
+-p;)C"]
?"](/'*>'"_/.''.
Ce
where
(7 is
this
is
the same as
Ti
[{<r,-
- p/) .?" +
"]
Ce
(H)(^,a2,)-2 4- /">"'
+t,*2p-3,n2-3+ Ifl,)
+ v^i."l
4-
+yX2p-3,aip-3
(H)(j/t, <'2p-2
where ^ n^ denotes
if
jj
such quantities as
iC""'
*'"i''''>', i
+ "'p
"^i
p)
thus,
',
'
we put
and
we
recall the
formula
175)
infer that
VF'^)_^@(;
where .S
Now
is
a quantity independent of
in fact
are constants.
places Xi,
x.
+ ^- Vyf'
2j)-2)
...,x.ip-3
we
we
F is independent of x and A^
to
..., a;,j,-3)
. .
FA^
(x, ,, ....Xy^,),
. ,
X^jj-^
at the
where
-^it-,
(x)
VFi;U(^i)
^Yfixi).
i is
VF"' vanishes
^Yf~{x),
which
1^')
given by
W>/Fr +
where ^ Fi
i<7,
in
+ ^ Op)
1, 2, ...,
_g '^{v;
{v
2p
- 3.
jo-, jo-')
^p.
^p)
Hence we have
392
[250
To apply
by
"p
x^,
is
...,
a;2p-3.
is
given by
problem expressed
such equations as
^^1'
where
/it,,
the places
..., fip
''1+
-^-ifP'l^U,
aJy^)+
where v 1\
places Xp^i,
...,
+Ap^,JYfl^{x),
Ajjf^
(^0
form
v are of the
if- ''2/>-2+
w, where
B, Jzf^ (x)+
A,-Iy['\x)+
+ Ap^,^Ylf^),
B,Jzf(x)+
+Bp+,-s/z^;i,(x)
...,
places Xi,
...,
4/j
(a;),
of characteristic (|cr,
^cr'),
be
the
A^^f\x)+
of which
x^_3.
zeros are
known,
+Ap^,^^'l,(x).
Xp^-,
...,
x^^,
of VF'*' {x).
may
A bd'schen
Functio-
The
\>.
( 274).
problems of the contact of curves. See, for instance, Clebsch, Crelle, lxiii. (1864), p. 189.
For the theory of the solution of the final algebraic equations see Clebsch and Gordan,
AbeUBche Functnen. (Leipzig, 1866), Chap. X. Die Theilung; Jordan, Traite des Std>stitutioiu (Paris, 1870), p. 354, etc.; and now (Aug. 1896), for the bitangents in case p = 3,
see Noether,
ii.
p. 380.
p. 368,
and Kltin,
252]
CHAPTER
XIV.
Factorial Functions.
The
252.
present chapter
is
As in that case, it is convenient to consider the integrals and the functions together from the first.
In
order, therefore, that the reader
may
the argument,
it
obtained.
All
is
desirable to
explain, briefly, at
integrals
starting, the
results
The
finite.
the simpler theory have, beside the fixed singularities, no infinities except
But the
poles.
sides of
i^
quite
is
characterised
reciprocal.
We
linearly
no
The number
independent
and a system
consider two
functions, those of
their relations are
functionsX-
We
is
parallel
there
may
denoted by a
+\;
the
system by o- + 1. As
where finite factorial integrals. The number of such integrals of the primary
system which are linearly independent is denoted by cr, that of the associated
system by vr'. The factorial integrals of the primary system are not integrals
in the case of algebraical integrals,
The
u,
subject of the present chapter has been considered by Prym, Crelle, lxx. (1869), p. 354
In
xiii. (1890); Bitter, Math. Annal. xliv. (1894), pp. 261374.
these papers other references will be found. See also Hurwitz, Math. Annal. xli. (1893), p. 434,
and, for . related theory, not considered in the present chapter, Hurwitz, Math. Annal. xxxix.
For the latter part of the chapter see the references given in 273, 274, 279.
p. 1.
(1891),
enter.
t In particular the theory includes the case when k = Q, and no such places
view
of the comparison instituted between the development of
in
made
statement
is
This
X
factorial functions have
the theory of rational functions and that of factorial functions. The
SUMMARY OF RESULTS.
394
[252
= Mu +
where
fi,
constant and AI
fi is a,
is
the factor of
is
system, FdG'jdx
is
may be
it
number
system
j)rimary
o-
number
0,
^+
is
impossible, or
is
can
always be constructed
number
is
ay'
when they
\,
+ 1 = 0,
When
cr'
is
when
Similarly
1.
which a
o-'
the smallest
two numbers a
In the present
1.
is
factorial func-nr
+ 1.
Of
the
>
l is
0,
We
can also
construct factorial integrals of the primary system, which, beside the fixed
singularities,
the least
number
number
the least
is cr
+ 2. And we
of poles of arbitrary
numbers
is
are
a+
a +
infinities
For the
2.
2.
It will be found that all the formulae of the general theory are not
The
exceptions,
for
imme-
and
their
footnotes.
The deduction
267 of
278,
which occupies
271
deals,
this chapter.
With a more
The
by examples, with
Riemann theta
development.
The concluding
section of the
chapter deals very briefly with the identification of the present theory with
the theory of automorphic functions.
They may be
Ci, ..., c*
We
are excluded
effected in a
occurring instead of
180,
the places
Ci, ..., Cj
manner
253]
395
Zi, ...,
Fig. 9.
which the
We
call
{a,),
The
postponed
to the places
left
when we
pass from
to
c,,
upon
and by
(6^.),
we mean
..., cj
respec-
181.
Cj
Cp_i.
Of. Forsyth,
Theory of Functions,
396
254.
of 2p
A;
we take now a
hi,...,hp,
gi,...,gp,
an integer, or zero
. .
value
on
left-hand
the
side
of the
period
loop
(a;)
t^'^'tpi,
Ci
and
<}>{
c,-,
is
uniform,
and not
finite,
(4)
which
is e"-"**
Let
X,
y=^gnpm +
wherein
Xy
...
X,
x,a
is
x^a
x,a
+ np^^,,-n.,,,-...-n<.^_,-27rJ[(A, +
H,)i;i
+...
+ (ftp + ffp)r, ]-
integers.
It is clear that
(i),
is
this function
A,
and that
..., /Sjv,
in the
e^""-*^,
...,
Hp
are
this expression
k
x,a
SXinj.,,^
x,
N-M=i\i,
= e-^^^ij
a-^,
..., a^^,
is
and
where
= /-"'-F...-^A"'-<"'"-...-<--l(/^.+if.)T,.,-lx,/-"',
r=\
r=l
and
is
A:
\t to
to be
(3)
{hi)
SX
+ Xj
except poles,
whose
(2)
form
Xj
. . .
which are uniform on the surface thus dissected, and have, thereon,
(1)
infinities
g-27riA,
series
quantities
Xi, ..., Xj;,
no
[254
e^"''^'
if
only
k
vY'
+...
+ V,'"
-Vi'
-...-Di-"
-2X^1)/
r=l
r=p
= gi+Gi+ 2 {K + Hr)T,.,i,
(ii),
r=l
Gi being an integer.
* It
surface
is
places.
...,
Cj
should be in the
finite
part of the
255]
It
397
(i)
and
(ii),
such a
Conversely
it
capable of being expressed in the form here given, and that the conditions
(i), (ii)
are necessary.
primary
system.
The
quantities
e"^""'^', ...,
e~^^*, e~^*',
...,
e~^'^,
e'^S', ...,
er"^i'i>
are
fundamental constants,
^i',
W,
+ \i+\=0, hi+hi=0,
gi
+ gi=0;
these functions will be said to be functions of the associated system. The factors associated therewith are the inverses of the factors of the primary system.
255.
Riemann
when the
/d log/,
is
is zero,
a generalisation of the fact that the number of zeros and poles of a rational
is the same, and the condition (ii) expresses a theorem generalising
function
Now
kind.
of the integrals
Riemann
surface,
constants.
enquire
first
first
but
differ
we
In what follows
are expressible by
Riemann
are expressible by certain factorial integrals, provided the fundamental condefine then a factorial integral
stants of these latter are suitably chosen.
We
as a function such
that
dHjdx
...,
is
a factorial
FACTORIAL INTEGRALS.
398
thus
dH/dx has
[255
dH/dx
c;,
is
where
<pi
is
Similarly we define
Cf.
H', to be such that dH'fdx
factorial
is
X/ +
1,
. .
X-t'
1,
hp,
/i,'
gp,
. . . ,
</i',
or
\,,
if/ be any
thus,
...,
\i,
of the
way
associated
system, /'
The
sides of
dHjdx
...,
ffp-,
gi,
hi,...,hp,
Riemann
And
special.
system, and
is
surface, for
any
factorial
integi'al
of the
primary
I'ational function.
H = ,i.H + D
where
/a
is
e^-'''*'-,
and n
Near d,
e^^'-,
is
a quantity which
is
is
of the form
where Ai is a constant, (^j is uniform, finite, and, in general, not zero in the
neighbourhood of Cu and Gi is a constant, which is zero unless Xj + 1 be a
positive integer (other than zero), and may be zero even when X; + 1 is a
positive integer.
After a circuit round a,
will be changed into
S = Ai + e-^"^i
thus,
t-'^i
,f>i
+ 2-rriGi + Gi log t
when Gi = 0,
is
e-2'^^i),
1 is
a positive integer,
H=H + 2mGi;
in either case
we have
H = yH+r,
where y = e-^'^, and F
is
(ji).
all
independent.
256]
As has been
one of
Xi
X,, ...,
stated
is
we
399
an integer, or
zero.
256.
we speak
which
fall
fall
there.
at the places
or
If
a; =
2^
namely 2p
iV
so that the
Now
factors,
number
let fo
+ 2n =
dF is
of zeros of
2n
2p
+ 2
+ 1),
(Xi
+ S (X^ +
1).
Cj.
Then,
if
like <-<^i+ii
(/)f,
an everywhere
where
4>i
finite
factorial
is
uniform,
finite,
integral
function
Thus
all,
Z, =f~'^dVldx, will
Z satisfying these
number of
the
existent functions
V which
jZfdx is a
function V.
is at
least
=p-l+
No + 2n + 2p-2-{2n+Mo)-p+l,
provided this be positive.
functions
do
exist.
To
We
if
number,
rational
* These
let
if
J?dF
exist, is ( 37,
is
(K+l),
V^ be one such
function
conversely
when
k
I.
with
poles
then
in
the
be a rational function
a function F.
Chap.
Thusf
III.)
=^ - 1 + 2 (\ + 1) + + 1,
o-
But
their
difference
1,
as here defined,
is
400
where
o-
the
1 is
number
[256
<T
+1
vanishes
when S (X. +
1)
>
0.
Ex.
i/^
= {x-a)(x-b)(.v,
l).2p,
the factorial integrals, V, having the same factors at the period loops as the root function
V(.r-a)
{.v
h),
and no other
by
J{x-a)(x-b)
[
^=p-
and
l)p_2
...
dx
c^.
257.
than
k=0
Here
1.
(x,
when
zero,
it
zero, there
greater
is
are
cr
For
and
if
dVi, ...,
of the
first
be an everywhere
d^+i
dV
whose behaviour
finite
dvjf.^
at a place
in the
' dV
'
like that of
Cj is
neighbourhood of
^^
Cj,
^^
<j>i,
where
namely of
without poles.
<j>i
t~''i<^i,
uniform,
is
are clearly
Conversely
if
K'
xK'dV
become
infinite.
\s
Denoting
it
by
v,
dv vanishes at the
2jo
S(\-i-l)
zeros of
dF
as well as at the
+ S V. = 2 (X +
1),
zeros of
K'
(cf the
=i
F we obtain o- + 1
when <r + 1 > 0, the quotient of two differentials
dV, dFo can* be expressed by the quotient of two differentials dv, dv^, we
cannot thus obtain more than a- + 1 functions K' while, conversely, to every
function K' we obtain a differential dv which vanishes in the zeros of any
assigned function F; and, as before, we cannot obtain any others by taking,
condition
(i),
functions
K'
254).
and
since,
Chap. VI.
98.
2^^]
INFINITE.
401
The
258.
equations
and
(i)
.:"'+. ..+.^'"
254).
(ii) (
case,
'
- n, p (hp + Hp),
(iii)
iV=- 2
and
(Xr+1),
r=l
where G,
Gp,H...,Hp
are integers.
zero
integer or be
t^(\r
...,
1) v"""*
Hp,
+ hpTi,p = 0,
+ 5ri+ A,T,-, +
,
when these
conversely,
Hi,
conditions
are
(i
1, 2,
...,py,
fulfilled,
it is
wherein
o-
+ 1), we
F=
functions V, given by
l^o
dv,
where dv
2+
is
+ 1 (to be
= 1; and
+ 1 = 2+
1,
(iii)
is
first
kind
it is
easy to
When 2(\ + l)
equations
dV
differential
7
there are
immediately
+ <r +
l is verified.
let t
dependent
differentials dv
K'
is
B.
If
Chap. VIII.;
This
K' be one
95,
2+
is
1,
= s p \-t + \
number of
then the
Chap. VI.
26
402
dvi
[258
dvt
it is clear
rdvi
J
'
K'
conversely
an
Fbe
if
any
is,
are functions
JK'dV
is
integral,
v,
of the
Hence
o-
=2+
t,
1.
The number,
0-
+ I,
of everywhere finite
to the number of
which vanish in the 2p 2 + "^, {\ + 1) zeros of any differential
21, Chap. II.) o- + 1 is less than p, unless l,{\ + l) = -{2p- 2).
dV;
hence
Also,
when
F, of the
o-
+ 1 > 0, the
= S (X +
1),
s,
+ 1 > 0,
number,
is
vanish in the
which
o-
primary system,
it
Similarly
is
follows that
when
s',
equal
to the
zeros of
whenf
o-
When
0.
= - 2 (V +
tit,
1),
= 2X, = - 5 - k,
that
there
>
is
0,
we can
prove, by
everywhere
are cr'+l
fi.nite
+ 1 is the number of
differentials dv vanishing in the 2p-2-1\, = 2j^ - 2 + s + k, zeros of any
differential dV
and that, when
+1>0, the number
of everywhere
factorial functions
when
tions
K unless
o-'
cr'
57',
where
zero.s
system
is
equal to the
of any function
K.
number
Hence
we have
existence of functions
is
9i + Gi+iK + H,)Ti,,+
'
Which hold
form
+ {hj, + Hj,)Ti,p = 0.
<r
+1
(i
= 1,
2,
. .
p),
. .
260]
where Gi
G'p
In such case E^
are integers.
is
403
0-'
r'
+1=
Of these the
1.
given
>
(ii)
259.
cr
cases
and
(iii)
be
to
= 0,
o-'
+ 1 > 0,
(i)
cr
(iv)
a-
+ 1 = 0,
+ 1 = 1,
are reciprocal.
made
in
When
connection.
this
( 2.57)
by
dv^+i
dvi
dvi
dV" dV'
~dV"'
that of
t~'^'(j)
=^,
where the
+Xc,^'r^'+\.
=\i.,F/+
(i
= l,
2, ...,(<r
+ l)),
obvious otherwise.
cr'
For when
o-
cr,
compatible.
Similarly,
when
o-'
+ 1 > 0,
...,
F^r.
It follows t therefore that of the cr periods of the functions Fj, ..., V^,
at any definite period loop, only st-{<t' + 1) can be regarded as linearly
independent; in
fact,
linear functions
by
Ki,
...,
260.
o-'
of the functions
of the remaining
cr
(o-'
F^ may
...,
+ 1),
and
be replaced
of the
functions
A^+i.
A factorial
two
It is clearly
assumed that K\
is
not
a.
X,
. .
X*
is
zero or au integer
(a
is
1)
must be replaced by
excluded.
262
p.
404
[260
by subtracting one
one of the periods
Tj, ..., Ti zero; and by following the values of the factorial integral, which is singlevalued on the dissected surface, once completely round the sides of the loops, we find, in
arising,
two at
least
from the
it is possible,
yk^y ^^**
virtue of yiyj
Thus
For
[Qi {I
+7i72
yk~i^k-
factorial integral.
Suppose now that V is any everywhere finite factorial integral of the original system,
and Vi is any one of the corresponding integrals of the associated system. The integral
Vd Vi, taken
in turn equal to
1, 2, ...
V and
we thus obtain
ar',
by
(ir-f 1)
By
equal to zero.
is
is
tr-t-l
constituted by
we obtain an
c?'
...
is
By
taking
V'^.
linear equations
As remarked above
;
so that
these coeffi-
we thus obtain
most
at
pendent of one another and independent of the two reductions mentioned above,
2,p + k periods of Fare linearly expressible by only
it
follows
that the
2p-2-(-i-[cj'-(o--|-l)]
periods, at most.
Now we
have
sr =jo-l-l-S(X-H)-l-o--)-l,
T!r'=p-l-S(\)+a' + l,
and therefore
si+'S!'
= 2p-2+/i: + + l+a' + l,
(T
so that
2p-2+^-[izr'-(o-)-l)] = ia'-(o-'--l).
or-((r'-fl) is the number of periods of a function V which appear to be linearly
independent; and, taking account of the existence of the functions E^, ..., Ec'+i, this is
the same as the number of independent linear combinations of the functions Fj, ... V^,
Thus
But the conclusions of this article require more careful consideracases ; it is not shewn that the linear equations obtained are always
Obtain the lineo-linear relation connecting the periods of the everywhere finite
F, T", of the primary and associated system, which is obtained by
i.
factorial integrals
Ex.
ii.
Vd V
vanishes.
first
identically satisfied,
We
can therefore form linear combinations of the periodic functions V, for which the inde-
1, 0, ...,
0; 0, 1,
oj'
(ir
...,
+ l), 2p-2 + k,
p and
2p.
262]
405
We
261.
enquire now how many arbitrary constants
enter into the
expression of a factorial function of the
primary system which has
poles of assigned position.
N-p+ 1+h+l
arbitrary constants, one of
them
F, so that iV==ilf 4-
and A +
X,,
additive,
where iV
number
1 is the
r=l
is
the
number
of zeros of
of differentials dv vanish-
But in
fact the
For
if
is
clearly a factorial
V,
integral,
is
dV
dV
FodV is
an integral of the
first
kind,
v,
of^.
Thus, the number of arbitrary constants
primary system, with
given arbitrary poles,
M+
K-p+l+h +
1,
r=l
where iV
is
differentials
the
number
in
a factorial friction of
In particular, putting
M=0,
in-'
the
is
(cf 258),
+1
+ a'+l
the
number
'
of
already obtained,
We can
equations
262.
(i)
and
(ii),
254.
M
M given poles always
smallest value of
for
which a
exists.
last
formula
406
M=
When
ot'
V,
is
it
[262
of the
form
V'
wherein
A-,,
+ 1 = 0.
therefore A
nr'
A^'
...,
= A,V,'+
+ A^>r'^,,
M arbitrary
Thus a
places
and
last Article,
ct'
+ 1+ 2 \^-p + l,
=<r'
+ 2,
arbitrary constants.
When
cr'
+ l =0,
number
this
and the
is 1,
when
cr'
We
+ l =0
how
is ct'
Hence we
infer that
1.
we can be sure a
formula of
Ki,
st'
a'
> 0,
there are
+\
...,
+1
254.
When
priori
poles
is
of the form
F+\K^ +
where
X,
. .
\^'+i
assigned poles.
+ \^+,K^.+
is
any
cr'
+l
in
Suppose that
tr'
Firstly let
the
o-
> 0,
associated
+1=0;
...,
to obtain such
K^+i.
K', of
system, and
o-
+1
<r+l
the
of
arbitrary poles.
everywhere
differentials
finite functions,
dv vanish in the
Ar
2^-2+ Z (K + l)
is
with the
factorial function
zeros of
any
differential
Then ot'=^ 1 S
\=p l+s + k,
We
and
Hence
dV.
take
it
first
is
s,
= 1
the case
(X^
when
+
>
1),
0.
possible to determine a
r=l
rational
It can therefore
*
tr'+l=p + s + A;
function contains s
+ +
k
l arbitrary constants,
places.
This
being
additive.
one of these
For J/ = '5r'-)-, we
shall
have
ft
+!=
?,
and, theietoie,
arbitrary
M-zj' + h + l +
(r'
and
will
+ l = a' + 1.
then
264]
everywhere
factorial
finite
s+1
function
arbitrary constants.
K'
of the
407
K'
then the
an arbitrary
constant multiplier. Denote the rational function thus obtained by R.
Then RjK' is a factorial function of the primary system with the sr' + 1
assigned poles, and is the function we desired to construct. And since the
ratio of two functions K' is a rational function, it is immaterial what function
K'
is
is,
This reasoning applies also to the case in which cr+l >0, s = 0, unless
= 0. Consider then the case 'in which o- + l>0, s = and k = 0.
also k
There
is
or
o-
Thus
ct'
sr'
Nextly,
K' and
let
o-
= 0,
as well as
o-'
+1=
Then
0.
But, provided
inapplicable.
is
<(:
can apply another method, which could equally have been applied
o-
+ 1 > 0. For
+ l assigned
or'
if
poles,
be the
and
be one of the
Conversely,
(cf.
37, Ex.
is
ii.
if
ra-'
kind,
dV
P -j
and at the
is
a rational
2^-2
zeros of
first
1 poles of
57'
2, we
when
2p-2-l\ zeros
of
dV,
R jdV
-^
It contains at least
+ 2^ - 2-p + 1 -
(2^
- 2X), =
],
In case
and we may
the least
264.
isr'
fall
<
2,
so that
iir'
= 0, 2X = /) -
1,
number
of poles
254.
V,
In this case
is 1.
We
408
Suppose
hypothesis
to be such
> 0,
or
let
V be
factorial integral,
an everywhere
primary system.
Then dG/dV
2p-2 + 2(X+l)
zeros of
of G, say
is
with
o-
+2
[264
poles, and,
under the
the
c,
%=C{1 + A+Bt+...].I>.V.
where
is
quantities G,A,
to the infinitesimal
dG
where
this is the
=E
dV
DcV
D^F
JJ,V
j=^^f=-
E= G/DcV.
same
c,
the
as
Thus dGjdV
place
is infinite
finite
when
at a pole of
like a constant
multiple of
if
Now R
is
y}r,
o-
+2
is
assigned poles of
like
^+Arr+
+4,rr+A[A.itr-0rr]+
d:
+ B^
wherein a is an arbitrary place, gj, ..., e, denote the zeros of dV,Xi, ..., x^+i
denote the assigned poles of G, and A, A^, ..., Ar, 5,. --, -Scr+2 are constants;
the period of R, in this form, at a general period loop of the second kind, is
given by
Afii
(e,)
+ Arili (er) +
B,
+ B
where
i
D,,ni{x,)-^ni{x,) +
[-0.,+,"i (^.+2)
p.
1, 2, ...,
Now
Chap.
II.,
ej, ..., e,
and
^Y^i
this
(^.+2)
must vanish
there vanish
cr
for
1 linear
periods of
are zero,
we
infer
265]
the ratios B^-.B^:
i-B^+a-
unknown
is 11
Thus we
1.
constants.
409
r+ 1 p + 0- +
is
ra-,
infer
1,
of the
which
is
the form*
+ C^V^ + C,
+ C,F, +
[G]
The proof
</>,-_
in the
first
kind, and ii
(say
M)
(say
N)
zeros of /,
we
take
iV>
1 (cf. 37,'
Ex.
is
of the place c
it is
same
zeros of dv
It
results.
is
and the
necessary to
x.
d[r^^+Auty].
a constant
when x
is in
the neighbourhood
of the form
l-t
where
infinite in the
JK'iXc)
which
Riemann
III.).
'(x)
"^^im
c,
we
indicated.
a,
is
Chap.
instead of iRdV,
may become
ii.
if,
265.
may be
Then
neighbourhood of a.
h^h
K'{c)
omitted
this is the
same
1
as
f(l
F(7)J1^+L
DK'icm)
^'(
M^''
is replaced by a Riemann normal inteIn the ordinary case of rational functions, where V
for the general period of R, vanish for
expression
gral V, the coefficients olB,,..., B.,+2, in the
sufacient to enable us to construct
<r + l ( = 1) pole is
Thus
V=v,.
when
namely
one value of i,
kind.
second
the
of
the factorial integral; it is the ordinary integral
410
hence
if
At the
yjr
is
-v/r
infinite like
is
infinite at c like
A
-^rr--.
= ?:'' +
^^
A
[265
log
ty,
^f
where
ty
is
7.
U^",
(C)
where
a,
A ^+2,
Bp are constants,
Bi,...,
that
(i)
+ A^^,I),Ml-;^^,y + BA{x)+
A,D,M:-S,+
S (X +
1) zeros of K'{x),
that
(ii)
the
condition
first
v^"
v^' ",...,
constants A^,
...,
A^+2,
+ 2 + (p or)
o-
if
the condition
first
kind
Bp, and
-Bsr+i, ,
if
If
we suppose*
whose
differentials
at the place 7.
(i) will
thus,
finite at
is
is finite
to
+ 5pn/^)
be reduced to
+ 2+p-'sr + S(\+l), = 2;
(ii)
[G]
C,V,
+ 2 1 =07 + 1, and
+ c^v^ + C
+
(i),
the
number
is
as before.
Ex.
s,
= - 2 (X +
A-'(.)[2>,,j/:;;+
The
266.
+i>,^
If
we have
A {ir'(..)[i^-+
...
+r-}.
represented by
This
the text
.<:;]=
a;,, ...,
EG{x...,x,+2)
integrals,
is positive,
may be
...
1 ),
is to
whose
is
+E^V^ + C.
+ E,V,+
In general
it is
(i).
to that of
IN
267]
where E, E^,
...,
singular places
E^, C
the
it
constant
we may
<f>
simplify the
consequence
the
.4,;
c,,
one of the
factorial
+ <~^'
411
Near a place
the
of
represented by Ai
from
(71) is zero
further there
by subtracting
integral
that the
is
is
period
additive
obtainable by
is
following the value of the integral once round the boundary of the dissected
surface (c 260).
most 2p 2 + k.
We
F^,
suppose
of the
periods
additive
integral
is
at
the functions
of
The
zero.
is
the
linear aggregate
^iFi
+ E^V^
...
C^-(^'+i) Fa--(..+i)
+ F^K^ +
We
+ F^+lK'+^,
...
is
particular case*
is
that of
Riemann'.s normal integral of the second kind, for which there are p periods.
As in that case we suppose here that the period loops for luhich the additive
F.G,{x^,...,X+2)
to zero
+ F^K, +
+F^'+iKy+i,
vr'
-{a +
Av+2) has
is
to be taken once
The
267.
..., a^'+i.
ct'
-l-
assigned
+2
assigned poles.
C^_
so that the
cr' - (o-
-f-
all
zeros.
Of the
and 2p-2 +
result.
The
result
is
+ G^'^,T{x^+u x
be the assigned
constants C,,
...,
...,
.r.+O
poles,
a factorial function with x...,x^'+, as
k respectively.
x^,..., x^'+:,
of the aggregate
1) linearly independent periods
are
Let
be
amended by the
{<r
1)
412
[267
E,V(x
+ E^T {x d^,
...,
d,+i)+
...
+E^'^^r
(x^'+^.d,, ...,d,+,)
same time the aggregate does not become inaddition, to the result, of an aggregate
Then the
FiKi+
Oi ,
. .
a'+i
Oj,
a^'+i.
...,
z,ti,..., t^).
of
yjr.
sum
of
its
Near a pole of
is zero.
at which
i/r,
t is
the infinitesimal,
^ + B + Ct +
/
where
Fi, ...,r^'
]dt,
is
(BV).
at the poles
^1 iDV'^.\
where F/,
denotes a difi'erentiation.
value of
{{DV')-\-t{D'V)
...,
ct'
everywhere
finite factorial
integrals of the
z.
Thus,
if co^ix)
tr,
and
denote, here,
EADV,\+
+^^-(i)FV)x,
wherein the constants Ei, ... E^' are chosen so that to^itr) = 1 and ca^its) =
when tg is any one of the places ti, ..., t^' other than t^, we have A^ = w, (z).
,
Hence we
infer
by the previous
article (
..., t^>)
is
equal to
r(^,di, ...,d,+,)-&>i(^)r(i, rfi,...,d,+i)* Cf. 122,
Chap. VII.
-a>^{z)T{t^<,d^,
etc.
...,d,+i).
268]
where
arbitrary places.
413
..., t^> in
means of
it.
of the
first
where A^,
...,
A^,
+Am^(x;
B^,
...,
tj'
dent, the
factorial
if
may
there be A
function
+1
The
contains ilf4-
+ B,K, +
+ -Oo'+iilo'+i,
condition
not be infinite at
1, 2,
or',
+ 1 or'
o-'
that the
tr is
= 0;
+A^io^(z^)
AiCo^{zi)+
in case the
..., t^')
but
Zyi,U,
be linearly indepenconstants;
arbitrary
form
G,dV,'
M assigned
poles,
+Aj,a>r(zJ
.4,0)^(0,)+
GWFV,
(A + 1)
number
of arbitrary
is M + a' + 1 sr' + h + 1,
Prove that a factorial integral of the primary system can be constructed with
o- + 2 places, but with no smaller number.
i.
Ex.
where
where K,'
and
G {x^,
ii.
t is
is
...
x + 2) become
one of the
denotes
x^,
G (xj,
<r
x^,
+l
everywhere
...
finite factorial
1
GdKr,
equations
+2
2 EiDKr' {x,)=Q,
(7
t=i
D denoting a diflferentiation.
Ea+2
'^^^
268.
From
he expressed.
The theory
many
We
space.
exhibition of examples of its application would occupy a great
can
expressions
explicit
limit ourselves to examining the case ^ = 0, for which
be given, and, very
briefly,
two other
cases (
268
270).
414
Consider the case
here denote by
>-,
p = 0,
= 2.
a, /?, 7,
and two
/CO ~~
By
zeros.
~~ c\
I CI
'
^\x-yl a-yj'
Cj,
...,
Cj,
of such a function
functions
iV'
is
the
of
system with no
associated
shall
Xj = 3/2,
factorial functions of
but which we
Xi = 3/2,
numbers* being
the associated
logr
= S,
poles
have so
[268
is
found to be
+ Bx + G
NM
M=0
cr',
and the
function
factorial
et'
+1=5
is
clearly given
of the
N M =
This function
b.
by
{x-a)i{x-^f{x-yr
p(^)_
{x
(x
x^)
'
Putting
f (x)
{x)
f{x)
{x
+^(x- /3)-'],
and putting
111
S\, = 0, 2a;,\i=l,
Va;jXi</)(a;i)
= 0, 2 a;,%(i)(a;,)= - 2,
1
1.\4>{x,)
= 0,
2 X,
[1
The
T,
V,
factorial
is (
cr
=4
poles,
265) given by
e(^.?,..?)=/^)C|U[^4^-<^(.)iogj^;]|,
* It was
numbers X,
\ are integers,
was excluded.
268]
refers to t, |,
zero, this
^4-^
c,
and
(ii)
41c
the expression
(i)
A,<}> (t)
is
17,
ZERO.
IS
f,
(D
r),
^)
the expression
TA,
1
(^3-^).+ '^<^)(^:r:;:-^_
when x
zero
is
(ii) is
(i),
infinite
is
is infinite
the
when x
to
we
find,
expression
always
by expanding
is
infinite.
in powers of -
1 ^1
[1
r4> (t)]
= 0,
Ia, [2t +
T"-<i>
(t)]
= 0.
G (t,
that
for
K'{x)I)G{r,^,V.O
(a;
t)-
X t
(a
"
(V).
(1).
(?)
1+t^(t),
2t
-H t^c/) (t),
in which the second, third and fourth columns differ from the
The
factorial
integral
G{r,
^,
i),
tj,
is
only in
a) (x y3)
-Ail)
T
ry + X
^
(x
verify that
'
V(^^:)F^)
+
V(a;
f)
first
t.
For we immediately
/-dx{x y)
f in place of
2t
'T^<t>
(r)
JY
V(t
a) (t
(t
/i?)
(t)
- 7) ^ (t) +
</>
(t)
L - "-J^ 4 V(a;-a)(a;-/3)
+ 4 (^ - 7) (.^3^ + .^Z^
xlog
'Jjx
- <8) (t -
a)
+ V(a; - a) (t - /3)
Va; T
416
<f>
found
for
G (t,
<?(t,?,'?,?)
^,
rj,
f)
(x),
;
if
[268
we
we obviously have
7T
X T
= V(^-)(^-/3)
(a M.
<^(t),
T<^ (t),
+ t^^(t),
.2t
+ constant,... (2),
(?)
first
respectively of
t,
differ
f, 17,
from
We
f.
11
r/If
fi,
fx
y-T 7-f
we introduce
ji
7-?
-fJL
2\,7-a
7-^8/
/3,
7,
+Bx +A =
-/3
{x-^)ix-y),
we can immediately
A(f> (a:)
+^
2
verify that
A'(l> {x)
5' [1
+ x<^ {x)] +
C" [2x
+ x'<f)
=-
(x)]
(x-a){x-0)'
7-j+
X T
[A+(x-a)
A']
<ji
(t)
+[B +{x-a) F]
[1
+ t^ (t)]
7T
- t) (^ -
t)
a;
-T
thus
G{T,tr,,l;)^{x-af{x-^f
1
7T
(a
t) (/3 t) T
constant,
(3)
1
2t
t^(t),
+ t''^(t),
268]
now
clear
it is
the form
cubic polynomial
17,
417
^)/'^{x-a){x- fi)
is
of
and
where
f,
ZERO.
iy(x -t){x-
(oc,
when X =
G (t,
(2) that
IS
coefficient
of
G (t,
rj,
^) vanishes
Hence we have
y.
is
when x =
y,
and similarly
of the place
y,
G (t,
7],
f)
N+(x- yY
where iV
at
is
In the neighbourhood
/3.
+ m'
- 7) +
"'
- 7)' +
J,
In the general case of a factorial integral forp=0, i=3, the behaviour of the integral
/3, y is that of three expressions of the form
o,
A + [x-a)-*-[P+Q{x-a)+
...],
C+{x-y)-'{F'+q'{x-y)^
provided no one of X +
C may
A, B,
integral
+ 1, v + l
...-],
becomes a
We
1, /i
let
...],
factorial
them
when
all
of A, B,
C are
The
factorial
zero.
be at t, t,,
^,
tj,
f,
1?,
f)
in the form just found, the factorial function can be expressed in the form
when
?, V,
For
clearly
^^^:rr^^^x-Tr'^{^-T){x-T,)'
where piss, constant, or that the expression
C{x-fi){xB.
T,)
C,
(a;
t) {x
fi,)
27
418
divide by
(ar
Thus
7)=.
(7
Ti _ 27 T
(7 - /^l) (7 - ^)
M) (7 - Tl)
27
(7 -
= - (7 - t) (7 - /Xj)
Oj
[268
- m) (7 -
'^i).
^nd
/i]
/ti
'
or
1_^
J^
ry
this condition
is
satisfied
fj,
1_
7 /tl
T,
'
From
7T
,/
J_
x,
it
. .
to proceed as
however better
is
,
Then
ccr,.
P (x) can be
expressed
in the form
by equation
+G,G{x I
G.., ...,
V, ?)
+ C.
(1),
XrY
_{x
4>{a:r)
X,.) A (f
{X
V,
0.
E,
F,
kGr<j>{Xr),
<l>{^),
<f>{v),
<f>iO
?<^(|),
l+V<t>iv),
+
I
K'(x)BP^kc,
kG,[l+Xr<j>{x,)l
?</>(?)
wherein
of the
(f,
first
r],
element of the
....
(ri)
{x
- 1])-' ,G =
11
and E = {x ^)~^
{x-^)-' + <f>(^){x- 0~'
first
row,
+ <]){^)(x
If
(1),
^)~'^,
now we take
Os so that
2C,^ {Xr) = 0,
^Gr [1 +
Xrj> {Xr)\
= 0, ICr [2x^ + x/
<f>
(,)]
= 0,
this leads to
^A ^y^^'V? i)P
= G,DG (x
x x
X,)
+ G,DG (x x x
x.X
if
ra-'
=4
everywhere
dV is a
definition,
-j
the associated
factorial function
269]
ZERO.
IS
419
which has at a the form {x - a)-^<f), and similarly at /3, and has at
7 the
form (x y)-^^. Further dV'/dx is zero to the second order at x = x.
Hence we have
(^. 1)3
F'=
{x
t^a;
- af (x - ^)i (a
(x - yf
and
Thus
pr'
_ p
Lx' + Mx +
dx
J{x-a)i(x-^)i{x-y)'^(x-y)(x-a)i{x-^f
where N, M, L,
are constants, K' {x), K^ {x), K,' {x) are particular, lineai-ly
independent, everywhere finite factorial functions of the associated system,
and Fo' is a particular everywhere finite factorial integral of the associated
system.
Ex.
(r
i.
+ l = 2,
jo
= 0, ^ = 2,
that
that
it
in the
y.^{ (Ax+B)dx
J{x-af{x-8f-'
X=
Ex. ii. When we take p=0 and k, =2n+2, places c^, ..., C2+2, and each
i,
prove that the original and the associated systems coincide, that o- + 1 = <r' + 1 = 0, c7 = nr' = ji,
that the everywhere finite factorial integrals, and the integral with one pole are respectively
where
f{,x)
Jf(x)/(x,
oi,
...
hand
= {x-ei)
l),i
+ i;
The
(^-Cjn+j)-
factorial
function
with
fewest
poles
is
(.r,
l) +
i,
2-l-l coefficients
on the right-
side.
One
t = n^;/., +
e*,
Til'Ja,.
where
- 27ri ^{hi +
Hi) v'
EXAMPLE OF RADICAL-FUNCTIONS.
420
where
Hi
fraction,
is
an
integer,
and 7
..., aj^
the poles, hi
is
an arbitrary place.
zeros and poles
is
The
[269
a rational numerical
The
singular places,
expressed by
/"" +
where G^,
...,
= g + G + T{h + H),
+v^'''''''
since, if to
be the least
common denomi-
=p
w'
is
1;
i.
cf,
where v
and *i,
is
^i
of
Ex.
form
Chap. XIII. ).
sum
Chap. XIII.).
Cf. 245,
267.
any
in 2''"2(2''~'- 1) ways.
In the hyperelliptic case, when the numbers g, h are any half-integers, prove
07' -f 1 =p poles is given by
ii.
(xi, y,),
...
are the
^{x) = {x-x^)
and
the
a, b
...
ix>les in
{x Xp),
yjr'
question,
{x)
= dyjf {x)/dx,
u = {x-a)(x- b),
are two suitably chosen branch places*, and Ui = {Xi- a) {Xi b).
the ftinction
'^
,
Shew
that in
e-iO'-').
<r(M-)
In the case in which the factors at the period loops are any
c,, ...,Ci being still absent, it remains true that the
270.
number
and there
Sr,--t-Ti,,
in
will
is
equal to the
finite function of
number of
the system,
K+
which
may be an everywhere
Gi,...,
Hp
-Ht,,p Hp],
{i=l,
2, ...,
p)
by
* For the association of the proper pair of branch places o, h with the given values of
the
numberB g, h, compare Chap. XI. 208, Chap. XIII. 245, and the remark at the concluBion of
Ex.
i.
270]
then E~^
<T
+1
an everywhere
is
+ 1 = 1,
tr'
TO-
isr'
finite
= p.
It
is
it is clear
421
is
being a
Then every
function
of the form
ER,
rational function.
When
(r
o-'
+ l=0 we
have
= p 1 = or'.
ct
having p
zeros,
If ,
poles.
and the
. .
ajp
is
z^, ...,
Zp the
+v'p'^p
i;^''*'+
There
means
= g + G + tQi + H).
expressing such
function,
whose correctness
is
immediately
way
of
verifiable,
namely
(.u-g-6-Th-TH)
H (m)
wherein
u
and m, m^,
factor this
...,
is
ir'."'
^iii
,,^^,
^fv
Chap. X.
Omitting a constant
the same as
e(M-ff-TA )
@(^^-
.^
_
,-<^W,say,
r- log
()
4>
is
a rational function of
x,
and that
log
The
2/3
(u) is a periodic
x,
(u)
ii.
Ex.
(f>
constants
a,
</.()
represents
(M+2a.) = 4</.(), <t> (+2o.')=^'<^ (), where A, A' each
equations
given constants, and the notation is as in 189, Chap. X.
satisfies the
Cf Halphen, Fond.
Ellipt.,
Fnnc
Prem. Part. (Paris 1886), p. 235, and Forsyth, Theory of
By further development of the results given in Chap. XI.
formulate the corresponding
it is clearly possible to
p^l.
422
We
[271
have seen
zeros
is
function,
differentials, dv, of
of the
is
the
number
of
When h+
function.
linearly independent
NR
in
NR
+...+Vi
Vi
-Vi
- ^
-...-Vi
X,V;
-igi
+ KTi^i+...+hj,Ti^p)
r=l
= -{Vi
a being an arbitrary
Now,
place.
+Vi
let
),
when
O^F,{x)+
where
independent
then,
when the
zeros
ySj,
. ,
Gj,Fj,{x),
and Fi(x),
/3ji_i
...,
is
A{x)=
F(x),
,Fj,{x)
Fi|3^),
,Fs(^,)
Fd0R-^),
,Fj,{0j,_,)
..., /3^_i,
Let
and
given by
Hi= t
take any places
Xr'oV "
...,
+gi + Kri^^+
hpTi^p
m and
jj
function of x
@(/-"'_^,V""''_
if
_n+.- ">,.+.
+ ^.."+
+/-.-"_^. ',.+=_
are ( 179,
-V'"'""-H=-V''"'"'-
_j;^"./'
EXPRESSED IN FACTORS.
271]
or,
what
423
is
/""+
+/''-''''_^-""_
it
_^.-'-''_
jy
follows,
A (),
7i
We
M to be as great as ^
suppose now
we take
269,
p.
= r +p
1,
1,
say
as in 184,
all
the
is
^flg_ ^
Further*, in this
e {v'' + in,,
'
so that
The
function
Then
X {x, z) must
say
article,
e^''^'"
,.)
is
by \
poles Hr+i
Oj,
at
{x, z),
\ (f, /t)
of 238.
h+\
(V)
=A
i^
X {x,
7j)
[\
(.r,
Cj)fi
'^
j^pi
U\(x,^j)
In order to prove this it is
that the right-hand side represents a single-valued
J.
is
a quantity independent of
sufficient
to
shew
where
function of x on
(i)
x.
( F),
that the right-hand side has no poles and has only the zeros of
and (iii) that the two sides of the equation have the same factor for every one
(ii)
Now
places Ml,
. .
?i.p_,.
poles
(/3)
its
z,
and the
A (a;),
(a)
\ {x,
a,),
times in II \
{x, jj),
and
X (x, z) has no
The places Wi,
the function
p. 134.
424
*
[271
\j
times in
11 [X{a;,
Cj)]^-'';
hand,
M-(p-l)+h+l+l,\j-R,
times, that
not at
is,
all.
Thus the expression on the right hand may be interpreted as a singlevalued function on the Riemann surface dissected by the 2p period loops
for we have seen that the places i, ..., np_, do not really occur, and the
multiplicity, at Cj, in the value of such a factor as [\(a;, Cj}Y> is cancelled by
the assigned character of the factorial functions F(x) occurring in A (a;).
Nextly, the zeros of the denominator of the right-hand side, other than at
i, ..., ?)j,_i,
are zeros of
A (x),
A(a;), other
than
Wj, ...,
)?j,_i,
finite.
A (x)
It
...
7^+1
beside
@ ( V).
the zeros of
Uj),
A,
...
y8/i_i
loops.
The
txMKX
t?
11
=1
the factor of the right-hand side at the i-th period loop of the
e*,
first
kind
is
where
k
yjr
= l-rrihi
-\-
rTris/
+ {h + 1) Tris/ + irisi 2
'Xj
{R 1) -jrisi
now
R = N-p-V\-\-h-\-\ = r^-^\j-\-h-k-\\
thus
</r
=-
t-nihi -f nrisl,
and
3=\
e*
= (,-mlh^-\s(^
the
t'-th
{x, z)
/4
=-
it
Qj.
^jjg
factors of the
at the
ltd \yi
I'-th
two
first
''
-^\si-\r is/r,-,
. . .
4- s'pTi_
is
+ ^Tf, i]
e**
-)-
be proved, at
where
iri {s^Ti^
1 -I-
. . .
+ s'pTf, p),
follows that the factor of the right-hand side at the i-th period loop of the
second kind
X=
^-mrji
is
&
27ri
where
wj'
> -|-
v-
'^^
+2
\pi
-jri\r
Ti^ i)
27ri
Vi
'
- X
v,
'
+ h-^l+ 2Xj-iJ-H
(t/,
i=i
= 2Trif/i mi {si +
>
-f-
2 V;'+ 2
\jVi
'
- 2
Vi
'
*;),
EXPRESSED IN FACTORS.
272]
425
now we have
Vi=vf"'-vr"''-...-vY^'-"''^^+vf'"+...+v^''-'-''-vr""^'-...-vT""
-Vi
-...-Vi
-Vi
-...-Vi
-gi-Kn^i-
-i.\jVi
...-hpTi^p,
and
\
{si
Si't,-, 1
Sp Vi, p)
J'"" '"'
...+r'r'''^'
'"'"'' '"
"'-' '"'-'
thus
T^f-i(Si
+ .V,,> +
...+SpVi,p)
= r +
t^r''
-2.XjVi
-...-Vi
-Vi
+ ?;'" +... +
-gi-hiTi_i-
...
?''-''''
-hpTi^j,;
J=i
further
+(/t
-i'
+ l)r)i
-(R-l)Vi
+ t
+rvi
XjVi'
hence
Vi -iisi
- ^i\T
''>
t-r
+ i
XjVi'
v7
''
j=l
J.
+S
"'
j=l
j=l
'-Qi-
IhTi^
...
/vT,:,p,
j=i
or
(/v
^s/)
^M,
1^.-
= - ^i + hi + S
a;-'
'^
"s
t^!'
^>
+ S
.;' ">
And
\ {x,
yfr
before passing on
z) be
it is
everywhere replaced by
\ (x
'
z)
A {x)
and
be replaced by
i/rA {x),
being any quantity whatever, the value of the right-hand side of the
In particular
equation.
We
111.
can
assigned places;
now
Vi
may be a
function of
x.
*/r
first
...,
Let
a, a,, ..., o^
be any
kind, of which
all
and
Vi
...Vi
=t
(,*i
+ *J^', +
1
+*P~'.P'''
V'
^' ^'
>P)'
h+\
be the
nip being such places as in 179, Chap. X.; let
in
zeros
of
which
vanish
the
dv,
differentials,
independent
linearly
number of
a,-, ih, -, p-i as
having
a,,
...,
system
primary
a factorial function of the
VI,
iih,
426
number
the
poles, or
dV,
of differentials
[272
of everywhere
factorial
finite
which vanish
...,
Mp-i,
k
a,
cLr;
,
.
let 71
by R, and let*
. .
x^,
yh+i be
. ,
..., Xjt
Hi, ...,
Cj,
where
y}r
...
...,a;jt)
(a;,,
and
(x) denotes
Fj,ix,)
F,ix,),
Fj:{x,)
Fi(xji),
Fj,{xj,)
Z
which
a.,
a,
and
X (x,
Vr(,),
y.,a
j=l
,^{'k),
c.,u.
j=l
let
be denotedf
poles be
let
''i^
j=l
independent of
is
let
= FiM,
j=l
+A+
a,.,
X,-
+ Cj,Fj,{a:),
C,F,{x)+
wherein
r+ 2
denote
if,
definition of
we put
z),
{X, Z)
we have
nn
wherein
Ci, ...,
273.
to
(7 is
Cic,
n\
h' \
aj)
(a;;,
1=1 (j=i
(*;, 7^)
j=i
a quantity independent of
The formula
function which
have
its
just obtained
application
may be
...,
jTi,
it is
Xjc,
[\
(a;,-,
c^)]''
j=i
is
ni::
if
{xi^ Xj)
examples of
We
log
{x'
x',
272,
. .
x)/{x' z),
Px-i of 271.
(/i
Js').
be denoted by
273]
X, Xi,
i27
...,
surface,
x;
x',
2'
z,
(x
= log {x'
now
X
zj
z
z
- x) jz' - 2) e (^-^+
n,,^ )
('.
|n,, ^)
X,
z which,
firstly,
a-',
^'
we suppose
log
e(^;^--
+ in,,,o@(t^'--iii,,,-)i
X () X {z)
- ( - zf
where
D denoting a differentiation,
(Chap. X.
but we have
175)
properly interpreted,
are
roots
we
obtain
Iiin.x.,x-,
z=2
^/XjxfT^)
When
the places
Denote the
x,
z)e
{x
(i)
x, z, respectively,
'^'
suppose x =x-\-
*'"+',
if
by
t, <i,
z =z-\-
and,
ti'
7 denote an arbitrary
place.
n-l
Hy/zr
r=l L
- log
r=v>+l
rf^'t7'
tions
^
'
@=
(v^''
.^=../=.V-*'^
to be
+ ^n,,,).e'^'^''-
ttiX{x)X{z)
lim
n x',^
r = Mi,+l
= -iog(r)-log(r)-i-iog l-{x-zy
where
t.
1=1
as x,
z'
= (x-z) e^^.
^X{x)X{z)
t,
t,.
respectively
(n)
428
The equations
(i),
@ (v"' + | ilg, )
'
one position of
is only
vanishes;
order, only
(ii)
is at z,
>ip_i,
vanishes only
VX {x)
^1
is infinite,
and
there
for ( 188, p.
x for
there is no position of
e"*"^'
namely when x
x,
[273
(. ^)
n-^,
{if' + ^^s,i,)
and
...,
first
'
m^-i
there
no
is
Putting
is infinite.
= (^j^-M-JiVsOi!^'
-JX
{x)
X (z)
we have further iCi (x, z) = tsi {z, x), and if t denote the infinitesimal near
For every
to z, we have, as x approaches to z, limitj;_j [cti {x, z)jt] = 1.
position of x and z on the dissected Riemann surface ot, (x, z) has a perfectly
determinate
value,
equations
the
(i),
save for an
value
this
(ii),
There are various ways of dealing with the ambiguity in sign of the
{x) be any (^-polynomial vanishing in
function bt, {x, z). For instance, let
an arbitrary place m, and in the places A^, ..., .42p_3 (cf 244, Chap. XIII.),
<f)
and
let
the dis.sected
may
determination of sign
for
E,
(a:
z)
^ZMZE) <f^i^f^+^^^^
<f>
(i)
(x)
^ (z)
/* {x)
'
on the dissected
z, (ii)
E^
(x, z)
..., rtip,
* (z)
Riemann
= Ei
surface
it is
it
a:
it
^',
and
is infinite
first
and
2)ri(t)/"' + ^T,,,)
a single-
as a function of
{x,
z)
advanced in
which the ambiguity of
Suppose that the Riemann
way
in
273]
429
( 227) which are unaltered by the linear substituLet the places in the region (S of the f plane which
correspond to the places x, z, x', z' of the Riemann surface be denoted by
?i f. ^. f- Then by comparing the equation obtained in chapter XII. ( 234),
dv dv
d"f
d^
integral of the first kind, which vanish to the second order at n,,
..., Hj,_,,
-.(^,^)/A/|.^ = -(f,0;
now we have shewn
any one of the
that ^^(f, f)
infinite
of X, has a continuous
hence
it is
itself is of
and
to^i
by the
(x, z)
BTi {x, z) is
factor
conformal representation.
and
possible, dividing
ambiguous
is
number
is
a/ -ft
The modified
upon a
we do not need
is
single
function
which by
-tI,
will
conform-
f plane in a manner
227, 230*
homogeneous variables.
Klein has proposed to deal with the function Wi (x, z) by means of
xxxvi. (1890) p. 12, and Bitter, Math. Amial. xliv. (1894)
pp 274284.
is well established.
4:30
of the places i
independently of z
Ex.
but this
is
:c.
Ex.
Prove that
i.
ii.
zeros at
/9,
CTi
^^,
(.r,
g) OT,
c)
(2,
+
X^,
...
^li^,
X,,
suffix in
function sr
274.
is
(f, f)
As an
+\f;
a fixed place.
be apprehended, we shall
and denote it by xr {x, z). The
the expression
is
to
where
X'^' (x) is
we take the
differential
zeros of
244,
(cf.
of
coefficients
having 3(ju
O (x),
of
the
case of the
and the
the 2p
oj,
a-x)
omit the
n,, ...
n^iC'^, "i)
where
[273
first
is
kind.
Thus r
a,-
are
the
linear in
= 2p 2,
2p-2
1) zeros,
Oi, ...
which
ifr'i,
and,
{x),
the expression
Vx {x),
2p-2
-Jx^U-^)
2p-2 ip-2
n n
^/zg'_,(^^P-2)
i,j=l,2,...,2p-2
2p-2
Sr{Xi,Xj)
^(X,)
i<j
i=l
nn
Herein
v-X^i
<!>() is
<i>
or
(a-'i,
Oj)
(a;2p-2)
...,
is
Ap)
for characteristic.
immediately obtainable.
Weber, Theorie der Abel'schen Functicmen vom Geschlecht 3, Berlin, 1876, 24, p. 156
xxvm. (1887), p. 367; Klein, Math. Annal. xxxvi. (1890), p. 40. For the
volume.
Abel'schen Functionen."
cxi.
cf.
276]
A general
275.
may
functions
be made by taking
p 1.
Then
Oi, ...,
Ar
...,
expression \IX
(x),
result
+ ^r,
/r
(a;)
in general
the
upon
1
CT {xi, Hj)
i^^
J'r-lC^,), 1
s/Xlx,)...X {xr)X{a,)...X{ar).
i,i=\,...,r
i,i=\....,r
r^(xi,Xj)
T^{ai,
a.j)
many
276.
the general
(cf.
n n
^r-l-f.-l (^)
and we take
if
be
F,{x,l ...,Fr-,{x,),
F.C^A---.
are constants,
we obtain the
a,,
/i
A,F, {x)+
where A^,
431
We
particular cases*.
proceed to
obtain a more special formula, deduced directly from the result of 272.
Then the everywhere finite factorial integrals
Let tti, ..., a^=ni, ..., Jip_,.
of the associated system are the ordinary integrals of the first kind,
which vanish in the places a^, ..., a^,
and the number, A+1, of
vanish
to the second order in the places
which
is,
that
...,
Til,
n^i,
dV
Wi,
+ + 1, =p. The
<^{x)IX(x), where
is 1.
...,n^i,
mVi
i^
<>
X (x) is the
n^^
[/
_ ^s _
is
equal
( 175,
ss'
is
expression
^Ts'
f/-
^s
+ \ts') =
e'^'"'+''''
F+ fi,, s)
Chap. X.) to
g,'(r+o...')-2.(F+iT,)
since
Further
Hence
place.
g (ir+}+iT/,
kind.
7 being an arbitrary
first
F{x)
an odd
VZ {x),
0(7),
integer.
X {x,
A(a; ...,)
=e-.i<'(F+.)
0(F), = _e--^r*'e(F),
z) is is {x, z),
*.(^i),
and
.^i>(^0
^^!X{x,)...X(,x^),
.B8.
^{x)
the
432
where
*(a;),
...,
't>p{x)
'
n 'ff
....
"'jdt,
.M^iy-.^'p)
@^v)=
Ce-^m-
denote dvi
dvp "jdt.
f^ |-^ ^^._
,^)_
[276
...,^{xi, 7i^_0
^ {xu
7)].
^{xi,wj)'=^
4>,(a;,)
by
poles,
fi
(xi)
is
depends
,1^(Xi, Wj,-i)
CT (Xj, Wi),
in fact
surface,
we
shall
^p(x,)
^e-y v@(V)=
^i('^p)
^p('v) V(^i).
n n
-. M(a:^)t!r(a;i,
or (xi, Xj)
ft,
7)
nrjxp, 7)
(7)
where
p
1
,j,^
.g^fa.7 ).
m(7)
and p
VX (7)
kind
is e^i^'ip-p-J)
factor of @(ir.-
factor of
'*!',
Thus the
factor of
At the
or e"'.
+ ins,,.)e'^'''''''''
is
^ at the i-th
be
suffi-
is
upon
7,
the factor
first
-v'^'"'^)^^yy.m_^m,_
kind, this function, regarded as
_ ^, mp)
depending
it
has
277]
CASE.
433
but since
it
follows that
is
equal to
thus, changing
into x,
(, i)
or
it is
set
down
is
same
zeros,
of
x,
given by
and no
poles.
completely established*
277.
arises
= 4(ar'^=+
y2
Then the
places i,
of the places
...,
Oj, ..., a^
d^+i,
+1
..., 6^+1.
np_i,
rfi,
(m +
....
1)
d^+i,
J' ;
of the places
6,, ...,
then
less
than
fi,
Xi, ...,Xr
6^+1
...,
a^j,
such a function
/i
x = di,
is
and zero in
d^+i
It
the
regarded as a function of
!, ..., a^, di, ...,
We
at
and that /a
which = 6,,
).
is
272),
(
...
Provided v
nj,_i,
is
not
of the form
(xi-n,)...(Xi-np_i)(xi-di)...(Xi-d^+i)(xi-bj)...(x,-bi,+i)(Xi,l\_,_^+yi{xi,l\_^+^
(xi-ni)...{Xi-np-i){Xi-di)...(xi-d^+i){x,-ai)...{xi-a^)
the terms
is
(x^, l),_i_^
we
when
x^ is infinite.
When v<
shall
suppose
fi,
W^hen v
fj-,
v^
fi.
The 2v
\ ratios
',
274,
and
in particular Klein,
p. 39.
{x, y) is (x,
-y)
28
434
to Oj,
Hence, save
..., a^.
a factor independent of
for
[277
the determinant
Xi,
by
V^(a;i)
{xi-ni)...{xi-di)...(xi-a^)...
i/-i-,i
-a;
(a;) = (a;
= jr7i/r (a;), and
wherein
</)
(x)
this
i/r
determinant by
'-V.IrK),
ni)
. .
(a;
-V^|r(aO, ar'+^V^CaO,
...,
rij^,)
(a;
....
,;,,
V,^ (,)
6^i+i),
denoting
(
272)
is
therefore equal to
B,
i=i
(a;i
Hence, from
tti)
{Xi
. .
j - a) V(a:i
Wi)
taking
272,
yjr
(x)
M-M-(^.-W
.
(xi
- Uj^^) V
(a;
(ajj
-d^)... (xi
Mp_i), so
that
- d^+,)
tn- (a;,
^) will
denote
V(a;
'
...{z-
rip^,)
we have
Li
D^,^
where
Now,
is
independent of aij,
if b,
and hence,
h)~ %
(tr- *;
places,
\s, |s')
place,
|s, |s')
if
V\
/3i
a^)
^f =i(?i + /T,-,, +
where
ID- (a;;,
..., a;^.
ts {x,
n n
= i (A + A'tC, +
,
m (x,
BT
(,
d)
are integers,
+ VTi.A
(*
1,2, ...,p),
+ ^/Ti, p),
we have
( 175,
Chap. X.)
BY THE FUNCTION
277]
where
is
Vf {x, z)
independent of x.
/&
g-^H-_fl-l^,a CT(ag, d)
V!{x,h)
which
are
no poles or
clearly has
all
The square
1.
expression itself
Therefore
cti,
...,
of this function
V.'
by
Jo-
= i(o-i + <Ti,,+
and the
+<Ti,j,),
is, ^s')
{Xi, bj)
ra-
a;;
- dj
Further
..., x,.
- iro-'
Chap. X.
175,
therefore a constant,
independent of iCi,
(m
is
if
j=i 3=1
is
x-d'
a constant.
is
2
where
435
= ^e""'" @ [m
(s
_ ^),
('
.')]
C%\i^'-i; ^{s-a),^{s'-a')
= D^_^ n U
T^ {xi, Uj)
i=lj=l
where C is independent of
independent also of Oi, ...,
n U{xi-
CT {xi, Xj)
i<i
a;i,
..., x,.
oLj)
= 1 i=l
-nr
i<j
Hence we can
infer that
is
(on, a,),
'
in fact
For when the sets iSi, ..., a;^, a,, ..., a are
interchanged, D^^^ is multiplied by (-)'^+--' = (_ 1)m, and, since v! {x, z)
^ Briz, x), this is also the factor by which the whole right-hand side is
a^.
by +
1.
It
that the
follows
characteristic
odd according as
hence,
be
(s
if
\{s
/i
is
cr),
\{s'
a),
even or odd.
by
denoted
by
- 0-),
i(s'
^n + if^'" +
^fl,
a-')
the
half-periods
associated
with
the
characteristic
!)*'+
t)V+i'"-|-i)''""+
-I-
w'^M+i'",
282
436
while
which
i^,
is
of degree
{x-ni)...{x-
j)
np_i)
+ 1 + 2/i,
is
[277
equal to
(x-bi)...{x-
(x-di)...{x- d^+O
6^+i)
a), \(s'
a)
can be
where
+ ;>'+
are given
also, in
taking
all
by
4(a;-e,)
possible
odd half-periods
and
(I
two
201) that
all
many
^fij,
j',
all
possible sets of
n^-^.
...,
i,,
results as
p+1
2/jl,
+1+
2fj,,
We
E l^^
- ep+,_2^);
(a;
have in
corresponding to
fact,
/i
= 0,
1, 2,
...,
=-
^iViHV^HIt!)-
It has been shewn ( 273) that the expression or (x, z) may be derived,
by proceeding to a limit, from the integral TQX Hence the formula that
has been obtained furnishes a definition of the theta function in terms
of the algebraic functions and their integrals, and has been considered from
this point of view by Klein, to whom it is due.
After the investigation
given above it is sufficient to refer* the reader, for further development, to
Klein, Math. Annal. xxxii. (1888), p. 351, and to the papers there quoted.
Ex.
i.
^ (' -
<r),
<r')]
ii.
In the notation of
200, Chap.
XL,
2 D>
t=l
4r+2
l> *
...
Up, Uj is
V'
4r+4 ,
S V'
v>
of the places
a,,
4r+3
V
.
1=1
t=l
i=l
...
a^,
c, Cj
any one of
an arbitrary integer
,..., Cp, a^ is
and r
is
278]
IS
437
GIVEN.
whose least value is zero, and whose greatest value is given by the condition that
be greater than p+ 1. Prove also that each of the sums
of.
the examples of
cannot
303 (Chap.
XVII.).
Ex.
By
iii.
putting nj,
e(g^'''
...
taking v=p4-l,
^t
...
Mp-i,
h,
+ i^i'i +
+v!^-%)
a, a,,
...
replaces log
CT"
\Xy
are, as in 203,
ttp
= 0,
Up,
_-Jy\r{x) w{x,
+ ^p'''p)~N/^/r(2)
e(v-''+2f=i.''i+
for a^,
^r
ttt)
j^
C7
!-.
a)
y\t
P {x-Xi ){z-a^)
= {x-a)...{x a^,
{x)
...
Xp-,
and
By
d priori.
Deduce the
e(v
,33,
log
,,Xi
J,--!
e(j/' "-?;^""'-
+1'*'" "", Zi ()
...,
^p, conjugate to
fi.Oj
results
3^1 /ci^i
dui
.j;, z
dx
Xp
dui
dt
a.
-i;^'P)'
dx
dt
'
J-',...
X.).
I,,
tt
.,Zi
_2,%>-p) e(jr^'"--y'
xi
278.
''
li
,".__
_a;
where u=if'-''' +
''
the substitution
{x-Xi){z-ai) ^-n'
i,ai=e
(x-ai){z-Xi)
-Tf^'^.
.+n:^'''-''
finally
Xi)
(2,
= iP'^,
result
x-ai^i{x-a>i{z~Xi)^
ot(z, a)
z-a
i,
ifl,,^)
(''
187)
+ in,, ^)
/-v
iris's*'^'
2iri
X y
SS(Ai + H<)r.'
'
w {x,
z),
Mr
nw(a;, ^i)nU(a;,
1
this
may be put
l-i
Oi)
-
"1
["
n
1
Br(a?, Ci)
-^
"^
438
Ex.
i.
y =(^)
if
Ex.
or
2 is
follows
it
z)=^{x z)e^
x,z,
(j>{x)
if ^, ^i, ...
= {x-k)
and
{x-ki)...{x kp)
and the equation associated with the surface be y'^=f(x), where f(x) = <^{x)
we take places x, Xi, ... Xp, z, z^, ... z,,, such that
,
,x,
then
X,
z,,k,
and
Zp,kp
z,
s is
\T^-'
yjr
(x),
and
if
Xp,kp
,Jri
it is easily
273 that
a branch place.
ii.
l)2p + 2)
to the place x,
SI {x,
unless
[278
v<^
{X) y,
/in
...,
(.2^)]-J-[y<^ {z)
Xp&a
zeros
s(f> (x')'],
z.
iz)^-Jw)^^^) _-^K.
.'..
+x;!.;
2 ilf{x)f{z)
Ex.
iii.
a, Qj, ...,
so that
( 200),
gn^'^
+n^'^
--i,'^."P)e(i)^-''
-'"-
''"''"'
+2)*p> "
= *> "+^Q,
and
is
-/p-'^)
-,;%%>)'
-i;%'.P)e(i;^'''-'i;^"'-
t)"^"**'-!-
-i-?/^J"''J'
= ir'''' + |Q,
+2/^''^ = '/'''+^C,
if"'^^-^
eQQ)
The
results of
Ex.
f,
yj,
...
V- (-,
if z, f, Cj, ...
Cp be arbitrary places,
y^ are coresidual with the places 0, o,, ...
<r
^,
...
cp)=
.^^^'^l
MKU^
is
279.
an arbitrary
A^.'^.""
and
y,,
Cp,
then
...
yp be such that
^K%.
We
rational functions
iv.,
+?^f.;
place.
The theory
point of view.
where a
2{x-z)4/_m/(z)
i, ii, iii
Prove that,
iv.
the places
examples
'J<j>{x)yl,{z)+s/<i,{z)^{x)
-^(^.^)
279]
439
(6,), (6/),
of the polygon, to the two sides of a cut (7) are to correspond two sides of
the polygon
Fig. 11.
C be
Then
it is easily
seen that
if
Riemann
surface,
circular arcs
t-C
?!"
t-
where 27r/l is the angle G of the polygon, (l being supposed an integer other
than zero) as we pass from a point ^ of one of these sides to the corresponding
point of the other side, the argument of the function [(f- G)/(^ B)]' increases
by 2Tr if therefore t be the infinitesimal at the corresponding singular point on
;
the Riemann
surface,
1
so that
we may
t,
(^
G)/{^ - B) = t',
^-C=f(G-B){l-f)-K
Further
if ^,
^ be
corresponding points
Math. Annal. xxi. (1883), "Neue Beitrage zur Eiemann'schen Functionentheorie "
Math. Annal. xli. (1893), p. 4; Bitter, Math. Annal. xliv. (1894), p. 342.
+ See Forsyth, Theory of Functions, chapter XXII., Poiocarfi, Acta Math. vols, i.v. We may
suppose that the polygon is such as gives rise to single-valued automorphic functions.
Klein,
Bitter,
J Supposed
440
d^ = -^{C-D)ti~'dt,d^
= j(G-L)f~'e^^^"ht,
[279
or d^'ld^
t,
= /^^^''^^
i_
whose
first
We
term
power
series in
or (^ C)/{^
D),
is unity.
now
shall suppose
..., Xj,
Then a function
given by \i = rrii/li, where mj, k are positive integers.
whose behaviour near Cj is that of an expression of the form t~'' tf), will, near
Gi,
behave
times.
the k singular points Ci, ..., ct, m others, ei, ...,em, for each of which the
numbers \ are the same and equal to e, so that, if t be the infinitesimal
at any one of the places e^, ..., em, the factorial functions considered behave
. .
of the f-plane are also cuts passing to the interior of the polygon, as in the
figure, so that at the point
Riemann
of the
Factorial
surface, ^ is of the
^,
functions
differential of
p-\
first
?!i,
..., r)j,_i.
zeros
top.
of
^=y|.-*i(-a"--?"-^^!"^
where
a,
are
arbitrary places,
this function is
first
e
(?-
c,
nowhere
order only at
?=
oo
infinite
;
at a singular point
OA?-
-D),
whose
and p
is
determined so that Z
is
not
or
first
for
term
is
is
expressible
unity.
The
(ci),
...,
as a
it
vanishes to the
power
series in
<'
or
one another by a substitution of the form f ' = (aS + i8)/(7i; + 8), wherein we
suppose aS-^y = 1, it follows that Z^'jZ^ = y^+B. If then we also introduce
AS AN AUTOMOBPHIC FORM.
279]
the function Z^,
^Z^,
we have
for
441
the form
Z^ = r^Z, + hZ^,
Z:
= clZ, + ^Z,.
f = K/zl
where
is
+m
singular points,
and
iJ=27?ie/jo.
1
At a
singular point
(f (7)/(f D),
Ci,
multiplied by (^Gi)i;
is
or unity
at a period loop
where
/x.
is
it is
fij,
or
t^
or
its
JEi,
multiplied by
the factor of K.
series in
at a singular point
behaviour
power
fj,
(y^+
S)~^,
number
is,
is
one of the
finite
is
a rational function of
The equation
fj.
connecting the values /', /, of the function /, at the two sides of a period
loop, may be put into the form
The
function
difference
is (
in the variables
factorial
254)
Xx..s(-5)_....(-.)-i,^,..(-^-)-ii,,-I,-is(l-^J,
at
adding the proper corrections for the zeros of the automorphic foi"m
Functions,
we
have,
Theory
(Forsyth,
645)
Oj,
p.
of
the angular points G^, ...,
for the excess of the number of zeros of the automorphic form over the
number
of poles
"
1
m+l-(S7+m
+
'
2x + 2^' = -f 2p-2+A: +
R 2o-2 g-S+
2
where o =fc + m +
We
*
may
l,
2- = 27-+m+l.
known formula
The quantity R
is,
for
automorphic
- 2m.
442
functions
m(n 1 S-),
n
Theory
Forsyth,
= 2iV" - 1 +
g',
there given,
we
of Functions,
we
substitute,
648
p.
by the formula of
m (^N- 2 + g - S -
obtain
a cycle by
in
if
for
[279
formula
the
608, 293,
p.
Em
a cycle
points Ci,
...,
by
itself,
(cf.
Forsyth, p. 596)
the
sum
Ci
is
the
itself,
sum
There
may be
is
k cycles
sum
is
is
27r2 ^
of the angles
is 2ir*.
way
...,
in
is
= a;x\;
= Z^x^
may
^'"
say, is
Thus,
which
is
(i),
e^, ...
for
,
e^
the cuts
and may
(ii),
where
Shew
a, c
. ,
(e^), or, as
put
is
fi
may
homogeneous
we
^fj-
a homogeneous
Riemann
Hx.
. .
also
automorphic form,
sideration of
(ci),
factorial forms.
is
unbranched at
M:=2ir+^{^,-q)-^-^
(,.. "
is
ej, ...
e, that
to be chosen so that,
z is unity.
At
it
when x
)" where
+,^. ) + ^2
x,,,
.;>%,,
- q denoting the number of circuits, made in passing from one side of the period loop to
the other, of x^ about X2=0 other than those for which x encloses places ei, ... , e^, and r
q'^
z.
The formula
is given by Eitter, Math. Annal. xliv. p. 360 (at the top), the quantity there
denoted by q being here - 4 p. We do not enter into the conditions that the automorphic form
be single-valued.
t The reader
will
Eitter,
The
quantities denoted by
tr
It may be
p. 291.
here used, does not coincide
+ l,
y-l- 1, as
him by
if,
tr'
to
correspond respectively to those denoted here, for the factorial system including the singular
points !,
...,
e, by o-'+l
and
la'.
281]
CHAPTER
XV.
280.
As
Introductory.
of several variables,
any adequate account of them a matter of difficulty in this volume the plan
In the present chapter are given some preliminary
is as follows
general results frequently used in what follows, with some examples of their
The following Chapter (XVI.) gives an account of a general
application.
method of obtaining theta relations by actual multiplication of the infinite
series.
In Chapter XVII. a remarkable theory of groups of half-integer
characteristics, elaborated by Frobeaius, is explained, with some of the theta
relations that result from these the reader will perceive that the theory is of
great generality and capable of enormous development. References to the
;
adopted
literature,
characteristics, are
Let
<f>{ui,
...,Up) be
a single-valued function
of
independent
variables Ui, ...,Vp, such that, if a,, ..., ap be a set of finite values for
...,Up respectively, the value of ^ (ifi, ..., Up), for any set of finite values
Ui,
of Ml, ...,Mp,
is
by a converging
expressible
positive powers of Mi
- Oi,
itj
aj,
...,Up
ap.
analytical function.
series of ascending
Such a function
<j>(ih,
.,Up) has
is
for
integral
an integral
each of
if
its
m be any
<l>(ui, ...,
Then*
Ua
+ m,
...,
...,
Up)
= <j>{iii
Up).
by an
infinite series of
the form
'^
n,=
*
-<x,
np=-m
-^Ki
njj
<^
>
444
[281
wherein
n^, ...,np
positive
and negative
values,
u^, ...,
^ is
and An,
independent of Mj,
...,
Up.
Put
a;i
,p)
Xp
e^'""", ...,
= e^'"^;
then to
a'-plane lying
values of m for all points within the annulus Ta may be successively represented by such series; the most general value oft/., for any value of a;, is
of the form Xa
Up)
<j>(ui, ...,
where
m is
an
integer.
which
Xp,
Xi
+ m,
is
is finite
and continuous
denote
it
by
i/r (ajj
. .
by points
So considered,
integral
P'
""
the variable ta be taken counter-clockwise round the circumand clockwise round the circumference 7a (a =1, ...,p); when ta is
upon the circumference F. put
(i)
let
ference r,
= S
ta
when
ta is
ta
*a =
ta
A +1
'
</
ta-Xa
then the integral
is
^Xa
Xa
Xa
'
*a="*a
equal to
EXPONENTIAL SERIES.
281]
445
the circumference 7, since the component series are uniformly and absolutely
convergent, this is the same as
;
where
00
QO
...
An^
p=_oo
ni= 00
where
and the course of integration for <. may be taken to be any circumference
concentric with r and 7., not lying outside the region enclosed by them
let
(ii)
whose centre
the variable
is
of radius p,
circle,
...,); putting
\,
4 = a;. + /3e**a,
we
The values
we have
..., Xp).
<l>{u...,up)=
...
where
,^=|
Jo
By
+ ...+n,,)0(y^
...f e-2(",.
Jo
values of
2*1
..., tip,
uniformly convergent.
is
Up)du,...dUp.
absolutely,
If m.
= fa +
where
Ex.
ii.
p),
and
:)
Wa, we have
Vp
+ iw^)
..
In the notation of
174,
gi7r(Tn,2+...+2Tn,i2 + ...)^
(v^
+ iWi,
...,
dv^
...
dvp = 0.
Chap. X.,
/
g-2(n,, + ...+np,)
.'0
(^^^
__^
^^^
^) ^^^^
'0
interest to
Prove that
jg-2(M,e, +
g^ I
Ex.
...,
and
*w (a
also find
177,
it
of
and
446
Further
282.
it is
necessarily unique
a;i^"'~
...
centre at x^
...,
Xp'^~ dx^
= 0,
...
^xi
-4,
...
is
=0,
by
...
[282
dxp,
...,
,,
be
For multiplying
zero.
this equation
of radius lying
^ = 0.
{2-niyA,
it is
obvious
now
For an
equation
ni= 00
s=l
or
i
where
'iiriUi,
...,
np=-oo
is
...
1,
and
k,, g,
denote rows of
= 0,
^^e^"^(^^,^,+...+tr,if,)
Aif,
hu,
A^^
j^^
is
given by
a-
where the summation includes 2^ terms, and N^, ..., Np take the values
arising, by the various values of n and kg, for the quantities 2n + kg; it is clear
that the aggregate of the values taken by 2n + k, when n denotes a row of jj
unrestricted integers, and k, a row of quantities each restricted to be either
or
1, is
Hence by the
n and
values of
Np = 0, for all
of arbitrarily chosen
Therefore,
kg.
or
g-6(n+J*,)i>+wX(n+it,)
1,
\ denote a row
if
we have
jl^
^ =2
0,
|n+i*)
i, e''W.+*)
=Q
adding the 2p equations of this form in which the elements of n are each
or 1, the value of kg being the same for all, we have
either
2 Gg,, k, ei*''*>'.+'')
where
/xi,
...,
fJ-p
[1
-I-
e*""]
. . .
[H-
...
(1 -f
e'"'''')
is
e""'-],
fi
given by
fjLi,
(^
..., fip
= gg + \;
are even.
2^'*]
that
447
283.
2=*
have
( 190), if
m and q be
a (m
and therefore
-I-
theta
may
with half-integer
we
integral,
n; ^q) =
eA.(u)+(m?'^'j)
if
functions
k',
^ (j^
Q=
1^^^
k,
Therefore a relation
leads to
where Q,
2^
the
= q,+k,
Q,'
= 9/ +
k'
and m'
is
either
or 1
then as
= [1
2e(.'-'e.),
is
+eic.'),]
[i
...
+ g^ie-'ip]
that
is,
= k, q/= k', we
unless qs
2 2
Ce"""^''-"'*.)
^ (m
284.
We
suppose
now
(Q,%
that
^q,)
not zero,
of %,
(iv)
...,
(ii)
(k, k'),
we have
e-^(Q,),] ... [i
...,
C* =
form
is
0.
Thus
four matrices a,
has
2aa>,
the coefficients
all
to', 17,
rj',
each of jd
(i)
symmetrical,
o)~^od'n^
Jifc)
(iii)
its
-f-,
1}'
+ e-(Q,)p-|
an even integer,
(Q,), is
2=*Ci^ (u
have
thus, for
in the
[i
17'
2a&)'
h,
hca
^tri,
hw = \h
+ Which requires that the imaginary part of the matrix w~'u' has not a vanishing
terminant.
de-
448
[284
as in 190
Xi (w)
where
Em = 217m +
We
have shewn
27]'7n',
= 2fflm + 2a)'m'.
D,m
equation
^ (m +
m
m,
fi, q)
= e*<)+2-ri(w'-''9) ^ (u,
q),
and m' each denoting a row of integers; it follows therefore that, when
m each denotes a row of integers, the product of r theta functions,
satisfies
n (m + n^) = erX^luj+s^img-mQ)
wherein
Qi, Qi
are, for t
l, 2, ...,
p, the
integral,
and
Q,
of order
r,
o)',
g'*'"'.
11 (u),
which, for
17,
rj',
integral
all
We
sums
5<", ...,
to obtain the
integer,
JJ (m),
Q' denoting
we proceed
quantities,
gW),
^ (m,
5<^0
values
of
suppose r to be an
conditions
and
the characteristic
(Q. Q'}.
(it;
the
first
189, either
teristic; increasing
by S{u; 2m,
2a)/r,
'ion',
2i;, 2rij',
and
by a factor which characterises it also as a theta function of order r, with the associated
2(u, 2cb', 217, 2i)' and {Q, Q) for characteristic.
We have, also,
constants
5(m;
= $(u; ^,
2a>',
Let
ki
2,,
is
ZrA^S fru;
2oi, 2ro';
^,
2r,'\
= 3fru;
-,
h,
rb\
be the least positive integer such that kiQ/ is an integer, =/<, say
p rows and columns, of which every element is zero
;
Rend.
t.
zij.
(1855),
and a
by replacing
letter
ki, ...
by k; the
by
respectively
ibid.
t.
xlth.
Schottky, Abriss einer Theorie der AbeVschen Functionen von drei Variabeln (Leipzig, 1880), p. 5.
The investigation of 284 is analogous to that of Clebsch and Gordan, Abel. Fund., pp. 190, f.
The
(1865), p. 28.
2^*]
llh
the
ill
;
449
r.
u introduce arguments
v determined
^.i"i +
0'= 1,
+Kp'^p = hvi,
which we write hu^kv; then, by the equations
Ao, = \-^i, h<o' = ^b,
that the mcrements of the arguments
v when the arguments
kUjn
denote
shall
the
Now we
follows
it
= jrim + hm'
right-hand
side
of
by
equation
this
Urn
...,p),
u are increased
rows of ?/ defined by
we
by
equations
T,
thus
have
(m
ft
2hm'
this
is
equal to
2hm') {u
^n,)
to
\n
(w)
or
^m
thus,
by the
ll(u
now
and
let
m'=
irimm'
Q (v)
+ n) = g-^""' n
denote e-' 11
Q (i; +
mQ;
2fo;m'
+ T^m'
g-raiu+CU)'
therefore, if
(m)
0,
f/m)
(m)
we have
e-^[''>mm'+2(t+jY)m']+2(mQ'-n.'(J)
(u),
= Q (v) e-'-['"'""*'+2(ii'+iV,)m']+27rt(mQ'-m'Q)
s denotes a
=ksQ = kyS^Q^'
+ kpSpQp = skQ' =sf,
=
=
'7rik~^m + kr^bm'
iris
thus we have
Urn
-ir
Q{v +
iris)
= Q (v),
or, what is the same thing, the function Q{v) is periodic for each of the
arguments Vi, ..., , separately, the period being wi; it follows then ( 281)
that the function is expressible as an infinite series of terms of the form
^e2i"i''.+ -.+pp), where 7I1, ..., rip are summation letters, each of
Cn^,n
+ 00,
C,,...,np are
Q (v) =
To
*
row
By a fundamental matrix
letters
B.
ofp elements,
11
e-^"'"
all
independent of
v^,
...,Vn.
equation,
if
/i
00
to
This we
n (w) = 2Ce^.
u d = JS v
be any matrix of
coefficients
u,
u.
29
v be
EXPRESSION OF A GENERAL
450
we apply
(7,
(v) of
[284
Q{v+ Um)
2(7
+ k~^Tm)
e2(+*-iYm)n _ rV(7
or
Q(v
grlnmm'+2{kv+iYm)m';}+i7n(mQ'-m'Q)
g2i"|
thence we have
summation
summation
such that
Si
thus s = n
we
-I-
IviTii
(i
= 2vn +
2rkiVimi, or 2vs
1, 2,
...,p);
2rkvm'
hence
obtain
therefore,
quantities
and
. (J
g2*-'Ym+>-(irimn'+Ymm'l w(mQ'm'Qi
By
the
eF"', ...,
are
= ni + rkiml,
denoted by
n,
m,
m.
taking the particular case of this equation in which the integers m'
we infer that the quantity
zero
all
- k-'Tn - mQ',
= . k-' (-Trim) n -
mQ',
1^ m,
[j
n,
- Q/
must be an integer
numbers (?), -kgQg')/kg are integers; thus, by the definition of kg, we may put
n =/+ kN,
denoting a row of integers, and /= kQ'.
With
k-^Tr^n
this value
we have
k-^ {"Trim) n
= (fc-'/+ N)
hence, as
giBrirmm'
mq = k-'mn-mN,
ijgjjjg
bm'
it-'n
bin'
= {Q' + N)
bm'
is
therefore
-lb{mr + N)'
e
Tq being
ttiQ
'^
+ bQ',
Cf+kimr+N)
=e
-Hn'^
'
Cf+ty
e^Ycm
_
or
^/+k(N+rm')
=e
i^/+kN',
= bQ'm'
284]
451
r.
rn
= D^
Dtf+rm'
by
D}^,
we
have, for
is
equal
N=
+ rM,
p.
may
_,
,t.^.,
r.
i(6iV2+2Y2V) + 2i<Q'+A')li
r*(jf+^y+2Yg(jf+^)+2Au(rJf+'+(')
for
term
is
now, as Tg
iriQ
+ bQ',
and 6
is
rb[M+fl+2T,{M + t)
is
R (u),
or e"'
is
where
e*+x^
and
Q (v), with
^J +
2i
[m + i^)
before represented
denotes a theta function differing only from that
h respectively into
b,
a,
matrices
the
of
change
in the
( 189, Chap. X.) by S^,
series
of
the
convergence
^ requires that r
ra, rb'rh; the condition for the
where
be positive
we
thus p
see, as already
= r;
remarked on
p.
292
EXPRESSION OF A GENERAL
452
6)
77,
r?7
r
with this notation then
we may
[284
are
n(w),
wherein
Ky,,
a row of
= D^e '"''
is
an unspecified constant
From
this investigation
an important
ft
coefficient,
/j.
denotes
r,
possible values.
all its
corollary can be
drawn
if
a single-
(p.
448), in
which r
is
Q, Q' are rational real quantities, be called a theta function of the rth order
Q'),
then*
aiiy
r^+l
theta functions
teristic,
It),
and
Stj'
same charac-
integer's,
are connected
linear
equation
or by
we have
Q'),
and
+ N' =v (mod.
if
say
r),
ijl
+ N' v +
rL',
= e'-"
^{u- Q
Qf
+v
and therefore
S^.^
(u;Q^N,^^) = ^H,^ (u
where H, = K^e^'"^~l^
r^
+1
Q'
The theorem
is
^^)
rf values of
is
the
+ t^
r-th order, with the same characonly by integers, and associated with the
Q,
attributed to Hermite
of.
Compt. Rendus,
t.
285]
same quantities
2o,,
2a,',
i.
(m
Q,
Ex.
all
^1-^) with
coefficients
independent of
follows as enunciated.
functions
j-p
are
2,,',
77,
453
r.
^{u;
^^\
Q,
Ex.
u.
The function (u + a; Q) 3 (u-a Q) is a theta function of order 2 with
(2, 2') as characteristic.
Hence, if 2p+1 values for the argument a be taken, the
resulting functions are connected by a linear relation.
;
Ex.
(a)
^^'^''
- 6)
ff
+ 6) - 0-2 (6)
(m
The function S
iii.
characteristic.
"'
cr (tt
Prove that
Ex.
The product
iv.
/*,
17'
(m
+ a) =
respectively, then
, are
o-'''
(m)
o-
(a
- 6) tr
(a +
6).
row
letters of
p elements
than
all less
r,
and each
5(m + m(');
is
o-
u,', rf,
- a)
{ru,
if
place of a,
?') ill
'
o- (tt
.9(m
(!))
+ kW;
<'))
44
enter as characteristic.
^=1
r=l
(T (ti
+ Ki),
1=1
V.
Any
(jo
+ 2)
^=0, where
five functions
II
1=1
(M + Mi
(r
r),
17'
r,
for
The
following,
+ +u+
it j
ci)'),
iV, is
(T (
great enough.
11
o)'),
^i+^*^+3 + A
is
2=1
associated constants a, a,
number
rr(M + Mi + <B),
v=l
Ex.
r=l
r=l
is
for
when
For the
is
taken
(p + l) theta functions
X (m
P,
ANOTHER PROOF
454
[283
where P, P' each denotes a row oip arbitrary quantities. The corresponding
quantity arising when, in place of o), ', r), r( we take other matrices m'",
?;i", ;'"' may be denoted by \'" (w; P, P')With this notation, and in
ft)'"',
case
ti>>,
b'<",
7?'^',
V"
rv
are respectively
,
where r
an arbitrary positive
is
-\
N, m,
we have the
integer,
following identity
=
s,
Jy.m
where
\<'>
m',
[m
+ ri"'
/t,
0]
+ X<"
[m
TO,
+ rN = 'm+
k;
TO^
- \"
[h
s,
- lirMk,
0]
the relation
s
it is
ii.,
190.
/) =
5:e--''*("+"F)-2"-^r
u(u + 2<o^,
=
Ze
arbitrary quantities,
and
now the
n
therefore, if ni
IT (m) denotes
w + 2a,^
+k=
+ nS:'
n
where
by
/)
2<b("s
=n
(m
2&) "' s)
by
+ 2<BiV-le*"^
definition
m + k = s + rN,
[m
P (w +
is,
r,
a,
and the
row of p
r.
we have
2a,'TO');
we
have,
by the
defin-
II (),
M+2<B-+n':' =
r
:n(^M + 2a,^^
k thus arising,
P (u + il,
value of
[u+2q)
\
summation extends
Consider
/)
2(b'to']
[+2""'s
If,
m')+27ri
{NQ-m'Q)
and hence
= 2e*n {u + 2<i)i')s),
s
in
which
,|r
= - XI"
[m
+ n^'
*, 0]
+ r\ [m + 2a>w s
i\^,
m']
jw'Q)
OF THE EXPRESSION.
285]
455
= \m [u
-jr
= \<
[u
m, m']
- \w
m, m']
[u
\<" [u
s,
0]
s,
0]
- 2Trim'k - 2mmQ +
27rim'k
2-iriN {Q'
-/)
_
-m
'
+ 2mf
r
we
iJ(M+niV;/) =
have, since,
further
if
'^e
>
'n(w+2<oi's)
=e
Hence R{u\ /)
satisfies
by
satisfied
respectively for a,
a',
T], rrj'
R{u,; f).
form as that
,
'
'^
11 (u),
w',
ij, t}'
R{u; f)
so* considered
r
function of the
first
we
Kq+^
is
a quantity independent of
matrices
is
a row of
u,
and
is
wherein
yu,
448) where
(p.
where
/i,
a theta
putting, in ac-
as characteristic;
/= Q +
is
fi
denotes a row of
series
r,
and
when
sum
\%Riu;
which, by the definition of
^.o-s^v^
R{u;f) may
2w, 2a',
2ji,
2rl
and
+ f^),
r^'"<'
Q'
jj /
is
equal to
2a,*) Se-'"''*r1
1^,
r,
456
in fact, equal to 11
is,
(u).
Hence
[285
as before
+
n(M) = 2^^+,a(; Q,^-^).
Q'
Ux.
286.
Suppose that
i.
an even half-integer
is
^1
characteristic,
and that
^s
by adding the
Thus when m
is an integral, or zero, characteristic, the condition is that the characteristic formed by
adding two different characteristics a,, a, may be odd. The characteristic whose elements
are formed by the addition of the elements of two characteristics a, h may be denoted by
a+b
when the elements oi a + b are reduced, by the subtraction of integers, to being less
are
s,
2'',
a,-
is
not equal to
i is
_;.
characteristic
zero),
a,
/3,
if
fi
/I,
be denoted by ab.
may
by taking m,
equal to
a, /3,
y and
is
therefore odd.
When p 3,
other characteristics
maK,
a,
/3,
y, <, X,
^,
fi,
toX/j.
XVII.
...,
resi^ectively equal to
TO,
m,
a,
(3,
...,
1^.
Now, by 284, every 2^-1-1 theta functions of the second order, with the same periods
and the same characteristic, are connected by a linear equation. Hence, if
p, q, r denote
arbitrary half-integer characteristics, and v, w be arbitrary arguments, there exists an
equation of the form
We determine
is
of the second
and of characteristic q + r.
the coefficients
A^hj
for m
5 {u + a
= e^(-
j)
^)-2''-^3
(M
P+q),
where
noticing,
what
is
verily
204
2S6]
we
OF THE THEOREM.
457
obtain
A[v, + w;
=
But
m-a^ + a^
since
characteristic when^'
is
we can hence
X,
infer,
an odd
is
shapes
different
momentarily by
is
k, so
according
need.
to
a^+p momentarily by
a^,
half-integers,
and similarly
M,
for integral
g),
we have
3[2v;
we
^-4K-hp)('-l-9'+r')5|-2
hence the
final
by
.
(i+j + ^)]
k);
e*"("''+''^("''+'+'->
.9
-1-
r)]
5 [0
[ + ,;;
i]
(q+r-p~a^)]B\u-v;
{p + aj\
5(0;
mai/Xj)
vanishes
206),
Ex.
ii.
When p = 2,
q, ,;
m,
a, ft y,
we can
in Ex.
i.
take
respectively equal to
a/3y, q, 0,
denotes the characteristic of which all the elements are zero, and
wherein
the reduced characteristic obtained* by adding the characteristics /3 and y.
and retaining the notation
general formula of Ex. L becomes, putting v =
/3y
denotes
Then the
for the
characteristic a/3y,
e47ri(<i^
+ in)(m' + 3')5(M.
q-m-a\)B{u;
m + a^)9{w;
q-af.)3{w,
A=l
*
So that
all
the elements of
^7
and
less
than unity.
ax).
458
Ex.
and
As one
iii.
ii.
[286
we put
therefore
-*P'
hence we
<-*(")'
find,
5 (m
where M, =(^^}^')
204,
/+ if
= e2"*V' 3{u;f),
consists of integers,
f=(ff/)>
Mf^MJ{-^MJi,
and
that*
^(m + w; j)=-52(a + K'), 5(m-7; 0) = 55(m-w), S(0; y + TO) = .9,2(0), 5(0; ot)=5oi(0).
5('M;5-wi-ai)= 5,2(M),5(M;m+ai)= 5i(u),S(w;j-ai)= -^Jw^aCw; ai)= ^j W,
denotes
5,2
-512
W>
5i4(w),
5o4(w),5(w;a4)= -SjiCw),
;
by substitution of these
52( + w)
where
9oiW;^('^; "2)=
and
3,^,
denotes
3f,^
{w), etc.
Ex.
By
iv.
ii.
respectively
is
in
iii.
by that of Ex.
iv.
we obtain an
is
Ex.
V.
5+
starting with
as the
a,
given characteristic
q,
we
ways
express
it,
a,
ft
to obtain
an expression
m and
the characteristics
Thus,
sum
in three
which
in different
sufficient that
is
?=^(io)'''^^^'"'
* In Weierstrass's reduced characteristic
t This
is
cf.
203, 204,
and
is
positive
and
p. 337, foot-note.
for instance
X For example, (ac) = (0,0) + (a jo) {c^c^) = (a^a^) + (cjc) + {cc^). See the final equation of 201.
six odd charticteriatics form a set which is a particular case of sets considered in
chapter XVII.
Moreover wc may increase u and w by the same half-period. But the additions of the halfperiods P, P + Uq lead to the same result; and, when q is one of o, /S, 7, the same result is
-1-
The
-t-
fi^ "<! o{
P + tl^+iin.
286]
(\
putting TO= J
^=*Gi). V'Kn)-
wherein
u,
459
S.
result
p=
we may take
2,
-903 (i")
iii.,
{u),
etc,
and
52
denotes
\, {w\
etc.
Comparing
namely
identity
by
Ex.
vi.
By
taking in Ei.
v.
the characteristics
*
and resolving
q,
m to
be respectively
x.
vii.
those in Exs.
1 01 12
w = 0, we
by taking
for
3ff^{u
+ w)
3^(u w) which
the characteristic i
^^
>
arise, similarly to
unaltered.
.Er. viii.
is
as in Ex.
v.
For tables of such formulae the reader may consult Konigsberger, Crelle, lxiv. (1865),
and ibid., Lxv. (1866), p. 340. Extensive tables are given by Rosenhain, M^m. par
divers Savants, (Paris, 1851), t. xi., p. 443; Cayley, Phil. Trans. (London, 1881),
p. 28,
p.
834.
460
Ex.
We
ix.
i.
[286
to the case
jo=3
taking the
,Ug
argument v = 0, the characteristics^, r both zero, and the characteristics m, fltj, a2,
where a, ft y, k, \, fi, v are seven characteristics
to be respectively m, m, u, ft
such that the combinations a, ft y, k, X, /*, v, ma^, maK, iX/x are all odd characteristics,
m being an even characteristic, and removing the negative signs in the characteristics by
,
./,
such steps* as
3{ w,
may
q + m must be even ; no w it can be shewn that every characteristic (q), except the zero
characteristic, can be resolved into the sum of two even characteristics {m and q+m)
in ten ways,
a set as
a,
k,
0, y,
X,
/i,
v (cf p.
309, Chap.
XL).
and
to,
effectively unaltered.
In
all
is
m = 0.
taking
Ex.
Taking, in Ex.
X.
when
of 205,
1, 2,
7,
a, ^,
ix.,
,
the characteristics
q,
that
Prove
iu)
(>)
also,
taking
that Si^{uJrw)9
{u-w)3i^3
is
equal to
5^ () 5,58 {w)Jrh {u) 5, {w) 9^ (u) \, {w) + 3^ {u) 3, (^v) 3^ (u) 3^ (w)
+ \W\{y>)St,iu)3t,{w)
^123 (-)
() ^2 (>)
K^ () ^317 (>)
- h () ^3 (w) 5, {u) 3^
where
5, ^45^
denote respectively 3
(0),
3^^
(w),
(0).
for 3^l^(u+u>)|3{u
+ w)
in terms of theta
Ex. xi. The formula of Ex. i. can by change of notation be put into a more symmetrical
form which has theoretical significance.
As before let m be any half-integer even
=2f, half-integer characteristics such that every
characteristic, and let a,,
a, be
,
is
<f,
is
written
\P1PiP3
and
287]
combination
mtajO,-,
which
in
i is
not equal to
let
j, is
J denote
461
an odd characteristic
let /, g,
h be
111),
.T=\{-\
1-1 1 1
1-1 1
1
1-1
1
1
u, v,
V,
W,
T by
the
or, as
= J{%H,
Vi,
Wi, 0),
(i
= l,
2,
p),
W,T)==J{u,v,w,0);
(U, V,
further determine the
new
characteristics F, O,
H,
form
{F,0,H,K) = J(J,g,h,m),
noticing that there are ip such seta of four equations, one for every set of corresponding
m)3{u;f)3{v;g)S{w;
5(0;
i.
that
h)
2 e*"'A\+2A'-'"+''+*'+'"')5(7';^-|-aj,;S(f/;
= I
e*A"''3(C^;
Putting
7?i
= 0,
we
x=i
wherein
u, v,
g,
Deduce from Ex. i. that when p = i there are twenty sets of four theta
them odd and one even, such that the square of any theta function can
be expressed linearly by the squares of these four.
Ex.
xii.
functions, three of
287.
The number,
r^
of terms
in
the expansion of
11 (u)
may be
arguments
u.
We
characteristic (Q,
For,
^^^^^
if
prove
Q)
first
is
that the
consist of half-integers.
e'"'-'"'
+=*^i'"'^'>
eD (- u),
din Variabeln
(Leipzfg, 1880).
462
while,
the
left-hand
equation
of this
side
[287
is,
equation, equal to
gg-A-,(-M)-m(Tn^(' m'Q)
hence, for
all
r [X (u)
must be an
y\
JJ
m'Q)
H-in (
m'Q)
(mQ'
be an integer
u),
integers.
11 (m)
Q,
= ell
m),
where
is
denote them by
-1-1
or
1.
q,
q'
and
equation
U{u) = 1.K^^(u;
since, for
any
characteristic,
^ (m,
n (u) = en (- u) = etK^^
g)
= ^ (
(-u;
V
u; q-2q,
q,-~),
q,
q'
^-
u,
we obtain
q),
2q'
^ + v+ 3_
- fi
(1
where
i;
is
vs-(/i +
2q'),
of
= e^^s'a
fi
(m
tK^t
therefore,
q, q'),
\u
by the
wherein M,
q,
M'
is
less
than
(mod. r)
190),
are integers,
^ (
+ M,
q'
+ M')
we have
n (m) = l.K^e~^
we
so chosen that
formula
for IT (m)
r,
-^
q,
U(u) = ^K,^{u;q,^!^)
the values of
2fj,
+ 2q' =
/m
that arise
(mod. r)
written congruence
is
may be
for
(i)
for
fj,
which occur,
'is
equal to
it is
applied
(ii)
thus,
those terms
288]
463
y+g'
'
for
which
2q'
(mod. r)
where
2/i,
2fi
the
first
(mod.
2^'
in the
term ^(u;
Hence
which
q,
= eZ^e"*"'
r),
is
for
for
/x
which
single term
form
11 (m) is
arising
The
of
Hence
when
even or odd
2fi
+2q' =
and
(mod.
if
satisfy the
r,
of
M+g'
''
these solutions for which also the condition e
number of undetermined coefficients in
(u) is reduced
=e
is
to, at
the
satisfied,
most,
y+i(rP-x).
288.
We
whose number
we
notice that
y vanishes when
is
The
x.
is
result
is
the half-integer characteristic {q, q') does not consist wholly of integers, or
(u) is ^ *
number of terms in
^ [1 ( X] ee*""'*'.
zeros, the
Suppose r
by taking
yu
is
even
M-^ q,
and
no other way,
in
2/i
2q
(mod. r)
is satisfied
integers.
Thus unless
(f
case
x=2p;
according as
'"
the condition e
e is
+1
or
is
by
all
e is
the 2? values of
in both cases
M,
when
or by no values of
= -IP-"- (1 + e) when
;
is
is
M,
not
464
integral,
y=
is
number
of terms in 11 (u)
2^"' terms
is
when
zero),
= 2p,
is
2^-1
e,
11 (m) is
an odd function
(2)
(including
+ e) + ^ (rP - 2^) = ^ rP +
2p~'
2P-^ (1
^rP
M can
the elements of
1 of
e"^"**^ = e
dition
when
[288
integral,
a;
2p,
= 2^"',
and there-
fore the
+ ^(rP-2P) = rP,
2P->
when
(3)
terms
is
wherein
iV^ is
number
is
is
= rM,
)iT
of
fi
= rM-2q'
^i- =
,
integer
M to have
M-2q'
^^
2N, where
-f-
f='-^^*-.=*+,-(.-i);
this equation, since
the value of
fx,
N uniquely
hence x
= l.
The
than
r,
determines
hence the
condition
>'+?'
^
~
4no
determines y = 1 or y =
number of terms in 11 (m)
=e, or
according as
is
l+i(rP-l),
according as eg*^?'
odd, the
immber
289.
-f-
or
= 6, or e"*''9' = 6
ee*'^^' = +1 or = 1
"*"'*'''"*')
or
i(r^-l),
It follows
numbers ^(rP +
if
when
(q, q),
is
1).
we take
r,
pro-
with
integral or zero,
-J-
289]
465
In case ^=2 there are six odd characteristics, and the sum of any three of
even* as the reader can easily verify by the table of page 303. Let a, j3, y,
f, f
denote the odd characteristics, in any order, and let a/3y denote the characteristic formed
by adding the characteristics a, /3, y. Then the product
Ex.
them
i.
is
fi,
n () = 5 {u,
{u,
a)
ft)
B {U,
y)
is
arising
1, 2, 3.
{u, aj3v)
Hence
integral characteristic.
this
v is
chosen so that
/i
+ i/=0(mod.
4).
= (0,0),
less
than
We
4.
(2,0),
of these
(2,2),
must be
(0,2),
/i;
fi
niu)=-'SA^s(u; 0,0;
comparing this with the
may
= SA^^B{u;
0,l^-3:(u;
0,
- 0]
0,^)],
these values
may
= (0,
1),
(1,0),
(1,1),
(3,2),
The number 6
,= -1, p=2.
Hx.
of
(1,2),
(1,3),
(2,3);
(3,1),
(2,1),
is
coefficients A^.
formula
j^rf
+ ZP-^t,
ii.
by putting r=4,
.-,
C,
(u, a)
/3)
S {u,
peC),
e,
9 {u,
y)
(u,
yf), 9 (u, S) 9
(u, 8f f ),
occur in
f be taken in the order in which they
^wW^mW.
*
This
is
a particular case
of
30
466
is
2,
[289
and of
by
integers
f, or,
coefiBcients
thus
do not
B
or,
(,
a80 B {u,
aSC),
B (u,
in Weierstrass's notation,
fiSe)
B (u,
if a, ft y,
B (,
fi8(),
e,
fi,
ydf)
t)
(u, (),
2,
and of characteristic
*f, or,
in the
For
iii.
p=i
the
number
diflering only
linear equation
whose
by integers
(
, ,
coefficients
thus
do not
is ^4^ + 2 = 10.
If j, r be any half-integer
any eleven functions of the form .9^ (m, q) B'^ (u, r) are
connected by a linear equation. Taking now, with Weierstrass's notation, the four
functions*
t=Bi,{u),
it
in
The
coefficients J,
H^
are independent of u.
respectively, to be
GS)^
(cf.
B^
5o (m)
may
be taken,
wherein
B^ {u
q),
B^^{u
z respectively
by
X,
making
this change,
that
Which
It
ia a linear
can be proved that these functions are not
289]
Thus the
467
V/i, fiJ
\0,
we obtain the
where
9t\
Q)
m, in turn,
\0, O)
\h
'
the half-periods
hj - Vo,
OJ
tt
= 0,
three equations
denotes
-95(0), etc.,
The value
functions.
\gu gj
'
of
Weierstrass's, as in
is
220.
By
these
u=0
in the identity
itself.
Thus we may regard the equation as known it coincides with that considered
i. and iv. 221, Chap. XL, and represents a quartic surface with sixteen nodes.
With the assumption of certain relations connecting the zero values of the theta fionctions,
proved by formulae occurring later (Chap. XVII. 317, Ex. iv.), we can express the
;
in Exx.
coefficients in the
We have in fact,
if
.95(0),
b,
^^{0), 5,2(0),
3(,(0).
hence the identity under consideration can be put into the form
d*+(*-a*-b*
'
d'^<?-a^b'^
^^'^
'
where the n 'denotes the product of the four factors obtained by giving to each of ,, t^
both the values + 1 and - 1. The quartic surface represented by this equation can be
immediately proved to have a node at each of the sixteen points which are obtainable
from the
fovu-,
by writing
respectively, in place of d, a,
(i) {d,
Ex.
iv.
-b, -c),
a, b, c),
We have in
Ex.
(ii)
iii.
{a, d,
(d,
-a,
in Ex.
i.
c, b),
(ui) {b,
c),
d, a),
c,
(iv)
(c, b,
a, d).
221
-a, -b,
b, c,
and
-c), (d,
b,
the functions
55(),
9^{u);
%W,
^04().
^i()-
302
[289
EXAMPLES.
468
There are* in fact sixty
these sixty sets break
a relation holds
and
up
of the sets of four consists wholly of even functions while each of the three other sets
two odd functions and two even functions. This can be seen as follows using
the letter notation for the sixteen functions, as in 204, and the derived letter notation
for the fifteen ratios of which the denominator is 5 (u), as at the top of page 338, it is
consists of
immediately obvious, as on page 338, that any four ratios of the form
in which the
letters k,
I,
formed in
ways
fifteen
by permuting the three letters c, Ci, c^ among themselves in all possible ways and nextly
nine such sets in which two of the ratios are odd functions, obtainable from the set
;
and instead of
the pair
CiCj
c,
the
number
of odd
will
remain unaltered.
Further
fifteen sets
the scheme
1
a,
"l^li
aoi
^ai
""101)
a^
a^^
"0ii3i>
^Oifti
where 1, u, ft o0 denote the characteristics of one of the fifteen sets of four theta functions
just described, such as 5 (), S^{u\ \^^{u), S^^^W, or {u), 3-].(u), \^{u), S^^^^{u),
a^ denoting the characteristic formed by the addition of the characteristics a, j3 ; and a, ;3,
denote any other two characteristics other than u, /3, or a/3, and such that a/3 is not the
same characteristic as ojft. This scheme must contain all the sixteen theta functions
for any repetition (such as a = /3ai;3i, for example) would be inconsistent with the
,
a^,
/3j
(would be equivalent to
a/3
= Oj/Si).
It is easily seen,
by writing down a representative of the six schemes in which the first row consists
wholly of even functions, and a representative of the nine schemes in which the first
row contains two odd functions, that in every scheme there are three rows in which two
odd functions occurj.
Ex.
V.
in
Each
may
be
+ The
letter a,
when
it
occurs in a saffiz,
is
omitted.
is
289]
have
469
= 2''-i
odd half-integer characteristics A^, ..., A,, and another half-integer characnot (integral or) zero, such that the characteristics* A^P, ..., A^P, obtained
by adding P to each of A^, ..., A are also oddt; suppose further that A is an even
)i
teristic P,
AP
and that
ia also an even characteristic, and that the theta
functions $ (u ; A), 3{u; AP) do not vanish for zero values of the argument. Then, by
288 the 2"-^ + ! following theta functions of order 2,
half-integer characteristic,
(tt
A,){u; A,P),
...,
AJS{u; A^P),
S{u;
all
But
in fact,
first
function
is zero,
and that
Ex. vi. In illustration of the case considered in Ex. v. we take the following
When
jo=3, it is possible t, if P be any characteristic whatever, to determine six odd characteristics
A^, ..., A^, whose sum is zero, such that the characteristics A^P, ..., A^P&Te also odd, and
such that all the combinations of three of these, denoted by AiAjAt, AiAjA^P, are even.
By the previous example there exists an equation
:
X, Xj, X2, X3
are independent of u.
Adding
to
u any
{u
A,P),
becomes
XS (m
A^Q) (u
A^PQ)
^3)5(;
A.I'Q),
Qj
''^s
we
infer
s,'
of Xj to X
Ex.
is
;
vii.
^k=\rq be
^y be
cj when Q is 42^3.
This equation determines the ratio
X and X3 X are determinable.
:
half-integer characteristics
*. = '-.+?.
-2^?
't.'
'-'
or ^,
and
or \, prove that
+ ?.'-2r.V,
{a
= \,2,...,p),
190)
q),
characteristic formed
reduced value, in
[289
EXAMPLES.
470
where
is integral,
If \r, ^a,
prove that
^j be any reduced
where
^ar. viii.
adding to n the half-period associated with the characteristic .42^3, and putting = 0, we
deduce by Ex. vii. that
Xei''*.'<'.52(0;
where
A,i^A^
in the
form
^k^,
.4i
SHO; A,A^A,)B'^{u;
= Jai,
.44=^04.
/ /]
+ (j=^j^)52(0;
where
^4^2J3)=X,e"*'''''52(0; A^A^A^),
A^A^A^)^Hn;
A,)
A,A,A,)S^{u; ^ 2) + (^'
j^)
Such a
^^ (0
A,A,A,)
relation holds
S'^iu; A,),
between every
...,
^5 be the odd
characteristics .4,$,
...,
A^Q
characteristics,
and
squares of every four theta functions of the same hexad are connected by a linear relation.
291]
CHAPTER
A
XYI.
The
290.
xi.
equations which
We
and
which the last includes the most general case possible. The first two furnish
an introduction to the method and are useful for comparison with the
general theorem.
The theorems of this chapter do not require the characteristics to
Lemma.
291.
A, B,
be half-integers.
If 6 be a symmetrical matrix of
f, g, q, r,/', g',
q',
r,
M, N,
s,
t',
p''
elements, U, V,
m, n be columns, each of
u,
v,
elements,
wfm = 2A^ +
q' + r'=f',
q + r==f,
-q'
=
+ r'=g', -q + r=g,
-n + m 2M +t',
s',
U+V=2u = A,
-U+V=2v = B,
then
2A{N^'-^) + 2b[N^^)\2^f[N^t^)
verify.
or 1, and
Suppose now that the elements of s' and t' are each either
integer
possible
positive
and
negative
all
that n and m take, independently,
=
2M + t'
values. To any pair of values, the equations n + m = 2N + s', -n + m
and M, and a pair of
give a corresponding pair of values for integers
=
2M
s' +t', s'-t- ' is even, and there2iV^2m
+
Since
If.
and
for
s'
values
Hence by
s' must be equal to f.
fore, since each element of s' and if is < 2,
means
sets,
each characterised by
s',
the pairs
a certain value of
s'.
(n,
m)
472
M)
[291
N+M
integer columns n, m.
has
x, ha?
its real
part negative,
=2
_M
thus, if
S(m;
^]
Tm;
\), or
denote
^^m;
M,
denote
2e'i"+^''+'"+'''i'+'"^ '+'''',
2e*<"+'^''+'*<"+*''^'^*'"+^'',
tiu-v; q)^(u + v;
r-)
%{u, \) or
we have
= S|
I';
+r
gj
where the equation on the right contains 2J' terms corresponding to all
or 1
all other
s', which is a column of p integers each either
values of
Therefore
of
if
elements,
p''
both sides by
^(u-v;
we deduce, replacing u by
e'+<"^,
q)'^(u
the result
+ v; r)='2%
q
|_
-q + r
]'
or 1,
/
'
g'
2a), 2a)'\ .
2T],2ri J
= 2^
(0,2a'
..W + q' + r')][a,,2a,n
q
Ex.
i.
When
the equation
2(B, 2a>'\
\2ri, 2ri'
i(6'-g^ + r-')|a,,2a,n
+r
the characteristics
q,
is
n denotes a
definite
row of
291]
or
1,
473
we have
for
2e'''+"'
= n [1 +e"l''i+i'l.
Ea;. iL
independent of
Ex.
that
i.
two functions
t \ obtained by varying
bAu;
Ex.
Prove that
iv.
^(J)],
u.
iii.
ratio of the
3[. + a;
is
t=l
if a, a'
',
are not
u.
be integral,
..[.,jQ]=5.-,.[o,j(-'-)]4.,iQ].
From this set of equations we can obtain the linear relation connecting the squares of
2" + ! (or less) assigned theta functions with half-integer coefficients.
Ex.
row
v.
is Xi^j,
prove that
|Xj,y|
?;,
i[)
any
t-th
half-
integer characteristic,
|.[....;
(:)].[..-..; i(:')]r|^.[".=
where
is
is
the
odd, {Ui,
is
71/}
rationally in
instance
sum
of the
{wj,
When
a.
t>,}
the characteristic ^
when p=l, we
^l
Mj-J-Mj-;
m^
,; ^l
\.
For
obtain, with a proper sign for the square root, the equation of
three terms t.
by a
\s\
Vi, ...,
y^ be any
columns, whose
cally.
("
When
2m
{i,
i ("
*
it
areconnected
J
follows that
if j, ...,
arguments,
m,
element
is
j)ih.
u-vp; j(
is
odd and
Scott,
Mj-|-t)y
i(
Mi-f,; ^f
vanishes identi-
t Halphen, Fonet.
m,
m rows and
t. i.
p. 187.
p. 71.
is
-474
may
Mj +t'y
i\
^i-Vj',
i[
{Ui,
The
[291
result obtained
may
be written
%}*=0,
m,
vanishes.
proof that for general values of the arguments the corresponding determinant
2^ rows and columns, does not identically vanish is given by Frobenius, Crelle,
{Ui, Vj}, of
A
Ex.
ii.
for the
given below
is
292.
We
292.
functions.
J denote
Let
the substitution
i(-l
1).
1-111
11-11
111-1
and
Jr,
is
in the r-th
s-th
column
Let
2 Jw /< =
then
or
1,
according as r^s, or r
= 8(r, s =
quantities
1,
2, 3, 4).
written
down
(M,
((Wl)i.
Let
and
(3)i,
elements.
rows.
Then
if
sets,
hu^Vi
2^
F3
+ A P^ F4
= hjt
A, [7^3
is
obtained by
+ hjt(u,)i{v,)j+hji:(u,)k(v,)j+hjt(u4)t{v4)j
[J,, ( FO,-
+ J^ ( F,),-
+ ^.3.(F3)^+J,,(F,),]
= hi,
1(2/%,)
U,)u
FO,
(2J- J^)
[(
t/Ot
( V,)j
+ ( U,% ( FO,] +
and
element oihU^Vi
The theorem was given by WeierBtrass, Sitzungtber. der Berlin. Ah. 1882 (i, xxvi., p. 506),
with the suggestion that the theory of the theta functions may be a priori dedncible therefrom, as
is the case when^ = l (Halphen, Fonct. Ellip. (Paris (1886)), t. i. p. 188).
See also Caspary,
Crelle, xcvi. (1884),
and
ibid.
zcvu. (1884), and Frobenius, Crelle, xcvi. (1834), pp. 101, 103.
292]
Now we
^ (Ml,
475
have
?i)
^ (Ma,
?2)
^ (its,
93)
^ (M4,
qi)
g2oM,J'+22AMr(lr+9r')+26(nr+r')"+l!>ri28r(r+r').
columns of
quantities
we suppose each
Hence the
exponent becomes
for
which
of
all
n^,-
= i (W- +
all
Wjj-
are integers.
But
Let
Tw,,-
iV^,^,
We arrange
w,3
since
= i (Wij + n^j +
%+
nij
= ^ e/,
say, (e/
iirj
or
the
1).
Then
e/.
N^ N^ = n^
since
- 2n,j) - ^e/,
every term of the left-hand product, arising from a certain set of values of
the 4p integers Urj, gives rise to a definite term of the transformed product on
the right with a definite value for e/, while, since
nrj
wii,-
- 2mrj) + ^ e/,
every assignable set of values of the 4p integers m,j and value for e/ (which
would correspond to a definite term of the transformed product) will arise,
Now we
which
is 1
if e
= (ej,
ep)
4-
m^j
+ e/),
the expression
is integral.
according as
or
In fact
i_ Se^*'
not.
right,
for
when
integers
[292
476
Hence
if
_ V n gd
2P
may suppose
the summation
every value of e.
is
If3+2*
tV
Pnr+Zl'r) +6
e~^' (Vr-f')
where
pr-k^-^ Qt,
Pr
= ie' + Qr,
so that
2;)/
Thus we have
2 g-.(e'+S<,V)^
O
2^
(,
This very general formula obviously includes the formula of Ex. xi., 286,
Chap. XV. It is clear moreover that a similar investigation can be made for
the product of any number,
k,
we know
of a
matrix J, of k rows and columns, which will transform the exponent of the
general term of the product into the exponent of the general term of the sum
of other products.
It
It
is
for this
is
is
elaborated.
should
consist of half-integers.
Ex.
If J be a half-integer characteristic,
i.
= Q,
say,
Q)S{w; Q)S(t;
Q),
we have
(t>{u
i (* )
istics, this is
putting
Q + ^r\ = R,
so that
also
all
possible
becomes
all
2*
half-integer character-
2*
half-integer character-
the same as
-|-a,
where,
if
^ = i(a)'
^^KJ)'
**'^"
ll
= ''
l^l=^'3',
\Q,R\ = a^-a'^.
v-a
R),
GENERAL
293]
CASE.
477
By
we
is
positive
and
/3S is
negative, p
= ab-fiy,
and
-^8rQ = ^^^e(uS-Vy;
where
e a
f
P78r|'*^^'')
e(^-^+a;
-pa^rl"^'')
denotes the theta function in which the exponent of the general term
is
row
V are
fi,
letters of
elements,
Ex.
/x
Taking, in Ex.
iii.
= i'<A+l,
k being
s, s'
We
293.
than
than
less
ff,
r,
subject to the
r.
.||;')
number,
less
/,
positive.
.e(.;
and
and
e,
now
proceed
or
1.
to obtain a formula*
any
of theta functions.
k,
(.-^l,j>
"2,jt
>
-^r, J
-^*i J/
~ IT
^3
y'''i,j>
*2. j>
' ^r,j>
^k,j)
^>
herein
r^
is
consisting of integers
matrix
is
zero,
and denoted by
1, 2, ...,p,
we may
fij
1, 2, ...,
bearing in mind
and the values of j
A;
^3
The
be denoted by
columns
-^ fi,-;
*
will
flj-;
first
be denoted by H,.
Then the
substitution inverse to
Xj,
the latter
is
\,~\
coj
is
478
any value of
If for
j a set of
integers, Pr.j, be
[293
known such
that the k
quantities
n
it is clear that an infinite number of such sets can be
we have only to increase the integers Pr,j by integ^ral multiples of
But the number of such sets in which each of P^j is positive (including
derived;
fjij.
and less than the absolute value of fij is clearly finite, since each
element has only a finite number of possible values. We shall denote this
number by Sj and call it the number of normal solutions of the conditions
zero)
/p.
Q,j (Pr,j)
integral
f-J
the same as the number of sets of k integers, positive (or zero) and less
than the absolute value of /tj, which can be represented in the form '^(pr,j),
it is
The k theta
elements
pr,j.
taken to be
first
those given by
@, =
wherein Br
quantities
is
Vr,j,
Nr,j{j
negative.
(r=l,...,k),
is
X, the
BrX^
2e^rrNr+BrN,^^
1,
Chap. X.) by
'\j(a;rj)
transform
the
sum
"Z
B^Xr^ into a
sum
r=l
hjXr^,
in
6, are
r=l
for real x,
brX^^ is negative.
m^^j, Vrj
= Xf' (Nr, j) = p
kk
the expressions
r=l
nj iA\, ,),
(Vr, j)
Vr,
j)
= ^ ^ ( Vr, j),
k
BrNr^,
determined by
S brm/
r=l
r=l
and
p -
Z\j(7nrj){Vrj)= SX.j(F,,,)(mr,j)=
j=l
all
rrirj
is
transformed into
'"''"^
where the
Nt
have every set of values such that the quantities \j{mrj) take
r=l
quantities
r=l
j=l
^.Vrrrir;
Jf,
293]
As
479
Let
{Pr.j)
we
now
seek
to
of the conditions
n ^i
,-f
(Pr.j)
= integral, = {p,j),
say
j,
{m,, ^)
(iv,, j)
fT,.
= (p
=
,-)
("r,j)+
r,n,. (if,,
,.)
^ n^ (i?;,
,.)
say,
(e'r.j),
where
("r,i)
{Pr,i}
+ r,n. (if^_
.),
(e;_
.)
and a
fOj
is
= Tj^i (E'^j)
We
for
do
we have
(,, j)
rj
rj
so that the values of ?v,j which arise are such that 'Kj{nr,j) are integers.
Conversely
integers
(n^
let
;)
are
Xj(n,.,j)
put
\j{nr,j)
= {Pr,)) + IJLj{Mr,j),
wherein the quantities if^j are integers, and the quantities P^.j are positive
integers (or zero), which are all less than the absolute value of /i; this is
possible in one way then taking any set of assigned integers {E\, j), which
;
are
all
a set of integers
Nrj by
{Nr,,)
and
less
{E'r.j)
{Pr.j)
11 0,.
r=X
we
can,
a definite set
{rir.j)
we can
define
(Nr.j), arising
'
"
'
''
of
such that
fij,
= integral,
X.;(rar,j)
are integers.
'Kj{nr,j)
And
and
are integral,
we
can,
by
association with a definite set {E'r.j), obtain a definite normal solution (P,.j)
and a
480
[293
a term C"
be written down
Sj
. . .
Sp times,
Si
. . .
Sp
normal solutions of the p conditions Xj~' (P,) = integral, then there will arise,
once with every assigned set {E\^ _,), every possible set {Ur, j) for which Xj {rir, j)
are integers.
We
introduce
integers
tegers
now a
factor
the conditions \j
(n^, j) satisfy
(-fiyj),
(rir, j)
zero),
and
less
according as the
1 or
= integral,
than
Take k
or not.
Tj
in-
put
then
t
J
S 2 -
/
e,.,
ir, j
1
e
+ -^ = SS -
e,.,
rrir, j
_;
SXj
(to,,
_,)
(Er, _,)
= l.(Nrj)(Er,j)=NrEr,
j
and
this is integral
actually occur
{Nrj)
when Nr
is
integral, that
= \j{mr,j) = -
B;(nr,i+
Tj
for
a given set
w^.j)
than
rj, is
equal to
the values
for all
(rirj)
which
by
Tj
{rir, j)
and a given
is,
defined
Hence,
iV,_ j
sum
all
when
ri ... r*
(iV^.j)
are
all integral,
and other-
which
is
zero because
tj (i\^r, j) is
all
l)/(^^^r,i
_ 1),
certainly integral.
Hence
if
'"
.>'
= l,orO,
integers or not.
values of
m,.^^-
arise for
we denote
C^TTT^l'
according as \j (n^j) are
far,
only those
factor,
293]
481
Tl^
YlffiVrXr+BrNr''
'
be unaltered.
series will
n,E,E
wherein
= ^j{E,.j);
each
zero,
Sj
a, b,
the
is
than
less
(Vrj) = rf^o)j(V^j)
matrices
number
...
the function
,. is
may
The number of
= rjnj(E'^j);
/u.^~'r,n,(Pr,j)
integral;
by
linear transformation of
^^--V
e'
be removed.
the number of sets (E'^j)
{e'^j)
by an exponential
K,j<rj;
|/^j|.
conditions
multiplication
E^',
^V.i<
the
transformed product.
;
of
\fij\,
equal to
is
{e^,j)
and
'
JV^,,...,iVjt,jE,
all
hand side
-Qg2''rli,+rmr=+2ifr ("1-+-)
*i---*p(''l---'>)*
is
the
number
is
of theta-products on
i.
We
test this
formula by applying
it
where
m,-
is
which
is
independent otj,
11
= (-l
rj
= 2,
br
= b, i = 4;
dependent of
write
),
=a)say,
etc.,
- (fVJ = (K + ^.;.
1-111
11-11
111-1
iO+ A.^'
'
is
independent of r and we
By
in-
may
the formula
31
482
5
(v,
q),
when
iV
is
integral,
Hence the
e'rju.
[293
(16)*''
= 2'^''
*
;
"Jo)
is
independent
all
the values of
reduce to 2" terms, in which, since {E'rj) = ^a>{^e'j + Lij, ..., Wj + L^j), all
values of f'(<2) arise. Hence there is a factor 2''' and instead of the summation in
|(a("r,j),
ne(
C.215P 2
E,
and containing
Now
c',
e'
i'-tiEr,))
2*" terms.
put
Mj being
integral then the factor of a general term of the expanded right-hand product
which contains the quantities ^oj {E^,,) is
;
where
k,,i=E^,i-\-E^,j+E^^i+E^^j-2E,,i=f, + 2{Mi-E,,,),
and
ime"SJCr
^ gTWe'
iritje/
'
while
227rt^r,j'lr,j
= rl22f,m,.,y
(mod.
'
Jr
=7rtf
2),
Jr
Sn^,
r
so that
form
ne (^r,
Hence the a*!" terms arising, for a specified value of e', for all the values of
2" terms, and there is a further factor 2^''
the right hand being
?'
)
'"^'
r,j, reduce to
where
conditions,
XJ^
than
8<
even, possible in 8*
ways.
= integral,
F,
(s)
of positive integral
and
equal to e^.^+^^'+^^a^H-a^.^^
n^
Ur, 0)
if
which
and therefore C.
2i8p
= i.
is
we have
Af,,
number
the
(a;)
substitution
which
Jm
for
Hence s = 8.8*=2i5.
Making now
the
C we must know
+ X2 + x^ + x^ = 2{Xr + ^/r)
of
^g ^^^:^^
= 1 26-""^'
m [ Q]
4
^^^]
Therefore
483
W n<>(r,+a.
is
easily
shewn to be ne-^<r.
J.
.), if
o),
j^)^^(^^_ ^^,j^,^^)
thus
e~'*''ne~'^<"^'?'-'e^('-.?i-)-2W'
g-'"e(2g'r+e')
"2J>,,,,
"2^e..
which
is
Ex.
ii.
More generally
let
X=
-a>,-
(Z,) = X(j;,.)give
+ Xi'^ = m(x^^+
^i'+
wherein
the eauations
equations ((fV)
Y.\ = X
\
we have
then, since,
+xt^),
by a property of all
linear substitutions,
(v.\
lead tn
to
(y^) ImH
also*
^1-^1+
Hence,
if
^'tJ^i=i(yi.'^,
+2/k^k)-
we have
AZj Yi + hX\V2 + ...+ hX^ Yi = 2
hi, j
where X^,
x^, etc.
2 Z^, J
r,,
=m
Ai, J
ij
!,J
2^^, j^r, i
= m {hx^y^ +
. . .
+ Aar^yi),
of jo quantities.
Taking a case
where
m = l,
rj
= r,
a,,
= a,
;=r*, Er,j<r,
E'r,j <\iJ.\<r^,
we have
so that the
new
Suppose now,
(Xi,...,2'...Xt)=i(
is
2-i
2-/E;
* Therefore mj;j/
= JS:r=Xa:
Xy = \\xy, so that \X = m
){x^,...,Xr,...,x^),
2-/fc
>Jm''.
312
484
[293
which gives
and
+A'i = Xi +
Xi+
The previous example
may
suppose
A\-X2 = x^-Xi,
etc.
is
odd so that
I-
+Xk,
rj
The
fractional parts of -
(e'r.y)
+ A,ji
f/
ne
> -^i'+^ic.i)
(vr ^^
in the
form
ne (vr,
'
,
The equation
shews that
instead of
values of - e/
all
|(i 1*^
7*''
(<
1)
do
arise.
Hence
for
a given value of
rP,
(jE',-,>)
there are,
H'^^iIp
divides out.
The
As
i(^'..y
where
t<1
The
is
= i<-(;,,,) = g(^..,+
corresponding
independent of
r.
Let
+ E,,i)='i+M
+^,,.)-^,w)=
(l^',
^,...), (mod.
Now
2
is
A-
1).
r,j is
therefore, as
\r\'^,
then
~(.,.,)
contains
number
or a factor of
^e,,,
= S (,,,) = e, + AJfy;
k,
and
1).
293]
Hence the
factor above
485
is
A=l ne
and the general term of the
right
'
ft
hand
1-/
_g
is
[-('"iS]'-""
Since -a{fT,i) = {-jr-^Mj E^^j\ we
we have
instead of ?*"
To
/fcP
and a
may suppose
cj<k
Hence
to arise.
=C,
...rp)~i (j...p)-*,
(r^
values of
all
say,
+.j;i)-^r = integral,
r.{^i+
namely, how
many
t(^i +
exist, for
where
may
are
<r,
utilise
3*
...,
x^o*;
+a:i)= integral,
let s
of terms
be this number
istics -r-
=9
wil^j effectively,
gp or 2'p to 2^.
We
repeat themselves.
) When
We can
""
For values of
q).
2e
'2/
> ^ we
k=r=Z
there
character-
=i and
Ex.
iii.
=(
Y, Z, T,
U,
V)=iu,{x,
110 1-1),
1100-11
o.-^
1-1110
0-1111
/i
u, v),
1-1
-1
1-10
Z-^T=^z\t,
11
-110011
= 64 and
^2+ Y^+Z^+T^+ U^+ V^=x^+f+z'''+t^+u'+v^
X +r +Z +T +U +V=x +y +z+t +U+V
Z-T=^x-y, U-V=z-t, X-Y=u-v,
X->rY=x+y,
0),
11-1100
11-11
1-111
which
y,
t,
-11110
lead to
z,
U-irV=v,\v.
by
aj
= a,
[294
CHAPTER
XVII.
For
294.
groups
of
to
is
not
expressly
the
stated
^^
is
\qi,
is
teristic of half-integers,
*
preliminary.
contrary
now
characteristics
letter,
q-2,
characteristics
;
Q+R
0.
To
(1890), p. 61 {aai
integers
is
considered
S = ^s,
Untersvehungen
this
If
in
qp
The
say Q.
denoted simply by
the symbol
considered
a characteristic
Prym
may
u.
be given
Prym,
Krazer, Math. Annal. xxii. (1883) ; Prym u. Krazer, Neue Grundlagen einer Theorie
III. (1883)
der allgemeinen Thetafunctionen (Leipzig, 1892), where the method, explained in the previous
;
is fundamental
Noether, Math. Annal. xiv.
where groups of half-integer characteristics are considered, the former paper
dealing with the case p = 4, the latter with any value of p; Caspary, Crelle, xciv. (1888), xcvi.
Stahl, Crelle, lxxxviii. (1879) ; Poincar^, Liouville, 1895 ; beside the books
(1884), xcvii. (1884)
of Weber and Schottky, for the case p = 3, already referred to ( 247, 199), and the book of
Krause for the case p = 2, referred to 199, to which a bibliography is appended. References to
the literature of the theory of the transformation of theta functions are given in chapter
XX.
In the papers of Schottky, in Crelle, cii. and onwards, and the papers of Frobenius, in
Crelle, xcvii. and onwards, and in Humbert and Wirtinger {loc. cit. Ex. iv. p. 340), wiE be found
many
much
larger generalizations
the reader
may
consult Weier-
Dec. 1869, and Crelle, Lxxxrx. (1880), and subsequent chapters of the
present volume.
t References are given throughout, in footnotes, to the case where the characteristics are n-th
In these footnotes a capital letter, Q, denotes a characteristic whose elements
are of the form q'Jn, or of the form jj/n, g/ qi being integers, which in the reduced case are
positive (or zero) and less than n.
The abbreviations of the text are then immediately extended
to this case, n replacing 2.
parts of integers.
'
'
2^^]
487
whose elements
teristic.
\Q\
\t,
= qq' = q,q,'+
+qpq;,
Q,
R\ =
-q'r = I
qr'
(g^-r/
- q/n),
i=l
(^]=e-''r^=
e-<9.v,+
...
+5V,)
further
differ only
the
Kv.
Zp+1
periods u"-
%"'
"',
295.
may be
lemma
preliminary
given at once.
Let
characteristics
1,
2, ...,
7 of
made
be integers, such
= a;,
then
if
have
2""'' sets of
tti,
..., ttr
2) for indeterminate
2, ..., r),
values of
x^, ...,a;;
= ai,
...,
Ur
ttr,
(mod.
a;,,
2),
..., a; is
either
or 1.
For
sum
l-^
X,,
[1
....
+ e"'^-"')]
...
[1
+ e"(c^r-,-)],
Xn
which
(i=l,
,,,
consider the
,,
...,
Xn
a;,,
is
..., a;
either
A term
1.
in
...,
When
of a factor 1
+ e"
'""'
...,
for
one at
we use a
factor
l4
a'
m we
+ m''+
use
number
number
w = em, instead
+/*""'! ^''ere
/n
= ai
'
of inoongruent
of e'"
"'.
of sets of
and instead
PROOF OF A LEMMA.
488
Now
by such a term
may be
/li
is tj'pified
as
1
where
[295
0,
or
iri(f7,-ai)+iri(J/2-n2)+...+iri (t/^-o^)
sp
or
1,
...,
or
-iri(a,
j)
this is equal to
and
+ ...+o)
m{c,x, + ...+i:^x^)
where
(^=1>
+a^,i,
Ci=a,_i+
2, ...,n),
when
now,
>
/i
^-(o,+...+a^)
^^ _^ g^,^ ^j _^
g<,,)
0,
. . .
Ci,
(1
+ e""")
...,c
must be =
(mod.
2),
fT^ is
of the series is
is
Or ^
1
''
or 2"
Ex.
i.
of all
If,
be denoted by
ff,
2'^"
half-integer characteristics,
^,
the
number
of even characteristics
characteristics, prove
Ex.
turn
ii.
all
the
If
2*
characteristics, the
(l+e-)
and
is
if
^a be other than
'2I
(!+'*'')
zero,
sum
2e"'^'
<21
= 2e'"('^"''9)
zero,
vanishes.
and
^ become in
For
it is
equal to
(l+e"**'"),
(l-fe"'*"'')
On
it
22J'.
We may deduce the result from the lemma of the text. For by what is there proved
there are 2*"' characteristics for which \A, | =
(mod. 2) and an equal number for
which J, 1=1.
I
it
can be taken in
2^1
ways.
Let Pi be any characteristic other than zero and other than Pj, such that
|PP,| = 0(mod.
*
Among
p-i (nP + m is
the
]
for
not divisible by
2);
i.
for
296]
sum*
A, P,P, = P
Pi
2^*-'
can be taken in
( 295), P.^
Pi Pa be the reduced
if
489
P., P,
I
0,
IP3, P,1
2^
ways
Pi
PiP,
P3,
I
also
by
(mod.
2).
by the
P3, P,
0,
(mod. 2);
0,
also,
P3, Pi
ways
of Pj, Pj,
definition,
0,
2'
characteristics
(mod.
2),
and
!
Let Pi be any
P3P1 =
-*
'Si "l"it
2)
0, etc.
than the
characteristic, other
"> -^It
which
P3,
2-^3>
-t
-^
S-' 1>
^l^S^St
such that
is
P,
P P, = 0, P P3 = 0,
2'^'"^ 2' ways, and we have
I
P,P3, P,
I
(mod. 2)
0,
= P
P,
+ P
Pi\
+ P
(mod.
0,
2), etc.,
and
PiP,P3, P,
Pi, P,
I
>
1, X'^2>
i^i^it
The
2'*^"'+'
r-th
2''""' = 2''"'
P3, p.
2''
(mod.
2).
"l"i"i^ >
characteristics,
1)
0,
characteristics,
of the fundamental
(2^-"'+''
=
I
ways
characteristics can
thus we
may suppose
r as
We
(22P
1)
(2^P-'
- 2) (2^^" - 2^)
or
(2^ ways.
But
1)
(2^-^ -
(2'*-^
-1
. . .
. .
(2=P-''+'
(2^-^+^
- 2'-^)/\r,
1
2*''('-iV 1^
any r linearly
these ways will not give a dififerent group
group may be regarded as forming a basis
all
we may
>
take, as basis,
wherein P^P^
P P
is
basis
^T
P^P^_ are
will arise
each either
or
J.
by the combination
GOPEL systems'
490
of
P1P2 and
P2.
and
other than
Pj',
0, P,',
P/, Pi'Pi, in
selected
is
. .
P/,
may
- 1) (2'- - 2)
(2'-
for a
given group, a
is
- 2'-i)/|r,
;
P,', in
2''
be selected
them, P/,
[296
2^
immaterial.
0,P,,P,,...,P^P,....,F,P,P...
2''
of
characteristics, in
syzygetic*,
is
(2'P
I) (2'^p-^
(22p-+2_ 1)
1)
"(2'^- l)(2'-i-l)
Such a group may be
name
2'' characteristics.
The
when r=p, such groups having been
Gopel group of
called a
is
We
now form a
(2-1)
p=
(cf.
221, Ex.
i.).
by adding an arbitrary
group (P) just obtained
let P, Q, R be three characteristics of the group, and A', A", A"', the three
corresponding characteristics of the resulting set then
297.
characteristic
set of
2*"
characteristics
\A',A",A"'\
\P,Q\,(inod.2),
resulting set
is,
is
its characteristics
thus the
0,
(mod. 2)
The
A + AP^ = AP^,
or
= PiP^.
there may,
Hence
*
it
When
(n^P-l)
...
(n2i>-i-+<!-l)
divided by (n'-l)
...
(n-1).
number of such
syzygetic groups is
297]
OF CHARACTERISTICS.
491
A\ A",
I
A'"
0,
(mod.
we
can, taking
P^^A^A,
2),
P,, of
which every two are syzygetic, and hence, can form such a group (P) of
2''
We
shall
To
0,
(mod.
2).
2''
A may
2^
characteristics.
But
if it
AP
g^. (2^
1) (2^-^
1 )
(2-^-^+-^
.
1)
(2^-l)(2'-'-l)...(2-l)
We
By
Frobenins, the
is
when r=p.
492
For
Ex.
number
jo
= 2, = 2,
>'
[297
And
is 15.
the
We
PQ = \P\ + \Q\ (mod. 2), or in words, two characteristics are syzygetic when their
even or odd according as they themselves are of the same or of different character.
It is immediately seen that the 15 groups given in 289, Ex. iv., satisfy this condition.
The four systems derivable from any group were stated to consist of one system in which
all the characteristics are even and of three systems in which two are even and two odd.
lent to
sum
We
is
Of the
298.
2^^^^
is
certain definite
and a certain
prove in fact
o-
p 7-;
being
2'"'
(2"
these
may
then be
systems which
may be
When^ = r,
there
For, if Pi,
is
...,
Prhe
must
satisfy the
...,
AP^
which
congruences
\XP,\ = \XP,\=
= \X\, (mod.
2)
\X,P,\^\Pi\,
as
immediately obvious.
is
when \X,
Since,
X, P,P, s Z, P, + X, P, s X,
I
P,
PJ +
(i=l,2,
Pj
|,
...,r),
and X, Pj = Pj
X, P, +
I
|,
P,
= IP.I+IP,1 + IP,,P,I^IP,P,I,
follows that these r congruences are sufficient, as well as necessary.
etc., it
2^^'- solutions.
If
( 295)
of the 2' characteristics forming the Gopel system
for
it
the definition,
{AP)
if
P,
Q = 0.
This result holds for characteristics which are n-th parts of integers, provided the group (P)
which either the upper line, or the lower line, ol elements, are zeros.
consist of characteristics in
299]
2^-^/2''
= 2^"^^
and the
characteristics A,
by addition of the
493
systems being
(mod. (P)).
Put now a = p
r,
equation
g+h
= 2^',
also
h = 2~^ 2e"""
[1
\ i+e"''-'*'-''''"'"'-''''] ...
+'''''*''''""''-''''],
wherein Pi,
of
noticing
For,
characteristics.
|
and
Pr are the
possible
RP =
I
the
...,
2*^
the
|,
it is
R is in turn every
the
that
one
congruence
characteristic
either
character,
which
for
all
R,
of
side
RPi,
...,
it
RPr
is
are
when
is
equal to
when
2"^e'"'*'
Now
if
we multiply out
It
wherein
is
true.
-Pa*
p,,Pj,
Pit
for
fj,
of the
...
characteristics
but
and therefore
where S,=RPi...P^,
will, as
becomes
all
2=?
characteristics
in
turn,
494
also
become
[299
all characteristics in
seen to be 2^
it is
r (r
l+r+ \,
2>'
I)
+1 =
'
Hence
characteristics.
2P[(l+^)'-],^,
2P^
These results
when
2'^",
o-
>
0,
will
2""' (2"
+1)
+ 1),
2?-' (2p
determines g
determines g=^, h = 0.
cr
make up the 2^
1),
possible character-
istics.
RPi
R,
are not
P^ + Pm
1
is
of the
all
1
of
n
is zero.
{1
+ e'rili'ml+TiliJ, Pm\\
= |P,! +
|P,-|,
2'
gyilPA+wHR,P,\
Qj.
gri:RI
i=l
Now
e"''^'''' is 1
or
e^i\IiPi\
i=l
1
according as
of 2' characteristics
and therefore
characteristics,
2)
be of
The
2'^
Chap. XI.) that the 2* characteristics can all be formed as sums of not more
than p oi2p + l fundamental characteristics, whose sum is the zero characteristic; we proceed to shew that from the 2'" Gopel systems we can choose
2o- + 1 fundamental systems having a similar property for these 2^ systems.
Let the
= 2^
the
first
Then we prove
that
for the
it is possible,
modulus
(P),
2cr-|- 1
character-
300]
495
istics, which we may denote by ^i, ..., A^^+i, such that every three of them,
say A', A", A'", satisfy the condition
1^',^", ^'"1 =
but
it is
=
^"
\,
(mod. 2);
1,
P be
^" +
P,
A'"
i",
(P),
|,
any
it is
convenient to prove a
lemma
if JS
be
characteristic
become
A^,
...,
then*
R
For
let
Z,
B =
1,
Z, P. =
I
Z, P, =
0, ...,
I
0,
(mod.
congruences
2),
and, corresponding to any characteristic R which is one of Ai, ..., Ag, and
therefore satisfies the r congruences \R, Pi\ = Pi\, take a characteristic
\
S = RX;
then
such as
pairs,
. "
...,
e'*
\Pi\,
Ag can be divided
-^^
'
'
= - e"
^'
into
This
I.
= 0.
The number
satisfy the
is
= 1.
|
equal to
H
*
and
We have proved an
K
be in turn
t If
iJ
B be
+ We
and
all
the
2*
characteristics,
2e"
^'
'
'
one of ^
...
A., so that
possibility only
if
[A-^,
Aj\
l.
if
B be not
Ex.
295,
"
Xe" = 2=p.
'
'
P we have s/'
=
Since [A^,
^3 ^1^2.
\R,P\ =
-^
^'
|,
ii. ).
P be
If
(P),
^ = e'"l^l 2-.
l
it
is
496
B is
wherein
satisfies
R, A-^A^
A^
A-^,
...,
is
is
Now,
the series
|,
[300
For a term
A,.
equal to unity*,
since
A^, A^,
equal to
R
the characteristic
A^ = A^P, which
is
gruent
series is 2^""\
We
A^,
consider next in
Ag other than
...,
= P,
then
A-^, A,,,
A^ and must
\A^,A^,X\ =
which, in virtue of the congruence
X +
A^, A3, X
A,, A3,
I
involve also
if A-^^A^
...,
for
these conditions
is
A3, Ai,
1.
\,
|^,.^,X| =
A^, A^,
X\ +
jIj
Ai, A2,
The number
must be one
of
1,
=
\
it
1,
congruences
satisfy the
Ai, A2,
A3
1,
equal to
M
or
where
R is
lemma
one of ^1,
...,
A A R\ + \A
A3,
is 2^"-^.
\,
But we have
A1A2A3
Now
it
to
is
....
-Bji+i,
these characteristics, and for an even number of characteristics B^, ..., B^,
the condition that every three be azygetic excludes the possibility of the
existence of any relation connecting an even
BiB^
we
I
...B:ji
have, as
B,B,...
0.
is
For,
B being
. .
A.
relation
,
B^,
easy to verify,
B^ B^,B\ =
* It is
B B^,B\ + \BB^,B\ +
B\ = 0.
...
+ B,^ B^, B
\
\,
^^^]
thprpforp
r
i =
inereiore, naIR
as tl^, B^i,
-
o
B,^,.
This holds for all values of m not
R -RR
- B,B,
^^
...
497
^i
0, we cannot have
greater than k, and
r>
To
number
find the
of
ways
in
A,A,A,
On
for
we
the whole
congruences
\A^,AX\ =
\AAX\
1,
l,
\AAX\ = 1,
The number
of ways in which
A, can be
Si
where
is in
that ^1^42-^3-44
number
of
ways equal
to the
sum
A^+^ can be
chosen
and therefore
in 2'*-<'- ways,
and that a
To
2o-+l
the
2^*
1 ways,
characteristics thus
A^+^ = A-^A^...
the characteristic
characteristic
A^ = A,A,
A^^+i
and A^+^
determined
if
A^
can be chosen
A^_, being
. . .
excluded.
In
in 2 ways.
it is
convenient* to add
of ^i,
. . . ,
A^+-,
we have
A^+2> Ai, Aj\
the expressions
the result
set
Aj,
...,
Ai, Aj,
Ai, Aj, Ai
From
the
2^
\,
Ai\+
Ai, Aj, Aj
Ai, Aj,
being both
characteristics Aj,
...,
A^
^20-4-1
We
zero.
it is
1,
have then
possible to choose a
\A',A",A"'\ =
l,
in
2ar (2^^
1) 2^-'' (2'^-'
[2q-
ways ;
Ai,
sum
...,
- 1)
+2
...
(2"
number
sum
is
A^^^
is
characteristics
a relation connecting
is zero.
32
ROOT
498
[800
SETS.
an even number, there can be at most* only one independent relation connecting an odd number of the characteristics A^, ..., A^^^^. And, as before
remarked, to every one of the characteristics A^,
Gopel system of
2''
..., ^42^+2 is
associated a
characteristics.
301.
The 2-" systems {A-^P), ..., (AsP), which have been considered,
were obtained by limiting our attention to one group (P) of 2'' pairwise
We
syzygetic characteristics.
are
now
still
further to
-42,^+2
be a characteristic of the set Ai, ..., A^, are ( 298) the r congruences
XAk, Pi s Pj in which i = 1, ...,; in virtue of the conditions uli. Pi
,
,
I
H Pi
X, Pi =
which are
solutions, but from
|
0,
a set 4/,
...,
A'^^^^, also
= XAi, XAj,
belonging to A^,
...,
A^.
Further
XAt:\ = Ai, Aj, At:\ = l and any relation connecting an even number of the characteristics A^', ..., A'^^^ gives a relation
connecting the corresponding characteristics of A-,, ..., A^^^. Thus the
A/, A/,
Ajc'
Ai,
...
from Ai,
...,
.a.2^-1-2-
Hence
all
...,
2''\2^-l){2"'-'-l) ...(2^-1)
|2o--l-2
root sets
of the root
302.
2'^ characteristics
to each characteristic
set.
Fixing
attention
upon one
of
-|-
of
its
characteristics,
Ai,
...,
-42^+1,
have
all
The
\XA,\ = \XA,\ =
s|Z^2.+i|.
the characteristic of which all the elements, except the t'-th element of the first line,
are
zero, be denoted by ;', and ; denote the characteristic in which aU the elements
are zero
except the t-th element of the second line, every possible characteristic is clearly a Unear
aggregate of Ej, ... , Ep', j, ... , Ep. Thus when <r has its greatest value, =p, there is certainly
* If
one
It is
only in case
all
any
2(r
+l
characteristics.
the characteristics of the group (P) are even that the values of
Ag.
...
can
FUNDAMENTAL
302]
SETS OF
2<r
congruences
\X,A,Ai\ = \A,\ +
\Ai\,
if
be a solution of these congruences, and
group (P), we have
XP. A,Ai =
\
\X, A,Ai
I
XP
so that
is also
were in number
r,
499
P, ^,
P, ^,
(i
+ l));
2,3, ...,(2o-
= ^, + 4f +
I
2
I
I,
and
any
solution, there
were
X
2''
Ai,
...,
Ai,
....
A^^i, of a root
A^+i, which are
set,
of the
all
form Ai = XAi.
Starting from the same root set, and selecting, in place of Ai, ..., A^+i,
another set of 2ar+l characteristics, say A, ...,Ai^+2, we can similarly
derive a set of the form
X A2,
...,
-a.2ir+2,
arises
Thence the
Ai, A2,
which
...,
= 2,
set
...,
2cr+
!,_;'
= 2,
...,
2a-
+ 1.
Since there
is
no relation con-
necting an even
expressing that
may
arise
whether such a
set as A^,
...,
the one
A^^+i, of
+1
2ff
characteristics of the
is
...,
of
2ir
will
appear below
303).
322
500
of fundamental
trriLiTY
[302
sets.
characteristics of the
from A^,
from
..., J-'aa+i
On
it
2."'
{^^ -l)(2'^-'-l)...(2'-l)
20-+1
different sets, A-^,
...,
2cr+l
A^^-^, of
characteristics of the
same
character,
Of the
characteristics A^,
(2^
1,
4- (2o-
1)
...,
3)
1,
+ (2o- +
..
+ 1) = 2=^
1, 2(7
number
These combinations
-4j, ...,
no
com-
is
all
X, P^
P^
for
;
j
And
A,A,
...
tio
...
+ Z^_
I
Pi\
= Pi
\
|.
modulus (P);
in regard to the
for
Ai...
wherein
form A^p
A2ici
= Ajfi^Am^i
is_a_characteristic of the
= A^A^...A^_-,P, and
...
Am+i^P,
0,
whereas
A,
...
...
|Z^^,
\Ai
...
jljp-i,
p +
A^^^^,
p
I
A^^, A^^j^i
I
+ ... +1 -42p_i,
consisting of
...^
A^+^jt. fundamental
...,
= Ai
...
^^2,+,,
set.
1.
.Asp, .^jp.,.!
We may
...,
Ag.
Wef
associate therewith
which
it
article.
fundamental sets,
combinations of the constituents of which, the constituents of the systems
(A^P), ..., (AsP) can be represented.
*
Where
(n, k)
denotes n{n-l)...{n-k+l)lk
2fl-
+2
characteristics (incon-
THEIR CHARACTER.
303]
The
303.
same
is
characteristics Ai,
We
character.
501
...,
on
Let
0-.
T ) be the usual
sign which
+1
is
or
sets,
according as
when
o-
is
cr
is
or
(mod.
4),
and
t =
"~ 1
when
s 2 or =
o- is
is
(-j, that
( -7
sum A^+i = Ai
of the
...
A^^+i
1
is
when
ff
(mod.
4),
when
when
when (t) = +
we may say
Or,
is,
3 (mod. 4)
<r
2 (mod.
when
For
if .4i, ...,
A^.i be
all of
A,A,...A^+^ =
cr
o-
character
Jj +
...
A^^^
Ai,
3,
where
Z,-,
Aj,
- 1) S
2i, Ij
...,
is
even,
is
even.
is
odd
+ 1^+, + S
1
=2
...,
A^^-^
li, Ij,
A^
A^^^
we have
0, .4,,
are odd,
odd
is
Ai, Aj
I,
also
Za
I,
...,
^at+i;
hence, since
J,Jj
=3
Au
...
...,
4).
the
character e or
(2<r+l, 1)+
(2.7
number
is
e is
this
- e according as 2A;4-1=1 or
of an odd number from
combinations
Thus the number of
has
-lai+i
(mod.
2^-'
.2(7+1
+ 2"-' when
<r
or
a=
(mod. 4)
otherwise
it is
A....;
a^O
SUMMARY OF
502
[303
RESULTS.
otherwise
6=-l
this agrees
The
it
sum
If A^,
...,
=3
(mod.
4).
thus obtained.
(2-v
r=0
or
=1
(mod.
4),
- 1)
now
It is desirable
304.
even when
is
It is possible, in
(2^'P-'
- 1)
...
- 1)
...
(2-
1)
of
2''
1,
Xa,
...
independent charac-
teristics Pi, ..., Pr, such that every two characteristics P, P' of the group
= 0, (mod. 2). Such a
are syzygetic, that is, satisfy the congruence \P,
group
is
is
(P),
It
is
same group
22p-r
(P),
we can thus
Gopel systems.
divide the
2*
possible
Among
characteristics into
there are
2^-*^
2p
2r = 2(7 we
systems by means of
2o-
we proceed
+1
to consider
1 characteristics A^,
...,
how
to represent these
...,
A, we can choose
fundamental systems.
A^+i
in
2<^'(2"^-l)(2'^'-l)...(2='-l)/i 2o-
EXAMPLES.
305]
503
of ^1,
this character
by
^ar+i, denoted
...,
combinations of
is
1, 5, 9,
...
of A^,
...,
that of (jj
is
e^^f^V
A^^^.^,
all
sum
the
Then
all
(^
and
the
all
the combinations of 3, 7, 11, ... have the opposite character. These combinations in_ their aggregate are the characteristics A^, ..., Ag. The characteristics
Ai,
To each
...,
A-,, ...,
may
itself.
The
This set
are azygetic.
2a-
+2
that
is
every three of
-|-
its
1 of
which
+2
J^,, ...,
A.^^^
satisfies
the condition
them
same
are of the
2<^^(2^_ 1)
...
character.
(2^-l)/\2a
the
There are
such root
sets.
It is not possible, from any root set, to obtain another by
adding the same characteristic to each constituent of the fonner set.
The
root sets are not the most general possible sets of 2cr
teristics of
2<r'+-2<r
(2^T
1)
As
305.
...
2^",
<r=p.
Of such
-f-
2 charac-
1)/| 2cr_+_2,
all
set.
When
(t
= 0,
(2>>+l){2P-^+\)
(2-fl);
constituents obtain one Gopel system consisting wholly of characteristics of the same even
character.
2''
/>
=2
289, Ex.
iv.),
by the addition of the same characteristic to each constituent, contain as many odd
characteristics as even characteri.stics.
When (r=l, we can, from any Gopel group of 2""' pairwise syzygetic characteristics,
obtain 4 Gopel systems, three of them consisting of 2"-^ even characteristics and one of
2"-! odd characteristics. The characteristics of the latter (odd) system are obtainable as
the sums of three characteristics taken one from each of the three even systems.
When
(7
= 2,
the
number
of fundamental sets
Aj^, ...,
24(24-l)(22-l) _^_
A^ia
504
^j
or
is
odd,
[305
IS 3.
is
Among
odd.
the
2^ = 16
characteristics yl
...,
characteristics A-^,
for
...,
Of the
in Chap. XI.,
P, = -P]
)
i ( 1 nrv ) ^^^ condition for the characteristics Ai, ..., A namely ^,
\000/
\100/
reduces to the condition that the first element of the upper row of the characteristic
symbol of
shall be zero
hence the 16 characteristics A^, ..., A, may be taken to be
where
"'
if
( AATv
i ("'
"^
represents in turn
all
p 2.
Taking next the case a = 3, there are ^=22" = 64 Gopel systems, (AP), each consisting
wholly either of odd characteristics or of even characteristics, there being
(P),
2'^~^
..., A
(2"
1),
= 28,
of these systems,
set of 7 characteristics
Ii, ...,Aj in
29
(26-1) (2*-
1)
_
-2^^'
(2^-1)
'
JT
ways; A^,
A^
...,
The
will
set A^,
..., A.,,
Ag
ia,
= (!)=
-!>
from one
which
root set
.^8,
sets
it
...
AgAiA^,
place
=36, root
...,
...,
Ag are azygetic
in threes
2),
l,
Hence the
characteristics
A^A^A^,
AgAyA.;, Ai, which, as easy to see, are not congruent to A^, ..^ A^ mod. (P), form,
equally with A^, ..., A,, a fundamental set, whose sum is likewise Ag they are derived
from Ai, ...,Ag by adding AgAiX to each of these. There are clearly six other uch
...,
fundamental
sets,
Hence to each
each of these even characteristics there correspond 8 fundamental sets. We can now shew
further that the even characteristics, thus associated each with one of the 36 root sets, are
306]
505
36 possible* even characteristics of the set J,, ..., A,. This again we shew
by shewing how to form the remaining 7 36 fundamental systems from the
system Jj, ..., A^, The seven characteristics A^A^A^, A^A^A-^, A^A-^^A^, A^, A^, A^, A^,
are in fact incongruent mod. (P), they are all odd, have for sum .^jidj-^S' which is even,
and are azygetic in threes for AgA2A^ia a, combination of five of A^, ..., A.,, and
in fact the
indirectly
A^, I^l
+ llj,
I4, I5I
+ II,,
2^, 35|
= 1,
Z4, I5,
Ag\=l,
modulus in each case being 2) ; hence these seven characteristics form a fundamental
There are 35 sets of three characteristics, such as A^, A2, A^, derivable from the
seven Aj^, ..., Aj each of these corresponds to such a fundamental system as that just
explained and each of these fundamental systems is associated with seven other fundamental systems, derived from it by the process whereby the set Ai, AiAgA.^, ..., AiA^Aj
is derived from A^, ..., Aj.
(the
system.
When
sum
their
will
(22P-2-
(22
1)
1) /
all
can, in
\2p+\
l^\
be of character e"* (j j
we
oi
We
306.
pass
theta functions.
^q)='$i{u
(f){u,a;
+ a; ^q)^(u-a;
^q),
where ^q
By means
wherein
immediately find
4,{u + in*, a
?)
= e^"'- **' q)
(
<i>
(,
P?)-
where \kq denotes the sum of the characteristics p, ^q; to save the repetiwritten in the form (cf. 294)
tion of the \, this equation will in future be
<^
when the
(M
contrary
characteristics,
K, Q, having
and
for
Thus,
+ ^,
is
KQ
each of
Q) = e^"""
""
q)
<^ (**'
^Q)
elements either
or ^.
Chap. XI.,
is justified.
506
We
shall
2''
[306
corresponding
61=
6"^
-P. I,
e/ =
...,
e"^'
^r
'
the signs of these symbols are, at starting, arbitrary, but are to be the same
|PP,i =
we may, without ambiguity,
the group the
6o
ei.J. *
2''
eij
1,
e,ej, ,
0,
(mod.2),
=
(p;),
compound
characteristics of
symbols defined by
(^, j,J
e^e,-
(^')
= e,e, e,
and
so that
e- j
= e;, y =
el />,!+- ^,
= .,e,,
p/)
Cje,;,- 1
>
e,-, i
^/^Q
1,
e, e,.
,
[^^^
etc-
* (a,
where
is
a;
A) = l
(^^)
6,-
<f>iu,a; APi),
Adding
to
if
then P^
PiP^, or
"
(/A)
P,-
becomes
order
all
we
all
D.p^,
obtain
= P^Pf., we have
G) ffl S)
(^*)
m a half-period
all
-.
= PiP^,
thus we have
<i>{u+np^,a; ^) =
6ie'"i^ii+2M;Pi)^(it, a; A),
= e-ie2MiPj)<I>(y^
* It preferred
the sign
,'
>
whose value
it explicitly.
is
1, may
a; A).
be absorbed in
tj.
But there
is
a cer-
307]
If
507
2fij,f
^{u + 2njr,
a;
A) =
eS^C"
* (w,
'^>
A).
Thus, X(m; Pt) being a linear function of the arguments w,, ..., itp, the
function 4>(m, a; A) is a theta function of the second order with zero
characteristic, having the additional property that all the partial differential
coefficients of its logarithm, of the second order,
have the
2''
sets of simul-
i.
If
>S
is
Ex.
ii.
If Pjt be
any
*
Ex.
When, as
iii.
(,
in Ex.
i.,
AP,) =
(^2)
'*
'
(/"),
prove that
*(">;
^)-
is
shew that
e"l'S'^l*(,
Conversely
it
II (m),
if
for integral
n
where
Vj
be syzygetic.
,
/j,,
. .
Vp are
for
in,)
11 (u),
= e''+'ii+-+>i>p n
independent of
Putting vu
A).
^m = P, \m = Q, there
{it,
A)^{v,b;
n(M-|-nj = e2^'""
with the characteristics
= PtUi+
u,
(u),
comparing
we
infer
this
+ fl^) = e^+'C'+i"".! n
{a
+ in,) = e2''+2"^+^-"'" U
(u)
and
I16I1C6
Q(w-^ np,) =
Q (u)
denote an analytical
satisfies
the equations
6,6-1^.1+2^"'^^.' q^^),
508
We
may
[307
combinations of the
the
all
2*"'' characteristics.
Then the
2^ characteristics of
the compound group are obtainable in the form PiRj, wherein Pj has the 2"
values of the group (P), and Rj has the 2p-^ values of the group (R).
every
2^+1
Since
are connected
by a
linear equation,
we have
i,i
and
G,
therefore, as e^e'^'^i'
all
Hence, adding to
zero*.
= ~^,
CQ (u) = t Gij
(^J^
forming this equation for each of the
et<^ (u,
2''
PiP.Rj)
we have
i,},k
herein put
Pa = PiP/c,
i,
Pjc
P^
also
and, therefore,
2'-GQ (M)
= 2 2 e,
[Sa,,,-
6i
e"i^.i]
(^j)
Q^ 4>(u,a;
P^Rj),
V-"'j /
where
and thus
2'-GQ{u)
= tGj^{u,a;
Rj).
Now
the
'2P~^
functions
"I>
(m,
for
by any
be of the form
* It
is
proved below
that, in fact,
is
not zero.
(u,
308]
509
2'"
functions.
308.
independent of
we
Putting for
u.
all
u,
u + Qq^, where
obtain
if ej, ..., tp
(), as before,
obtained by
together,
and
we have
SSfi'.^^^^f.^CK + a; AQiQa)S(u + b; AQiQa)=0;
now
let
group
Qj=QaQi, then for any value of i, as Qa becomes all the characteristics of the
become all those characteristics therefore, substituting
(), Qj will
we have
AQj)3iu + b; AQj),
must
Here
e^, t^,
where
Sffiu\
vanish.
f<,,-,
like
f^,
multiplying by j"\
The expression
as_;' is
we
cannot
characteristic of the
arbitrary,
'S.HicT^
we have then
and, therefore,
for,
we
infer
Hj=Q,
FIRST APPLICATION
510
Hence
[308
it
which
2y5(M + a;
for
AQ,){u+h; AQj) = 0.
When
u + a=
U,
u + b=
V (cf.
When
For instance,
if j,=
e**"'^'!,
it is
is
not impossible.
ft)
for all
and hence the equation does hold when A=0, a = Qq, 6=0, f = e*" *'
the values of fj, ..., ci^i, fi+i, ..., (p. For any values of the arguments u + a, u + b
we infer from the reasoning here given that if the functions S(u + a; AQi)S{u + b; AQi)
are connected by a Unear equation with coefficients, ffi, independent of u, then (i) they
are connected by at least one equation
'
j.
'
2fi5(tt + a; AQi) S
{u+b; ^ft) = 0,
for
one of the 2" sets of values of the quantities tj, eg, ..., and (ii) similarly, since the 2"
&(u + a; AQi) S (u + b; AQi) do not all vanish identically, that the coefficients
functions
by
are connected
25ier' = 0.
i
The
309.
307
result of
is
application (^ 309
important, case is that in which
examples of
its
of great generality
The
313).
o-
w^e
proceed to give
= 0, r=p,
and
to that
we
most
give most
attention ( 309311).
When
0,
. . . ,
<^{u,h-B,e)<P{a,v ,A,e)
^"'
By adding
'
'
<i>{a,b;
B,e)
e;,
except
we
obtain
2P"
"i!{u
e,,,
= 1,
+ v; A)'^{u~v\ ^) =
ej,
..., e^,
all
the
bearing in mind
<J>
S*
Wherein
it is
B,
e)
assumed that
vanishes.
a,
(m, h
B,
e) <& (a, v
Cf. 308.
A,
e)
^ia,b; B,e)
any one
of the 2p quanti-
309]
is
^(u,b;
By
511
B,e).
B,
e)
^ (a,
v;
A,
e),
or
"
f f {a) {b)
=!
'
^^') "^
2 (^^)
e"
^i
<^
(m,
^PO
f {a) (b)
all
BFj)
('''
'
(a, 6
..., e^,
e^,
we
results,
5P.-)
7 \AB)
increasing
and
u,
/j>p
6 each
'""'''^'
to
we have
be
'^ ^'''
^^^'-^
-^
<
''
^-^^'>
'
VIr,
T [ab]
taking
by the half-period
all
^""'^'^'""''
? (ii)
the possible
2^
^^' <^ (.
>
;
= 2-Pl2
=1
where A,
4^i)
5-PO </>(.
C = AB,
A)
(^^''\ e- ^^i
1
./.
(m,
RAPi) 4>{a,b;
RA PiC)
for (Ex.
ii.
S becomes
295), l.e"'^'
all
-^i'
2*
2^*
when Pi = 0, and
as
is
for
any
characteristics, so does
definite characteristic
RAPi.
APi,
C=
for
die
it,
is
Cf.
INDICATION OF
512
310.
[310
R
according as \A, Ki\
of 309,
2P-M2"Uiril^(^
S becomes
where
(for
= l,
fi),
J; Ki}<l>{a,v; Ki)
C=0
R becomes all
2*
2e
'
'
'
(^(m,
?;;
extends to
all
" 2
e"^ ^'^^i
1=1
AR)(f>{a,b; AR),
AR),
v; S)(l>{a,b; S),
characteristics,
= 2-''e"l^l2''2c"l^l<^(M,t,; AR)^{a,b;
where
in the formula
by Ki, we deduce
=^
where
...,
of that formula
= 0.
AB
<^ (m,
BKi)
<t>(a,v;
BKi)
here {K)
defined
may
be regarded as
315.
311.
The
function
<f){u,
v;
A)
functions <^(m, v;
is
;
a,
APi is reduced,
all its
in the preceding
elements being either
or ^.
supposed
reduced.
2f^(u +
v;
309)
Ay'i.U-v: ^) = S
'
^ ^'
'^^
^(a,b; A',)
Aj)
311]
COROLLARIES.
u h,
-B of 309,
+ V,
a v,
W.
U+V,
suppose a
u-\-v,
513
= 6,
v,
and put,
+ b,
for
b,
b,
respectively,
U,
then
we
V,
U+r+W;
0,
obtain
2P^(f7-+F;
ff
U+W, V+W,
A)^{U+ W;
A)
V and
adding to
^*^(
^+ ^; ^ ',Pk)^(0
2P[U, V; A,B][U,
1 2 Vi,,jt,j[U,
V,
^^~
W;
A,
W;
A',P,)
flp,
flf.,
this
becomes
(J\
W;
v,s\y.
A', B, C,
[V;A,B, Pj][W; A,
PJ [0;
A',
C. P,]
PJ
and,
'^
[U +
V;
A + B], etc.,
= e-i"(^'+v'H?,+?y),
In connexion with
st
'
then
V;
<,_ j,
(A
{, vi
e,-,
s^
= e-i'^iP'+T'i^i.
may be
W=-V, A +B = K, A + C = D,
Putting
(a)
an expression oi^[U
fii
K]%[U' V; D]
A'
deduced.
= D,
DPi], ^[0;
ZPJ, ^[0;
i)PJ;
DPi can be
Putting
^[U-
odd.
further
K=0,
we
obtain
an
expression
of
^[U+V;
0]
V; D] in terms of
'^[U; Pi],
%[V;
Pi],
'^[U; DPi],
DP,].
^[{/+ F; K]/'^[U+ V;
^[0; DPi], ^[0; KPi] with numerical quantities. In this expression the
it may for instance be taken to be zero.
is arbitrary
characteristic
B.
33
514
INDICATION OF
The formulae
by
6i,
as is clearly allowable,
following formula
2P'$t[u
Ex.
(cf.
+ v; K]^[u-v;
"7
[311
replacing,
and taking
D = 0,
e^-e"''^'
''<'
317)
viii.
0]
2 e^e-i^ia ^ (0
V + P.) ^
(0
'
P.)
K=^
where
,\
P,
= ^l""
1,
in regard to a extends to
all
is
the right hand does not vanish for any one of the 2* sets of values for the
quantities
e..
For p=l,
i.
HP
excluded
is
is
(cf.
zero,
where 5
.fi"^.
(u),
ii.
3p (u) denote S {u
By
0),
putting, in case ^
3 {u
F), etc.
= 2,
2
f1
fa
A=3,{u)3^{u),
A', B', C",
B=3^{u)3^^{u),
C=3^{u)3^{u),
Putting
D=3^(u)3,,(u),
a'.
= C, F=Tf=0.
22/Xi/.,i.,.
^' = ^, we obtain
[0
A, B,
P,]^
is expressible as a linear
function of the squares of the theta functions with the characteristics forming
the
Gbpel
^^{U;
A Pi),
system
{AP).
We
Cf. the
iv.
and
283.
311]
COROLLABIES.
Ex.
where 5sj=52
(7)
515
^ () =
< ^L
+ ^14
^2 ^0 ()
('')
- ^L ^O:
^?2 {^)
- ^l
^l ^l ().
(0), etc.
There
^ (m APi)
is
provided
+ b;
A, Pi)'it{a-b; A, Pi)
= 1e-^\Pi\^(u + h;
Pi),
APi
are even, as
u+v,
is
by taking
possible ( 299)
uv,
b,
b,
+ b, ub,
+ v, av
respectively
+ w, uv, a + b + w, a b,
then, putting a = 6 = 0, we have
u+
u+
+ w,
= 2e"l^'i^(M;
A, Pi)^{v; A,Pi)^{u
b,
+ v + w;
+w,
v;
A, Pi)
+ w; A,Pi)^{v + w;
A, Pi);
herein put
w=
fip^, v
= u + XI p,
1 (^/')e'"''^"MO;
...,
Pp
A,Pi)'^{0;
= S (^'p')e"^^<^^{u;
A,Pi)^{u; A,
P Pi)'^{io;
every characteristic of the group (P) can be given in one of the forms
becomes in turn all the characteristics of
Qb, QiPi, QoP^y Q>PiPi, where Q,
2^-"
characteristics putting
a gi'oup (Q) of
Now
we immediately
^{u;
Q,)
4,PQ)^(; A,PP.Q,),
find
= ^{u;
Qs,Pi)
= ^{u;
Q., P.)
= >|-(;
Qe,
Pu
^2);
where
Bm
0,
Pi, Pa, Pi
+ P-.-
e)
332
COROLLARIES.
516
we add
u the
to
we
[311
m=
t>,
a=
= 0,
the
result
A, e)^(0,u;
^^jS-1 ^(^. 0;<p(o,0;A,e)
^
^^^
d
PN^^n A,
4 P,)
P^ = .-p/-PA
%{2u;
^,P*)^(0;
2
/PA
"^
1 &'( 0;
U/7e,
A,
e)
^.g)
*(0,0; 4,
6)'
where
*(, 0; ^,
By
=2
6)
^(O, 0; A,
^ (2m
6)
=S
(-^'j
A, P^) given by
the formula above, there results the biquadratic relation* connecting the
functions
(8)
'^
(m
As an
APi).
indication of another set of formulae, which are interesting as
may
following
where \i,
and let
f{v; ^)
. ,
also be given.
g)
(u),
the
Let
(v)
= ^'
(v),
S'^ (v)
= ^" {v),
v,
^^-^> =
t)
and 6
0,
we obtain
?^^^^f^^'
wherein
Chap. XI.
217).
(for
Riemann
relation
d.
has also
Mat., 2* Ser.
312]
These propositions
312.
a; A,
{u,
now
Passing
517
e)
=p
a,t
to the case r
=p
there
l,
is
its
importance.
^i(ii, a ^,
'
There
= 2
6)
(J)
6i^ (u
+ a; APi) ^ (m - a
APi).
is
this relation
we have
a-
= p r =1,
2^"^
A, whereof three are even and one odd, which are such that all the
2^^' characteristics (AP) are of the same character.
Taking the single case
in which these are all odd, we have
istics
(u,
<E>
hence,
a; A,
= ^(a,u;
A,
and
e),
a; ^,
<t>(a,
e)
0,
if,
\4>
wherein
\,
fi,
a; ^,
(,
/4
* (b,
A,
/i*
A,
{u, b;
)-\-v<i> (u, c
independent of w, we put u
v are
e)
= ^{c,a;
A,
e)
= a, we
^{a,b; A,
A,
infer
e);
is
e)
<I>
(m,
a; ^,
e)
(c,
A,)^ (u,
+ ^{a,b;
A,
b;
e) 4>
A,
(i(,
)
;
A,
e)
= 0,
or
2P-120-1 ,p.\ fP.s
where
rfr(t,
j)
APj)
APj)
APj).
Adding together
all
2J.-1
Se-rfl^il-f (i
i)=0.
which all
l, this is the so-called equation of three terms, from
For instance, when
When ^ 2, it is an equation
relations connecting the elliptic functions can be derived.
p=
of six terms
fifteen
such equations,
all
expressed by
= _e''^l^^l
a, &, c
A and
518
[313
sixteen characteristics
such that
all
Denoting the
(AP)
are of
odd characteristics in any order by A^, ...,Af and an even characteristic by A, there is an
equation of the form
Xi<I> (u,
A^,
e)
them being
+ \^<^ (u, a
A^, e)
odd.
six
e)
have
( 306, Ex.
L),
* ( + fie,
0= -42^3,
we
A,
u,
and recalling
e)
= e^<>
AA,A
('^^') *(,;
e)
in
+ D,c,
where
infer
^( A
')^('^''^' ^i^2^3, e)
=
(
')
* (a,
A,A,A
e).
A2A3, we
/A A
finally obtain
L'^'j*(a,
a; A,A^A;)^{u,a; ^,)
IA A \
+ L'^'j*(a,
a;
Ex.
Ex.
*(a, a; ^,^.4,)*(!t, a;
A^
e is
289).
viii.,
53,534 (2i))
where 534=534(0),
314.
(A A
(^^'^'j
The
tion of the
etc.
When
v=
this is
j>)
5^^ (m
( + jj) 5((jj - v)
by a
repeti-
(P) = 0,
{Q)=o,
P,P
...,
Pr,
...,
Q Q,Q
...,
...,
P,P^P
QiQ^Qi.
...
...
respectively of
2''
and
2' characteribtics
the
2''
+'
314]
519
then with
thus
if
A be any
A,
E)=
* for
sum extending
(^^ Bi^^Siu + a;
(^
J)
f,
( J^ )
By
...,,
f,, ..., f.
characteristic
*Ka;
ej,
the theorem of
ARiJB{u-a;
ARtj)
2""
{u,
a; A, E) by taking
values of a and A, and the same group (PQ), are linearly expressible by
^-''-'='2''-' of them, if tT=p r, with coefficients independent of u. The 2' functions
* (m, a AQj, f), obtained by varying a and ,, are themselves expressible by 2" of them.
different
*(m, ; ^, .')*(o, 6;
A, E)
or
taking for f
we obtain
..., f, all
2e'^ljl*(M,
t);
AQj,t)i{a,b; 4y,
\X,Pi\ = \Pi\,
306,
Ex
i.,
(mod.
2),
...,
and
A^
let
e)
are the
C^=A^A
=2
'2,^"
e"l''
* (m,
2'
^y,()*(,; ^9;,
0-
then
u and
C, P;
6,
hence,
by the formulae of
we have
is
zero because
j=0, when
its
in turn,
si^ecial
value
is
22'*(,6; A,.)*{a,v;
.1,
^^".i.
')
K ^1
^^>
^)
520
[314
with
( 298, Ex.), if a characteristic S which is syzygetic
every characteristic of the group (P) be added to each of the 2^' characteristics A^, ... ^x,
now, as already remarked
,)
= e"^'^'^^i
= e"l<^l*(M,
i>;
^C, c)*(a,
AC,
6;
i),
and
sV
^-^^
(,
17
^ C^Qi,
A C,Qs,
,)^{a,h;
e)
70=1
220"
m=l
and therefore,
have
dividing by
finally,
factor
a,
2"
(there being
2'^
characteristics in ()),
we
22(7"
2''*(k, 6; J,
0*K
When
this
o-=jt>,
>;
0=
A,
becomes
the formula of
We
309.
c)
(a, 6;
infer
AA^A,,,,
).
are connected
e)
Ex.
A .v-^'^- K^;
J,.)-2
^"^biuv
^[u,v,
2
315.
The formula
a group of
A,E)^{a,v;A,E)
A,E)
*(a, 6;
characteristics,
2(*
text,
...,
R^
Let {R) be
\R,P,\mO,...,\R,Pr\ = 0.
Thus (P) is a sub-group of {R) the group {R) consists of (P), together with groups {RP),
whereof the characteristics R form a group of 2'*"'' characteristics, whose constituents are
;
be denoted by
conditions
is
iJ,, ...,
{R).
Then, if i^, G' be arbitrary characteristics, and Ri become in turn all the characteristics
of a group of 2**"'' characteristics of the group {R) which are incongruent mod. (P), we
have
^-1^
2V'^^^'*(m,
= 'ip-p'
GRi,f)i{a,v; GRi,
6;
2 e^\^GRi\ 2 e""^''"'*(u,
m = l,
Since \Ri,
S"-" 2
m=l
= 2'-''
P = 0, the
|
2, ..., 2^",
hence
i;;
,),
m=I
1=1
t)
are in
( 306,
Ex.
the series
2 e'"''^^^''+"l**<^"''*(,
i;;
is
G'C,,
equal to
1=1
2 e^\F<^\+^\Cm\
^^
2 e"l
*'"'''
^')*(M,i); f^^,
c)
(a,
6;
GC^,^);
316J
vM-r
iri\LR
now 2
case
value
its
where
521
is zero, iinless
'
is 2''-''
\L,
Ri\=0 (mod.
2) for
iii
which
equal to
is
5, satisfies
= 0, FGC,i,=S
ditions
the
number
for the
modulus
/i
conditions
(P) is 2^P~''"''
is
2^^"''
the
number
of these
2^"'+''"''.
stitute
""'*+''
characteristics
modulus (P)
putting
fi
for the
= r+p,
2P
i=l
= e^l-f<^l ^'s
>I-P,nl *
{u, o;
F.%,
m=l
0* (,
6;
FS
.).
groups (R),
316.
We
286, Ex.
i.)
formula can be obtained for the cases p= 1,2, 3 by the application of one
We give now a generalization of that rule, which furnishes results for
rule.
any value of ^.
Suppose that among the 2^" characteristics A^, A^, ..., A^ which,
Gopel system (P) of 2'' characteristics, satisfy the conditions
for
any
\X,P,\ = \P,\,...,\X,Pr\^\Pr\,
302
certainly possible
characteristics B^,
is
not equal to
of this chapter,
when
\BBiBj,
o-
1,
or
2,
...,
j,
and
or 3
B is even, which
an odd characteristic as
B^, B, of which
BBiBj
is
286, Ex.
i.,
is
and, since
\P\,
* The formula is given by Frobenius, Crelle, xovi. p. 95, being there obtained from the
formula of 310, which is a particular case of it. The formula is generalised by Braunmuhl to
theta functions whose characteristics are n-th parts of integers in Math. Annul, xxxvii. (1890),
p. 98.
The formula
522
among
[316
so that all
A),,
...,
Then by
307
\^(u,c; B,e)= S
wherein the coefiBcients
equation m
=a+
^bb-
X, \i,
\* (a,
Bi, e)
306, Ex.
e),
independent of
\^, are
...,
then we infer
X,4>(m, ; B,,
= \j<E> (a, a
Put
u.
in this
i.)
B,
e)
hence we have
*(a, a; B, e)^{u,
= I
c; B, e)
5,
e),
m=l
which
is
Adding the
2''
there results
%{u-c;
BP^),
m=l
i=l
the
i=l
where
Vr
ylr{BJPi)
c;
BPi)
B^Pi).
2a,
c,
c,
a-^-
c,
c,
+ a,
respectively by
U, V,
and thence,
Chap. XV.
p=
in case
Or we may put a =
0,
2,
or
and
p=
3,
xi.,
286,
so obtain
22
e"l'.-B^*i^=(M;
m=l 1=1
5Pf)y(c; B^Pi).
Chap. XV.
Ex. When <r=l there are three even Gopel systems, and one odd;
[B^P) be the three even Gbpel systems; then we have
i{a, a; ,
* (,
f)
B,
We may,
if
A snchthat M,
we
Pi|
t For similar
c;
B^,()^(u,a;
.)
+ e'''^^^'^ -l>{a,
= |Pi|,
(i
results, cf.
p. 327.
{BP), {B.^P),
e)
= e"l*-8.l*(a,
*
let
= l,...,
on the
c;
left
B^, t)*
(u,
a; B^,
,),
V).
Frobenius, Crelle, lxxxix. (1880), pp. 219, 220, and Noether, Math.
EXAMPLES.
317]
where *
a; B,
(u,
i)
523
for instance
when
ju
= we
1
obtain
= e"l*^'l5(a + c; ,)5(a-e;
5,) -9(^+0;
B^)^{u-a,
317.
If
i.
4 + ^),
5(m;
5(t4; .4)
a fixed
be
B^)
B^).
characteristic
prove that
and
if
^i,
..., Bjc,
X, P\=:\P\, such
5 be
that,
'B)
;i:+l=2P""i +
when
is
not equal to
j',
BB^Bj
we have
odd,
is
307) an
equation
2P
j5,),
m=l
where
is
*4,
P| =
putting M = Oij+S2^., we
obtain
B);
therefore
*(0; 5)*(; J) =
Ex.
0-,
=1,
Bi,
...,
Ex.
(5j*(0; A + B + B,)i'{,i;
B (Ex.
From
iii.
3 respectively, so that we
2,
i.,
286,
the formula
B,).
when p = 2, 3, 4;
know how to choose the
ii.
=p \,
Bi,
___2^(^^5^j
i.
in these cases
characteristics
309)
for a,
^{u;
where
.4,
R)S{,a + b;
R)d{a-b;
R),
A)3-'(a;
(^Q 5M;
R)^Ka; PR),
are any half-integer characteristics and R becomes all the 2^^ half-integer
we also have, from this equation,
for
putting
;
characteristics in turn
3^(u;
RP
^)52(a;^P) = 2-P2e"l^l(^)e"l^^--Pl52(;
iJP)5^(a; R);
therefore
or
for
Pif).
The
=0;
values of .R
may
EXAMPLES.
524
conditions |^| =
E = E,R = EP;
thus, provided
|-ftP|
^,
i'
5M;
+ P =
(mod.
A)S'(.;
by
(mod. 2);
[317
= 2,
and
2),
(i),
2i'-2(2p-i + 1)
wherein 5'^(; A) denotes 5^(0; A), etc., and, ou the right hand there are
=
=
(mod.
only
one
of the two
^P
which
for
values
of
to
2),
corresponding
terms
values, E, EP, satisfying these conditions being taken.
P=0, u = a,
Putting
.9'(w; yl)
in the
= 2-P2e"l^^l5V;
R); 5*(;
R
so that,
by addition,
2-"2e'''^*'[l+e"l^'+'^'l^'-P|]5*(tt; R);
5V; 4) + e"'^"^'5M; ^P) =
thus, as before,
p%
+ e"l^- -P' 5*
')
^P)},
(ii).
(P)=0, Pj, Pg, P1P2 be a Gopel group of even charac( 297) that the Gopel systems
(P), {B^P), {B^P), {B^B^P) constitute all the sixteen characteristics; each of the systems
{B^P), {B^P), {B1B2P) contains two odd characteristics and two even characteristics.
Then, in the formulae (i), (ii) of Ex. iii., if P denote any one of the three characteristics
Ex.
Taking
iv.
teristics*;
let
let
are ., P| = |P| = 0, E\
Pj, Pj, P1P2, the conditions for the characteristics
+ l), =6, solutions of these conditions must consist of 0, Q, and P,
|
2. 2P-2(2P-i
where
group
is
(P),
the
QP,BP,
defined
and
= Q;
is
[B^P), {B^B^P). Hence, when P=Pi, we may, without loss of generality, take for the
2p-2(2''~' + l) = 3 values oi
which give rise to different terms in the series (i), (ii), the
B^; similarly, when P=P2, we have, for the values of E, E=0, Pj, B^; and
when P=PiP2, E=0, P,, B^B^; taking .4 to be respectivelyt B^, B^, B^B^ in these
values
cases,
0, P.^,
we obtain the
Q)52(;
3M;
Qy^i;
o)52(;
six equations
o)-i-5(; Pi)+e"''-^'^'l[5(;
S^; 0)+S'{;
Cbb)'^'^'
P2)
^^^''^'
+ e"l^^-P>l[y(;
AA)+"'
^^^'''
o)+SH;
PiP2)+e"l^""'''' [5^(;
5'
'(//},)
'
B^{;
P2)^'{;
A)+"'*'-^"'(5f)>j5^(;
'
(;
Pi)^H;
p,)
-S'^i; B,B2)9^{;
'
B,B^P,P2)=0,
We
is
odd
is
^l'^]
EXAMPLES.
wherein
wuereiQ e'"l^'^2l_<,i"|B3iil_.iri|B,Bj,p,
e
_
-e
'
525
mu
=-!.
^
,,
'
all
r.
where
We
V.
'
represents a set of
= C-\-K^,
q, r represent the
we immediately
say; then
same
or
1,
half-integer characteristic,
=J
values.
{^\
+i (?),
find
( 291)
U.J
or
1,
independent of
is
= fi'+Ci'
(f'c')i
Tc^.
in
K^, K^,
Kp, K^K^,
...,
...,
KiK2K^,
...
/0,0,0,...\
""^
.(0,0,0,. ..\
J.
/0,...,0,0\
then denoting
included in the equation
wherein
= 2p,
tj',
..., f,'
The
order, the
^ to
is
5i
f' c',
m
for
'
When
ingly altered.
is
c',
and
is
this is
i;
c'gc'
''
is
f ''''
systems (CiK)
^j
in turn
by
Cf.
291,
Ex.
v.,
C.
one equation*
Chap. XV.
being correspond-
or 1, and has
either
f'^ denotes a row of p integers, each
matrix the element written down is the /3-th element of the a-th row.
in an assigned
wherein
We
e'
...,
*'
all
it
Now
^,
^''-i[o,...,o,i)-
left
and Caspary,
+v'^''
d.
values.
In each
CaKfi]
2''
^ [u'"' - 1;*^'
324
CaK^].
Frobenius, Crelle,
1882,
i. xxvi.
Ev
If in
vi
consist of
Hoo)
and the values of f
* (oo)
'
'
Hoo)
(0,0),
rooi
no~|
LiiJ'
LooJ'
~Loij'
_ron
'
nn)=(
(1,1),
(1,0),
(0,1),
final
-"2.
3.
"4>
equation of Ex.
"i)(
are equivalent to
v.
ft-
-^4'
^^y
^2)
^2'
ft'
ft'
^l
LioJ' LoiJ
".l;
W' bu
boJ'
LooJ' "LiiJ'
LoiJ'
-boJ'~LoiJ'~biJ'
wherein, on the
^ (oo)
'
are, in order,
[317
EXAMPLES.
526
left
hand,
"l.
"21
"3>
~4
ft>
ftl
ft'
ft
"2)
-"IJ
"41
"3
ftl
ftl
~ft'
ft
etc.,
and on
boJ
L^l denotes
+ ; i
jj j
- v;
^^ jj
4].
-.=^{;
respectively the
ft) ft) ftl ft lieing
Now
i{
A,
iQ].
''^=^i--'iQ]' -3=^.[-:
iQ].
''4=5{;
v.
first
linear equations
(yi) ^2) ^3)
^4)=^
'"4)>
(^li
%)
^3.
H)
,r^2+:r22
+ .r32 + .r/=(|3,2+/322+/332+ft2)
{,2
+ ,^i + ,2+,2)
2
1=1
2y?,.. 2
yL= 1=1
1=1
2 v" = 2.y^*"
"' J
j=i
, n
'
fw^
'^^
y,.
,.y, ,
= 0,
(-4=*, r, s
j=i
["i^z]
>
KC2],
-M
-[cc,]
I'^i'^i]
''
2, 3, 4),
>
Kci]. -[ic]i
[i]
-[cci]
-[cicj,
side,
algebraic proof
may
be given
cf.
and
we may
["ia]
[0]
An
= l,
2 7 ,v ,=0.
*
j=i
changing the second and third rows of the matrix on the left-hand
the result by saying that the matrix
the matrix
44
Denoting
is
inter-
express
Ex.
527
A MULTIPLICATION FORMULA.
317]
vii.
Deduce from
309 that
where P^, P^ are characteristics of a Gopel group (P), of 2p characteristics. Infer that, if
n he any positive integer, and A Pi be an even characteristic, S{nv A Pi) is expressible as an
integral polynomial of order n^ in the 2f functions S{v AP^).
;
Ex.
viii.
If
^ =4 (*")
>
-P.
=i (* )
deduce from
309,
putting
a=h = u U=vV=^Qt,
that
where
x{u,v}=Sf.e-i'^9-B{u; K+P^)S{v;
P.).
[318
CHAPTER
XVIII.
of
first
kind;
it
is
method
number
of positions
equivalent
positions
We
183, Chap. X.) of determining the places mi, ...,mp from the
it
when
is
by a
is
all
these possible
w,, ...,Up,
138
140
(cf.
mi,
2h(o'
=b,
But
i}
= 2aa),
it is
t/'
= 2aa>' h,
References to the literature dealing with transformation are given at the beginning of
Chap. XX.
319]
want
of
529
LOOPS.
to enquire
what
new periods,
new theta
original ones.
So
far as the
will
it
we
Riemann
when the
of this chapter
Riemann
is
surface.
The
is
following chapters.
Riemann theta
side*.
We
Chap. XXII.
319.
call the
direction
can be described, choose either, and call it the positive
described
curve
is
the
when
hand
left
the
side of the curve which is on
;
starting
now
[319
530
been described
is its
positive direction
The
it.
(.4i),
is
direction, from
is
(Bi)
The period
to its beginning.
integral on the left side of the loop (^,) over its value on the right side, and
may be obtained by taking the integral round the loop (A^) in the positive
direction,
may be denoted
respectively
by
fi,
and
left side.
These periods
fli'.
It is useful further to
(A-,),
is
reversal of the convention as to the positive direction of the (original) loop (A^).
periods associated with the (new) loops [^i], [Bj] be respectively denoted
we have, therefore, the equations
[aJ
= Q',
If the
by [Q] and
[O'],
[Q']=-Q.
These equations represent a process of interchange of the loops (.4;), (B^), with retention
of the direction of (Bj) which may be repeated. The repetition gives equations which we
may denote by
{O}
= [Q'] =-a,
{a'}
= - [O] = - Q',
from the
and
left
not necessary for the purpose of the dissection of the surface into a
simply connected surface ; but it aflfbrds a convenient way of specifying the necessary
condition for the convergence of the series defining the theta functions.
The
321.
is
pair of loops
...,
(^i), (B^)
to
(Cp-i),
curve.
Upon the p-ply connected surface formed by cutting the original surface
along the loops (Ai), (B^), ..., {Ap), (Bp), the Riemann integrals of the first
and second kind are single- valued. In particular if Wi, ..., Wp be a set of
linearly independent integrals of the first kind defined
W^
at the loops
(A^^), ...,
(Ap) are
by the conditions
all zero,
except that at
322]
(Ar),
which
is 1,
and
be the period of
if Xr,,
Wr
531
+2Ti2Wjn2+
TiiW,''+
is
+rppnl
with e.g. the retention of the direction of (B^) and a consequent change in the
sign of the period associated with (Ar), as explained above
if the
( 320)
loops (Ar), (Br) be interchanged without the change in the sign of the period
associated with (Ar), the imaginary part of the corresponding quadratic
;
form
is
negative !.
shall call
denoted by
(ar),
by 2a)i_ r, 2<o'i, r
and the periods associated therewith by
',
the integral
Ui,
Oj^ri
left side
j=l,
(r,
[2a)i
,.],
Ui,
is
associated therewith,
denoted by (Ar),
[2&) ';,.].
(Br),
may be
2, ...,
(bj)
is
,,
number,
eqviivalent to a certain
the positive
(aj) in
[a>i,r]=
(oOijOj^r
+ o'ijCtj^r),
(^=1,
2,
. .
p)',
i=i
similarly
J= I
(m. j^j,
a,-,
^'i,r
(r
1, 2,
. .
. ,
p),
';, r-
r,
Thus,
if
[to]
where
a,
a.',
/3,
/9'
<oa +
as
a,
[o)']
= o)^ + w'^',
And
p. 247, note.
contribution of - 1 to
in the positive direction furnishing a
+ As previously remarked,
+
circuit of
(b,)
o,,,.
342
If Li,
[322
532
...,
Lploe
a,
...,Up, as in 138,
and the period matrices of L^, ..., Lp at the original and general period
loops be denoted respectively by '2r], 2??' and [2r]], [2r)'], we have,
similarly, for the same values of a, a', /3, /8',
[t]]
= r)a +
t;'a',
= v^ + v'^-
[rj']
We have used the notation Up for the row of P quantities 2tP + 2<b'P',
where P, P' each denotes a row of p quantities we extend this notation to
the matrix 20)0 + 2(B'a', where a, a each denotes a matrix of p rows and
columns, and denote this matrix by Qa similarly we denote the matrix
217a + 217'a' by if
then the four equations just obtained may be written
;
',
[2]
Noticing
now
n.,
[2a,']
= ^^,
[2^]
= if.,
[2(], [2&)'],
[2v']
[2r)],
= H,.
(I-)
must
[2t]']
satisfy the
we have
by the matrices
2a),
2eo',
2r],
2r]'
and
similarly
- [ST [^'] =
[^'] [o-T
_
i (zr^n^
- ii^ir^) =
os^'
- ^'ys) ^Tri
thus we have
o/S'
a'/9
= /S'a - ySa',
aa'-S'a
respectively satisfied
by the matrices
for
by
= 0,
^yS'
- ^'yS = 0,
(II.)
to,
by
can be
= l=yS'a-a';S, a^-l3a=0,
a'^'-fi'a'
= 0.
(III.)
if
'-(r.)
denote the matrix of 2p rows and columns formed with the elements of the matrices
a, ^, we have (cf., for notation, Appendix ii.)
{a,P\f
B;
\a',^'J\-a',
-m_/af-/3a', 0a-a^\
/I ON
\a'^'-^a', ^'a-a'^J
\0 ij
aJ
>
a,
/3,
TRANSFORMATION OF INTEGRALS.
323]
and
533
therefore
and the
a = [at]fi' [a']a',
a'
=- [m] ^ +
form
[<'] S,
If
denote the matrix of p rows and columns whereof every element is zero, and
denote the matrix of p rows and columns whereof every element is zero except those in
the diagonal, which are all equal to 1, and if e denote the matrix of 2p rows and columns
given by
1
=(;:-;)-"'=(";:>
then
it is
(II.), (III.)
JeJ=f,
t/fJ=t,
where
'=a:^
and
it will be noticed that the equations (III.) are obtained from the equations
changing the elements of J into the corresponding elements of J.
It follows*
+1
or to
Ex.
1.
(II.)
by
from the equation JtJ=^( that the determinant of the matrix J is equal to
It will subsequently ( 333) appear that the determinant is equal to + 1.
p=2,
/
4,
"-(,
4,
-20\
ij'
/-3,
"=1-8,
a_/-29,
^-(,-28,
-6J.'
22,
-124\
^=V
56,
43;
20\
-TJ'
124\
323.
Wp, other
...,
integrals Wj,
= Mi^iWi+
Ui
...,
+Mi^pWp,
{i=l,
2, ...,p),
138) denoted
Dw?" by
iii{x),
in{cj)
by ^;
Cf. 142.
534
denote now,
Bwt "
also,
by
pt (a;),_and
/i
= pM,
[323
gives
(cj)
by p
138)
= p-^H"' - 2MaMw''-
ML"- "
"
thus the integrals of the second kind associated with Wi, ...,Wp are the p
integrals given by ML"''^, and, corresponding to the matrix a for the
integrals
L^'",
= MaM.
- 2^, 2^'
....
we
Lp",
have,
the
integrals
the
ML''-"',
matrix
and
that
m = Mv,
therefore have
<b'
?'
= Mv'
= 2MaMv = Mv,
^
it is
so
",
139)
?=2av,
we
for
^'
2ai;'-|(0-'A,
and
(IV.)
v,
v,
f,
vv'
v'v
The preceding
324.
- r r = o, vi - vr =
rr
0,
ttz-
^d'
^'v.
It
is
of importance to
Let
&),
Q)',
r),
the conditions
is
7)'
(i)
a symmetrical matrix,
(iii)
rj'
that
the
n^, ...,np
ca
quadratic
form
= r)a)~'^a>' ^Tria"^.
(ii)
that
(o~'ci)'n'
r)0}~^
The conditions
is
a>~'^a>'
has
its
a sym-
(v)
2hiD
are
= iri,
2ha)'
= b,
r]
2aa,
t)
= 2a&)' h,
^ /j^
<2')
the condition
(iii)
ensures
the
by
_ 2e"+2*K(+')+5(n+g)2+2inQ(+e')
[o)'],
also of
where
= a + m'a', = ifia,
a, a', /3,
say,
[&>']
say,
are
324]
integers*,
it
not vanish
Ui= Mi_^Wi+
or M
= Mw,
535
where
...,
[co]
does
Wp defined by
+Mi_pWp,
(i=l,
2, ...,p)
denoted by Vij,
when
ilf is
..., Vp^j,
Wj,
w.p
column of
2Mv =
The
[o)']
[o)]
= ila,
^(m; ?)
function
investigate whether
M,
matrix
is
a function of w,,
...,
it is
may be regarded
Chap. XV. 284).
(cf.
and the
as a theta function in
Let the arguments Wj ...,Wp be simultaneously increased by the constituents of the j-th column of the matrix 2v
thereby Mj
Up will be
increased by the constituents of the 9-th column of the matrix [2], and,
,
since
a,
= (H^yi>
Lj
(a)'-"
a,
yS, /S'
a',
2a)a
2&)'a', 217a
27;'a'
(m
q) will (Chap. X.
+ 2^ [(a)'-"'
(a')'J
Q'
(a')'-''
Q],
(fla)'-",
where
[u
and
e^i
respectively
_;-th
inw^, ...,Wp,
where
= r{2^,j,
(wi,
...,
similarly {vij,
...,
2fpj)[(wi,
...,
Wp)
+ {v,_j,
....
+ 27riKj,
Vpj)]
Wp) denotes the row letter whose elements are w,, ..., Wp, and
..., Vpj) is the row letter formed by the elements of the _7-th
function
Li'
S-
(m
|) takes a factor
= {H^yi^
[m
+ i (ft^)'^'] -
same value
Z;=r(2?'.,,*
We
vanish.
have
e^'',
o, a',
viu-^[i->]
of
r,
-rri
where
(^)"'
this can
2rp,>)[K,
j3, /3'
= iria + ba'\ we
(/3')'^'
- (/3')<^> Q],
Wp)
is briefly
+ {v\,j.
v'p.;)]-27ri^,-,
considered in chapter
XX.
7ria
+ 6a'
does not
536
where Kj,
?'i,,-,
..., f'p,,-
MiHaYJ^w, =r{2^^j,
or
In these equations
p.
2^p,j){w ...,Wp),
...,
(i,
respectively
...,
[324
?,
f'
denote
^HMa = i^.
and
21/u
therefore, in particular,
(V.)
comparing the
( j,
two sides
of these equations,
where, as before,
is
(u)'-"
necessary for the identification of the two forms of Lj and Lj, are
K; =
(a)o-)
Q'
this
(;S')'^'
Q - 4 (/9)'^' (^)'^'.
(VL)
matrices
fia',
p^'.
The equations
suflScient
efifectively
to
it is
it is
'
Iv-V};=0 =
I'v'
- V-'r,
Iv
- Vf = iTTi,
(VII.)
<^(w), it
where
Lj,
L/ have the
<fi{w
+ 2lWl +
The
...)
= C^i
(ji
(w), (j
...,Wp) or
= 1,
..
,p),
2v'm')
+ 2Vr_j,
<j>{Wi,
verification is Included in a
the
more general
piece of
in Chap.
XIX.
324]
537
constants 2u,
Ir,',
2f,
The equations
2f
'
( 284, p. 448).
v,
v\
f, entirely analogous to the conditions (ii), (iv), (v) of 324 for the
matrices u>, to', t), ij'. The condition analogous to (i) of 324, namely that the
f,
function, viz. that the quadratic form v~^v'n'' has its imaginary part positive
for real values of
rii,
..., rip, is
We
consider
first
(VII.).
In virtue of equations
(V.),
4r (?v' - V?')
= 2riri,
and, similarly,
HJD.^
but
and
(Hanp
this,
by the equations
(B), 140,
is
equal to
i7n-()S'-a'/8);
thus
0/3'-
0-/3
= /3'a-;8a' = r,
and, similarly,
oa'-oa =
and as before
_
0,
y9/3'-/9'/3
(VIII.)
= 0;
a^ = /3a,
a'^'
/8'a',
a^'
(IX.)
the relations satisfied by the matrices a, /9, a', ^' respectively being similar to
those satisfied hy m, m ri, r{, with the change of the ^tri, which occurs in the
,
r.
transformation.
Ex
(:.?.)(j:i)-'G:)-(l(-rO.
The determinant
+*.
of the matrix will be subsequently proved to be
of the
linear
538
Ex.
ii.
(M
Ex.
iii.
such row
If
x-^, y-^
y,
:;;,
letters,
[324
\ /2u 2u'\_/2<B
/a |3\
2o)'\
elements, and
Y, Xj, Y^ be other
JT,
such that
Z = ax+0y,
(.r,r)=(^,^,)(.,y),
Xj = a^i + 0y,,
then the equations (VIII.) are the conditions for the self-transformation of the bilinear
form xi/i x^y, which is expressed by the equation
when
Conversely
325.
function in w,,
(m
the
q
satisfies
Wp, of order
...,
matrices
a, a',
/3,
/3'
r,
equations
satisfy the
and with
the associated constants 1v, 2v, 2^, 2f'; and in virtue of the equations (VII.),
the determinant of v not vanishing, matrices a, b, h, of which the first two
are symmetrical, can be taken such that
we proceed now
to
The
quantity,
or
form hn?
is
matrix, obtainable
row
letters
respectively by t and
t',
so that b
+ iVj,
of real quantities.
cc'
where
x',
= ttit, b = ttiV
t/ + it^', where
be used
be denoted
be written,
will
v~^v' will
a~'&)',
are matrices
Then, putting
x denote rows
of
= VoMoCOo~'^x,
x^',
= T=ui}-\
we have
a)-^Mv
and,
x-^,
iCj
{x-i
or
to
[fiP
aT - i (SQ +
d.
Mat., Ser.
fi'Q')],
2, t. ix.
(1878
9).
326]
where P,
row
539
given by
so that
(aP
of
+ xf)
is
which, in virtue of
rxs (xi^
in r'x'x^
the
equations
Since x
in rrxxa.
may
we
the proposition
Of
equal to r
(PQ P'Q)
or
and
necessarily positive;
is
this includes
desired to establish.
326.
is
Ex.
(IX.),
is
equivalent to
and we
1,
is
of
ourselves
limit
we have
therefore, for r
1,
^(m;
We
suppose
satisfying
a,
/8,
= A'^{w;
l)
redundancy of disposable quantities we may for instance suppose <a, to', rj, t)'
and
to be given, and choose v, v, f, ^' in accordance with these equations
or we may suppose to, <', v, f and f to be prescribed and use these equations
It is convenient to specify the results in two
to determine M, v jj and rj'.
;
cases.
(i)
We
w respectively
replace u,
2v
by U, W.
1,
20)'
= T,
r]
b = 7riT',
(a
and, because
t}'
= r}T
a=
m and ? =
t;
-rria'
(o
W,W = ~ 0'-r'a') U,
0,
+ ra')-' = ~
0' -
r'a),
540
aU'^a'(B'-ra')U' =
-n-ia'
WU=7ria'{a+
[326
W\
ra')
These equations
@ (U;
g^(+Ta, Tf^^
where
is
I)
= A& {W
t';
when r =
to
deter-
is
The appearance
1,
(X.)
^)
K'
itself
the B-function, in
all
we put
If in 324
we
t;
lead,
mined from
would of
and
easily find
W^ = \r,a,-H^-\rCv-^W^;
7ria'(a4-ra')
thus
( 189, p.
a theta function in
2f=0,
(ii)
2f'
= -27rt,
left side
of equation (X.)
and
1,
is
2u' = t',
A particular case
of
(i),
a'
consists of zeros,
is
given
by the formulae
1,
2(a'
2u =1,
2v'
=t', ?
2<B
CT
Then the
= alf
function
= 27ri,
1?
= 0,
= 0,
= m,
=
-Ti,
f'
ri'
= a-> (/3 +
r'
(t/;
W, of order
function in
25"'
t;
')
t^'),
or
a = 0, h
a==0, h
t
@ [aF;
= 7n,
= 7',
(t'
-irir
b=
(ar'
-^)
a; |]
1,
2u'
and, in particular,
is
when r =
we have
independent of U^,
Works, Vol.
1.,
a theta
= t', 2^=0,
^{U;r;l)=A{W-T';l),
is
K'
where
7riT',
- /3) a,
r,
p. 156 (1889).
and
...,
(XI.)
Up.
^ = 1,
Cayley, Liauville,
x.
(1845),
or Collected
TRANSFORMATION OF CHARACTERISTICS.
327]
327.
It is
541
clear that the results just obtained, for the linear trans-
it is
the reader
who
and proceed
results,
desires
may omit
We
at once to 332.
now
= 1.
As
arising
a symmetrical matrix
is
ad{^^') + ^d{a<x')=rd{aB).
then we have
_
a'd
ad
d{&<x')diM) = ir+l)^d{j3oL') =
ir
+ l)-Zd{^'a)
(a'/9')
and
so that,
(mod.
ir
when r =
1 or is
is
z),
d(a^)d (a'y8') =
l)
(mod. 2)
The
= rd aa)
d'd (a/3)
(/3/3')
(mod.
2).
a)3j3'a-|3aa'/3=raft
and therefore
but if y be a symmetrical
where x is any row letter of jj integers, and z=dx, i/=^x;
yfi,
integers
+ ...+2y^^-it2 + ... is
letter
of
=y^itj^
row
^j
any
matrix of integers and t be
hence
for modulus 2
=d{y).t,
or
therefore
and
=yn<i+...+ypp<,,
= ynh^+--+ypp'p^'
,
(mod. 2)
{ad{^ff)-^^{aa')-rd{,^)'\x=0
(mod. 2)
"'"
and as
at once.
The second
any row letter of integers, x, the first of the given equations follows
in the same way, and
of the equations also follows from ^a-a'fi=r,
To prove
we
ba=c^+rc,
542
c=^d
[327
we have
p
1'
bi,,a.j^i=
j=l i=l
.)=1
i=l
c,-,yc,-,i
+r 2
c^j;
'=1
we obtain
i=l
1=1
!=1
Q = il,
Q'
= i9'>
so ^^^^
?> ?'
^^^^ consist of
integers
k = aq'-a'q-d {aa'),
In most cases
modulus 2. This
From
them as congruences,
to
these equations
we
Article
^ = dk+^'k' + d{a'^')
q = ak+^y + d{a^),
(mod. 2)
and
qq'
Further
if n,
= aii
(mod.
2).
(i
v'
we
= lck'
-v = 0/ - 0'^ - d (00'),
-a'ii-d(aa'),
rf (a'0'),
(mod.
2)
therefore, if also
<7'
we have
kj/'
k'v
by
or, in
which we
\A\ = \A'\,
2).
^\^j,i('^j,^y)hyA,B,C,
^(,),^[),^[],
A', B',
(mod.
transformed
call the
results ( 294)
\A,B,C\ = \A',B',G'\,
(mod. 2)
characteristic,
characteristic
syzygetic
{or
transforms into
azygetic)
an
characteristics
transform
characteristic,
into
three
and
three
syzygetic
{or
azygetic) characteristics.
Of these the
first
result is
into
-w.
r=\ from
the equation of
328]
543
(0
A') vanishes.
328.
^h
we may
write
=^4"' Mt)
*('')]
replace
m by w
i.,
n'
we obtain
= am'
'+m'\
'
where A'
m', q
is
n = 0m' 0'm,
u; h
q'
a'm,
+ inj
= ^'^
*(
k'
+ n'
+n
independent of
+m
namely
k'
+ m'=a
{q'
+ to') -
- (k + m) =M iq' + m')-
a {q
+ m)-d (So'),
B' {q +
m)-d{M^);
characteristic,
and of ^
theta characteristics
of
as
it arises
here the
when
sum
appear subsequently
as a iheta
as a period characteristic
(
n',
is
sum
of any odd
by integer
number of
characteristics
is
transformed as a
JX
'="+''
theta characteristics
JX^\
functions ^Z
are to be regarded as
in
""+'',
544
the
the distinction* thus explained somewhat diflferently, by taking as
We may regard
sfw+^Q,;
[328
S^u;
in terms
i
(^^
jJ
iQl, where
'-
r'
= d{aa:),
= d(^^),
f = /f+d(aa') = 5?'-5'g',
when
'
characteristic,' a
329.
case
where
r'
ments.
= d (oo'),
It ia
= d (^0),
aS =
which
will
make
olW
/3,
= /3'a',
the characteristic \
a/8'
- /3o' =
characteristic.
Any even
by
fCp
iC:
we may, momentarily,
the columns
may be
/k'\
call
then
of four sorts,
Q. 0.
but the number of columns of the
ffl.
last sort
0.
must be even
we
build
now
matrix
C
*
Theta characteiistics have also been named eigentliche Charakteristiken and Primcharakteristiken; they coDsist of 2'-^(2p-1) odd and 2p-'(2'' + 1) even characteristics.
The period
and Elementarcharakteristiken or
For them the distinction of odd and even is unimportant
while the distinction between the zero characteristic which cannot be written as the sum of two
different theta characteristics
and the remaining 2^P - 1 characteristics, is of great importance.
The distinction between theta characteristics and period characteristics has been insisted
on by Noether, in connection with the theory of radical forms Cf. Noether, Math. Annal.
xxviii. (1887), p. 373, Klein, Math. Annal. xxxvi. (1890), p. 36, Schottky, CrelU, en. (1888),
The distinction is in fact observed in the AbeVsche Functionen of Clebsch and Gordan,
p. 308.
in the manner indicated in the text.
characteristics
sometimes
relative Charakteristiken.
329]
of 2p rows
first sort,
Corresponding
we take
545
to a
column
ai,i=/3'i,
=]
but take every other element of the i-th row and i-th column of a and /3',
and every element of the i-th. row and i-th column of /3 and a' to be zero
corresponding to a column of the characteristic of the second sort, say the
j-th column, we take Oj.j = /3'j,j = a'^,j = 1, but take every other element of
the j-th row and j-ih column of a, 0, a', and every element of the _;-th row
corresponding to a column of the characteristic
m-th column, we take a^.m = Pm,m = ^m,m = 1, but
take every other element of the m-th row and column of a, /3, yS' and every
element of the m-th row and column of a' to be zero corresponding to a pair
of columns of the characteristic of the fourth sort, say the p-th and cr-th, we
take ap,p = jSp,p = y3'p, p = l, a,, = aV, = ^V ^ = 1 o-,p = l. ^p,o- = -l. a'o-,p = l'
^'^ = \, and take every other element of the p-th row and column and of
the <T-th row and column, of each of the four matrices a, a', /9, y8', to be zero.
Then it can be shewn that the matrix thus obtained satisfies all the
necessary conditions and gives k' = d (Sa'), k = d (y8/3').
and column of
/3,
to
be zero
o-
/O
^\0 0\
1\
\)''
is
546
Ex.
case,
is
i(j
[329
a=(
is
10
10
-10 1
^=(-l
),
0),
a'
= (0 0-1
-1
|3'
),
= 10
(
iO
1),
io
-1
5[;i(;
where
is
independent of
Mj,
...
J)]=45[-;0],
<^p.
We have proved ( 327) that if three half-integer theta characterbe syzygetic (or azygetic) the characteristics arising from them by any
330.
istics
2''
It follows therefore
Also the
Chap.
2^ Gopel systems
of
298, having a definite character, that of being all odd or all even, transform
And the 2(r -|- 1 fundamental Gopel
into systems having the same character.
systems
( 300),
which
any three
characteristics
chosen from different systems of these are azygetic, transform into such
systems
the
is
sum
all
What
are the
Is it possible to derive the 2^" Gopel systems of 298, having a definite character,
by
result: Let
An
D, D^,
...,
i^jp +
be
any 2p + 2
i<j,
331]
1=1,
2p + l, we have \D, A, Bj\ = l then it is possible to choose a halfand a linear transformation, such that the characteristics
547
ED, ED^,
ED^^^
...,
transform into
0, Xj,
where Xj,
...,
azygetic.
,.., Ajjj
i,
Xjp + i are certain characteristics to he specified, of which {hy 327) every two are
if
D, D^,
...
i^'jp + i
of which every three are azygetic, a characteristic E', and a linear transformation, can be
found such that, with a proper characteristic E, the set ED, ED^, ..., ED^p + i transforms
shewn that the characteristics X,, ...,X2p+i
into E'D', E'Di, ..., E'D^^p + iI* ^'^
200) by m"'
"',
"";
it has already been remarked ( 294, Ex.) that the characwith these half-periods are azygetic in pairs. The proof which is to be
given estabUshes an interesting connexion between the conditions for a linear transformation and the investigation of 300, Chap. XVII.
teristics associated
{:'.)
for
which
Sa'-a'a = 0,
^ ("'hr,
a'i,r,
^0-^'^ = O,
a'p,r\
where a\r
is
aff-a'fi=l,
\Pl.r) P2,rj
Per/
294,
etc.
a'=-a
and r= 1,
2, ...,
p; then
equation
is
the
(r, j!)-th
\Ar,Br\ = \,
\Ar,B,\=0,
whereof the
first
fourth each
\p{p \)
gives
a,
a,
fi,
\Ar, A,\=0,
\Br,
conditions.
may be
denoted by
(a/3'-a'/3)r,,;
thus
p{p-\)
{r ^s, r, s
conditions,
= \,
%,..., p),
Op \,a^hya^,
whereof the
first
ai'bjb.^,
a^h^a^b^a^, a^b^a^h^h^,
...,
by
a,'6i...6p_i6p, a^bi...ap'bp,
ai'6i...a'r-l*r-l*r)
for r=l, 2, ...,p, and a/6,a2 means the sum, without reduction, of the characteristics a,',
The sum of these characteristics is a characteristic consisting
6,, Oj, and so in general.
+ --'<.iy.i-- = ^'
= l,
2,
...,2p
Cjp +
352
i,
548
Thus the (2p + l)
Ci
from
c,,
[331
Cj,
...,
Cjp + i,
namely
= icj,
...,
= 1, ..., 2/), j = 2, ..., 2^-1-1, we have \D, A, Dj\ = l, so that ( 300, p. 496) there exist
connecting them only two relations (i) that their sum is a characteristic of integers, and
(ii)
a relation connecting an
If
-B, we
putting Ci=D'Di{i=l,
C^j,^-^
= C(C2C.^C^...G^f,--^C2-i
...,
...,
2p),
where
A = l,2, ...,2p-lY
C c\-\
taking
where C"2,.-i=
C'2r-i)
'^fi
2/)
equa-
tions
|C?m, ^2^ +
Thus putting
Ci
= Jcj,
..., C2j,
11
= 1,
(to=1,
2,
...,2p).
the i-th column of this matrix consisting of the elements of the lower and upper rows of
the integer characteristic aj or 6^, according as !;<p-|-l or i>p. We proceed now to find
the result of applying the linear transformation, given by this Abelian matrix, to the
half-integer characteristics Cj,
The equations
...,
Cjp + i.
to the characteristic
h'
-k=^q'-^'q-d {^),
= Mo
where Ai=^ai,
|-[rf(5a')]i,
Q=lq;
(i=l,2, ...,H
taking
=K'*i--'r-i*r-iar,
in turn,
-ki = \Bi,Q\-[d{m)l,
we immediately
*W(W^Hii.-.iio...o;'H<^w^Hii...ioo...oj'iU(W'))+Hii...i)'
or, say,
by
respectively.
Now
let
the characteristics
*'" Q'-
*(;)'"'oo-'
*(;)-0'"'
*(!)'
be respectively denoted by
A], X2,
...,
Xjr.j, X^rt
>
^zp + l!
X.-f-^Q, where
r=d(Bn
BBt
r-=d{aa')
331]
tion of the
and
sum
two
of
characteristic ( 327),
is
characteristics is the
the characteristic
it
Xj
^(
sum
54.9
),
hence, putting
E=^(
r\+DDi
the characteristics
EB,
ED
ED^^i
...,
X2P+1
...,
in
Sa' a'a=0,
which
(:A)'
a/3'-a'j3 = l,
^fi' ^'fi=0,
if all
is
but
infinite;
it
2,
is
number
is
(22P-l)22P-i(22p-2-l)22p-3
(22-1)2;
By adding
Ex.
the table of
p = 3, we
H-1
oj'
Similarly
(n^-l)n.
-1)'
\Ci,C^\
0' ^\
-1 oj' *(
1, 2, 3, 4, 5, 6,
Hi
...,
= l,
(t
oj' *(l
C7,
we
= l,
find, for
1)'
1)'
i<j, that
...,6;i=2,
...,7).
of the text
a'r-iK-iar,
Cjtr-j=ai'&l
ar=(V2
therefore, in this case,
/-I
"i
= Ul
<;i!r-3<''2r-2<'2r-l>
we
Another proof
is
C2r
Or
= l'*l
= CiC2
"'r-lVl^r,
C2r-3'' 2r-2'^2r)
find
/-ioi\
oo\
OOJ'
''>=("J-io)'
2.
obtain respectively
*(l
and
modulus
205, for
The equations
for
"^
= 1-1
ij'
^'={-W^)'
_/oo-i\
"^"Voo-i;'
*^=(oi
0)'
7 in
550
...,
for
text,
Cj is
(-1-1
[331
0-1
332]
COMPOSITION OF TRANSFORMATIONS.
551
are the chai-acteristics which arise in 200, Chap. XI. as associated with the half-periods
m"'"', u"'"'-, u"'" respectively.
conditions
S;, 2y|
= l,
for
The
characteristics Sj,
...,
2jj,
satisfy the
p{2p-l)
i<j.
332.
We proceed now to shew how any linear transformation may be
regarded as the result of certain very simple linear transformations performed
It
convenient to give
is
first
been shewn
It has
arguments
2r), 2tj',
first
order, in the
-K = fiQf -^'Q-^d
M,
_
2rJ ~
\ /2v, 2v'\
0, rM-'J \2^,
/2q), 2q)'\
/a,
27?'/) [a!,
[2v,
2<o',
Mw,
are
(^/9'),
/3
/3'
=
=
^irie ( 140, Chap. VII.), and
OeH
^irie, UeO'
=
re, where the matrix e is given by
equation
JZe/j,
-IN
= /O0-lN
Oj'
that also
AeA = re,
as in Ex.
i.,
324.
And, just as
in 324,
it
can be proved
(N,
\ 0,
sN-'}
last
(yy),
- Q, = BK' - I'K -
(2w 2a,/V_
27,
(2v, 2v'\
J - V2r,
h,^\.
2r7 W,
vfl,
^rd {W),
'
S'J
= UV, we
VeV = se.
or,
/2a)., 2a)/\
rsMN-'l
\2vx, 27;/'
/2a,,
2a,'\
^27;,
27;V
"
[332
SUPPLEMENTARY TRANSFORMATIONS.
552
from this equation we find as before that the matrix Aj, given by
satisfies
d (a,a/) = yd (Sa') +
d
{Mi) =
(Sa')
y'd
(/8/3')
+ rd (77'),
I'd
(y8/3')
+ rd (SS'),
(mod.
2),
Q/ =
a,Q'
- A'Q -
(AA'
).
(mod.
2).
rs.
may
be con-
order.
It is
is
equal
to
carried
first
is
associated
out,
factor.
One important
to.
The matrix
'-K-J' i)
is easily
Mi=rM,
A}c{k =
2ci)i
= 2(o,
ia called
2<a/
putting
it
Vi niay be taken
to be
= 2iB',
2);/
sxipplementary to
2i;j
= 27;,
(cf.
in place of
y, the
final
217'.
vii.).
2, 3,
-4fc
(/xi,
,
),
iJ.k\
where fxjc has unities in the diagonal except in the first and A;-th places, in
which there are zeros, and has elsewhere zeros, except in the )fc-th place of
the first row, and the k-th. place of the first column, where there are unities
let
333]
B=
553
-1
which has unities in the diagonal, except in the first and (p + l)-th places,
where there are zeros, and has elsewhere zeros except in the (p + l)-th place
of the
first
is
+1
is
let
indicated by
a=(i
),
which has unities everywhere in the diagonal and has elsewhere zeros,
except in the (p + l)-th place of the first row, where it has 1 let D denote
;
0-1
D = (l
),
-10
1
1
1
1
1
1
in the diagonal
first
is
1.
It
is
554
for
which
= /8o,
a/8
a'/8'
+ 2,)
1,
[333
matrices A^,
is
two processes, (i) the change of the fundameutal system of periods, effected
by the equations
[o)] = ma + w'a',
[&)'] = fB;8 + tu'/S',
= Tja + 7?'a',
\ri\
(ii)
[77']
'V/S
tj'/S',
by the equation
_
r = ^['7'];
leading to
[a,-]
of these
we
= Mv,
[a,']
= ilfv',
tu'a',
and
first process.
= wH +
[to]
K=M[v\
u=Mw,
[&)']
= &);S +
Q)'/3',
we obtain the
following results
= <Or,i,
[<Br,t]
[ftj'r.i]
Q>r,i.
= <"'r,i,
[w'r, t]
""'r,
(l"
= 1,
2,
..
p)
or, in words, if 2a)r,{, 2o)V,i be called the t-th pair of periods for the argument
Ur, the change effected by the substitution A^is an interchange of the first
When we
and
/;-th
of these
an interchange of the
first and second, (ii) a cyclical change whereby the second becomes the first, the third
becomes the second, ..., the^-th becomes the (/> l)-th, and the first becomes the ^-th.
Such a cyclical change is easily seen to be effected by the matrix
quantities can be effected by a composition of the two processes
(i)
E = (0
1
1
10
1
1
Other
sets of
be formed, can easily be chosen. One other obvious set consists of the matrices obtained by
interchanging the rows and columns of the matrices A^, B, C, D.
somewhat differently in accordance
+ We may state the meaning of the matrices A,, B, C,
iii.,
324.
333]
555
which
B we
obtain
\i=
so that this transformation has the effect of interchanging
[w'r.l]
=a>'r,l
<Br,l,
(r
= 1,
is
tOr
2,...,p
and
eo'r i,
introduced.
2, ...,p),
The matrix
I)
Thus
we have
{u;r\1) = A@{w;r'\^),
where
differs
Vr^
and
Uj.,
matrix
r,
and K, K'
differ
t^^s
Q and
Q'.
Thus
from t
differs
of the
first
is
and
equal
1.
In the case of the matrix (C), the equations of 326 (2) give
0(M;T|f) = ^0(;;T'||'),
where
save that t',,i=t,,,-1, and K'=Q',
u=w, t'=t
now the
is
equal to
g^nw(n+K')+inT' (n-^-K')^+m (n,+Q,')=+2iriJi:(n+Jr')-2w (Q,'- J)
_ g-tV (Q,"-
Q,')
g2mw {n+K'^+i^{n+Ky+2^K(n+K')
(m,
+ Q,')
g-w(Q.'2-Q,')
to
* See Krazer, Ann. d. Mat., Ser. n., t. xii. (1884). The number of elementary matrices is
by Burkhardt to be further reducible to 3, or, in case ^ = 2, to 2; Gdtting. Nachrichten,
stated
1890, p. 381.
556
326
(ii)
[333
lead to
@(;T|')=^0(;;T'|f'),
=r save
where u = w, r
save that
g2ir
^, =
+ Q^,
Qi
that
K2=
1,
=
= Tj^i 1.
now
we
have
Q2 + Qi
'''2,1
''1,2
t'i,2
gStir
We
now the
consider
(i)
the equation (a
+ ra!) t = ^+ tj8'
ri.l^
i/'Ti,],
'r i,r
or,
Hi,!'
-,.
K=Q
(r,
'"
''"i.r
"r
'"j,r/'''i,i,
= 2,
+171-1^
{n+K'y+imK {n+K')
equal to
is
which
falls
e-^mQ.'Qa'.
is
3,
i.r/''" 1,1,
also
Ti
transformation {E)
326.
Q,
sidered in
and K'
'Tr, 8
p),
Ty,s
that Wj
iti
),
T r,s
= t'i,
1 tti
Wy
= Mr t'i^^Mi, and
K'
q save
Z/ = - Q
that
we have
and
(w
T ^)
I
334.
= ^ (w
(i)
K-,
Q,'
the equation
g-v,, ^.^
Article
^ = Q save that
^').
t'
I
e2"''dw
or
1,
Jo
according as
(ii)
if
/S,
7,
S be real
quantities,
J-
where
for the
square root
is
to
a-i8'
is
positive -f-;
For indications
of another
a.
+ By the symbol Jii., where ^ is any constant quantity, is to be understood that square root
whose real part is positive, or, if the real part be zero, that square root whose imaginary
part
is positive.
335]
(lu)
u,
and
t, t'
557
Article,
Mn
= (wn),
+(t,,iW,
+Ti,pj,)w,,
w^n^
(iv)
the series representing the function (w, x') is uniformly convergent for all finite values of w^, ..., w^, and therefore, between
finite limits,
the integral of the function is the sum of the integrals of its terms.
case
(),
and
Ao= 2
M,= -ao
Jl,,...,
...
to
1,
we have
e'"'-i.i'i'+2'"(')i+2'"(T,, + ...+T,,,np)w,+iTO.2^y,
' J^
P'
Jo
BpJo
is
= 0,
?ij
...,
=0;
thus
l= ao J
rl
00
I= -oo J
J 00
ttit,
V
where the square root
is
to
is
negative
174),
"
-7rlTi,i
we have
Ti,i
is
positive.
Hence
V
and from
this equation,
gWr,, ,
,
by increasing
(
e-)
.^
Ti,i
by
/ J_
K+t'K
e2TfQ.Q,'
we deduce that
@ (w
t'
^').
I
|
Hence, when the decomposition of any linear transformation into transC, D is known, the value of the constant
A, can be determined.
But, save for an eighth root of unity, we can immediately specify the value in
for
when
C,
B to
have a
558
value which
Hence
is
M.
for a transformation
^\ = ...B'K..Al....D\..B'\..C<'...A'l
(",
if
[335
when
Q,
but
if
transformation
'
(viii)
(vrii)
we have M=...M^M-y...;
thus, for
any
'^"''<'<-'>=m<'[-''i'C)]'/\M\
where M=a+Ta', u = Mw, and is an eighth root of unity, r, r being as in
Putting 2a)U, 2vw for u, mi, as in 326, this equation is the same as
-^ 5 (
= ^=in=T
l""
'
2f
'
328, p. 544.
^ ('^ )1
is
of
some importance,
that,
c?)=(rj).
so that
[<i>r,i]
= w'r,i,
[<aV,i]=-<Or,<;
{r,i=l,2,...,p).
Then
We may suppose
Ttid' {a
+ Ta')w^=inru^=niuio= niT'u\
simple transformation
Bthus Apply
first
the transformation
where
'^2,2
/~^ /HI
V/
V r"3,,n,, V
r'2.2
where
|t| is
t.
y^'
336]
The
a/
\/-t ,
'
'"i. 1
x.
i.
''
oo
559
LOOPS.
we thus
find, for
integra-
the integral*
2,
oo
is
iiitan-'X.
-=1
where,
which
Xj, ...,
/V|rT|,
the matrix t be written p + ia-, |ttoI is the determinant of the matrix p^ + cr^,
equal to the square of the modulus of the determinant of the matrix
also
Xp are the (real) roots of the determinantal equation p X<r =0, aud tan"' X, lies
if
is
,-,
ii.
jr/2.
Of the fourth
for
according as a or
a' is
odd
'
as
where a
aS = aa
/
is positive,
^s'-^S = ^ff,
336.
Returning now
[]
/
/
(oa+
(o'a,
is
given by
-^<^+ia')^
a'{a + Ta')-
more particularly
to be taken.
and
Tta' -j- *
is
we prove
[ca']
first
= <ayS
in con-
to'yS',
-I-
where
a;9-y3a=0.
a'y9'
- /3'a' =
0,
ays'
- /3a' =
1,
can be effected by a change in the manner in which the period loops are
For this it is sufficient to prove that each of the four elementary
taken.
types of transformation, A^, B, C, D, from which, as we have seen, every
such transformation can be constructed, can
itself
be effected by a change in
The change
A^ can
clearly be effected
them
transformation by
Crelle, Lxxiv.
ELEMENTARY TRANSFORMATIONS.
560
differently
No
1.
1,
[336
further remark
therefore necessary in
is
can be effected, as in
w',., 1
['r, i]
=-
r,
[6,]
18
be altered
loop, will
(p.
(hi)
the change
is
shewn by
\1^
--<
Fio. 13.
to the substitution C,
which
given by
is
is to
This
may be
effected
(ttj).
[oj] as in
(cf 331).
Fiu. 14.
Wr, l] =
and
this is
effected
D the
<u'r,
2,
by drawing the
[w'r, 2]
loops
<o',.,
[oi],
Mr,
is
that given by
[a J, so that a circuit of
337]
OF PKRIOD LOOPS.
561
is
circuit of
Fig. 15.
this
diagram
(Fig. 15) is a
deformation of a loop
Fig. 16.
Abehan
integral m,
is
single-valued on the
[a^] gives
<a',, ^
- <r,
i >
to
i/' is
Riemann
surface,
self-destructive
and
as desired.
by a composition of the
any of the linear transformations of 326, that if the arguments there denoted by Ui, ..., Up he a set
of normal integrals of the first kind for the original system of period loops,
then
TF,,
...,
Wp
It is
immediately seen,
effected
for
new
transformation.
36
562
[337
...,mp.
Wli,
= 6)2+
[]
[a]
co'a,
w/S
first
when
that
a)'/3',
the places mi, ....Wp of 179, Chap. X., derived from any place m, upon
which the theory of the vanishing of the theta function depends, become
. .
.
nip
which
satisfy the
equations
iri,, [d (a'^')].
+ -+^n.p[d (a'^')]p,
(i=l,
wherein
Mj,
...,
first
kind
...,p),
for
the
For
let Wi,
girw' (a+ia)
vfl
first
ic^(a'y8')"
= A,@{w;
t'),
id(ay8).
put
..., nip'
when x
is
at any one of
= M*' u"""^'
...
u'ip'^^'
M*'
M^"'"*! ...
d{a'^)-]
u'T""^
\h
vanishes
when x
is
d{afi)]
...,
Xp-,
therefore,
y,f-
[d (;8^')]i
may be
by a
+ WiA [d (Sa')]. +
. . .
+ ^r'i, p [d (a')]p.
( = 1, 2
p).
338.
From the investigation of 329 it follows, by interchanging the
rows and columns of the matrix of transformation, that a linear trans-
339]
563
for
. .
mp, which
arise
by taking the
v""''-
"+
characteristic
M"'j>'
-^n^,
'^
(
in the equivalence
^.
be in turn all the even characteristics, can arise for the places m/,
vip.
In particular, if |n^_ be an even half-period for which (^0^, ^) vanishes,
we may obtain for m/,
nip' a set consisting of the place vi and p1
places Wj', ..., n'p_i, in which rh, ..., n'p^i are one set of a co-residual lot of
sets of places in each of which a </)-polynomial vanishes to the second order
(cf. Chap. X., 185).
to
. .
i^-
x. If in the hyperelliptic case, with p = S, the period loops be altered from those
adopted in Chap. XI., in a manner equivalent to the linear transformation given in the
Example of 329, the fimction e {w t'), defined by means of the new loops, will vanish
for w = 0; and the places m,', m^', m^', arising from the place a ( 203, Chap. XI.), as
TOj, ..., mp arise from m in 179, Chap. X., will oonsistt of the place a itself and two
;
We
839.
i.
be
when the
taking
make
<^
VZ"*
in place of
V P^.
VF"'
It will
244) by
244, slightly, to
uniform with that of this chapter, the results there obtained are as
follows the form X'" is a polynomial of the third degree in the fundamental
it
zeros of a ^-polynomial
<f>;
..., J-ap-s,
v.'^'-'"^
where
Ui,
Mi, ,
Up are the
is
if
f^p',
...+qp'Ti,p),
Riemann normal
(i
1, 2, ..., p),
integrals of the
is
first
that denoted by
^f
kind, the
jl
and+
defined by the fact that the function VF^/VJi^, which is single!)*', (- 1)*
valued on the dissected Riemann surface, takes the factors (kind.
respectively at the i-th. period loops of the first and second
it
may be
m/,
...,
Or
lot of sets,
let
remark of
m/
be the places
solution,
185.
being omitted.
X Integer characteristics
362
564
[339
which, for these loops, arise as 771,, ..., m^ arise for the original set of period
character
loops; let Zi be the form which, for the new loops, has the same
^<^i
second
to
the
vanishes
that
so
as has the form Z"i for the original loops,
order in each of
(
V,
A^p-s,
...,
.4i,
the equivalences
337)
^-.+
...
+;f''-'>>H H^(/8/3')]i
+ iT'i,,[d(aa')]i+
+|T'i,p[d(aa')],,
(i
= l,
...,p),
...,
The equations
+ oj'a',
Qa
equivalent to
e""
'''
'^^'' *'.
of transformation,
[ft)]
(ia') ]r
factors
(cf.
0,,^
is
[o)']
a)/9
that given
0)'^',
by
(&{)
new
and
loop
a'i_y
[6,] is
positive
where
Ir
is
an integer which
where
I,
I'
now the
VT^/VX'"
aq'
loops, are
aq'
a'q,
[-
qi
at the
Oi, r
+ qt
new
-l = ^q'-
(mod.
^'q,
e"^*", e"^,
2).
where
(mod. 2)
may be
vT^/VX'"
of loops
(mod. 2)
a'i. r],
given by
= 2
Ir'
k'
given by
new period
is
is
VF'"
is
for
at those
therefore i (,
the
equations just obtained prove therefore that the characteristic associated with
VF
is
charaxiteristic.
OF RADICAL FUNCTIONS.
^^^l
The same
result
may
be obtained thus
the
p equations
565
of the form
?"' + ...+'''"'"'> =
are immediately seen,
expressible
by means
- r'a') = 1
to lead to
'
by
p
r
equations
<'''"'
+ ---+y'T'''''"'^Hd{m')]i+ir'i.dcl{Sa% + ...+iA,,[d{aa')-],
(i=l,
VW),
...,p),
for the
new
immediately deducible.
k'
-k = Pq' - j^'q,
aq' -ctq,
(mod.
as a period-characteristic, as in
2),
328.
340.
Having thus investigated the dependence of the characteristics
assigned to radical forms upon the method of dissection of the Riemann
surface, it is proper to explain,
may be
somewhat
further,
how
these characteristics
The
case of a form
When
the zeros of a
is
process
is
values of Vj> (
Riemann
But
it is
surface;
elliptic functions,
'Jf{u + 2m^-eil'Jf{u)-ei
V^ (it)
e^
the
the
may be
VF<'>/VX'" at the period loops, because the form VX'^' depends upon the
places 7i, ..., mp, and we have given no elementary method of determining
these places the geometrical interpretation of these places which is given in
;
183 (Chap.
distinguish
X.),
the distinction in
fact, as follows
from
338, cannot be
made
algebraically
Nevertheless we
566
m-,,
may determine
F'^',
[SiO
...,inp
It is easily proved,
if
VZ
there be a form
and conversely,
if
g'
a linear aggregate o{ q
characteristic as VF'^'
in general, since a
3(p
1) linearly
\/F'''
= ^^',
radical
independent terms
form
all
where \/F"'
is
SP^*'
( 111,
V F''* is S{p 1), we have 2 + 1=0; in any case the value of g'
capable of an algebraic determination, being the number of forms
zeros of
is
4>'^'
places.
Now
the
odd
( 185,
number
is
of linearly independent
(^-polynomial
of
/ii,
Ml'.
..., fip,
>
function
Ai,
...,
factors
be (-
1 )*>',
l)*-"
first
then
respectively;
if
kr'
Take now,
Qr
(mod.
the possible
2^
m;
2).
where the
refereiKse is to
same form
all
cf.
iii.
(1894), p.
494
341]
by the change
characteristic,
is
GIVEN.
56t
even when VF/'* has an even characteristic and odd when VF,'" has an odd
characteristic
satisfies
for, clearly,
the characteristic ^ (^
if
i[]
a value for
is
which
for
we should have
(k
<r) (k'
+ <7') =
(k
+ q)(k + q')
(mod.
2),
or
k {a
for all
(k'\
(
and
qq'
- acr'
_ _
295).
following rule
form VF,'"
all 2'^
forms
is
even or odd
1 )*',
(-
1 )*,
+ k'\
is,
form VF'"
of
'JY^^'^J^IY^^'^
any
this being
is
even or
determined
the characteristic
loops of the
^y(+i)
341.
is
VF;
. ,
in
(p.
380).
568
[341
these formulae when the period loops are taken otherwise than in
Chap. XI., follows immediately from the results of this chapter. If the
change in the period loops be that leading to the linear transformation
which is associated with the Abelian matrix formed with the integer
of
matrices
a,
^, we have
a',
yS,
324)
where
k'
aq'
-a'q-d {So.'),
-k = fiq'-^'q-d (/9/3').
have
then
M*' "
giQJr.i
W^'"
liVr,i
+ qpfOr.p + g'l'w'r.i +
.-
+ qp
<>>'r,p,
we have
+lpVr,p+li'v'r,i+
...
where
I'
aq' -S.'q
= k'+d (fia'),
^1
is
(^0), d
a constant multiple of ^,
be denoted by
(aa'))
k'\
w; I
+ lp'Vr,p,
,
j
+ ;x
fi,
the function
and we may
/i].
208
is
equivalent to
where
is
= <'" +
..
+ <'-r
(r=
where
Tn/,
. .
.
is
...,;,).
209, 210.
It follows
nii, ...,m.p
of
fj,
of period loops,
It has already
Ajc,
B,
the
is zero.
342.
from
new system
179, Chap.
characteristic
206,
/u,
1,2,
= 4 (a; Ci) {x ttj) (x Cs) (a; ttj) ...( Cp) (x Up) {x c),
342]
so that
annexed
figure
and
6jf_i
6aj,
is
569
and interchange
c,
(oi), (bi),
as in the
(bi),
(Fig. 17).
Fig. 17.
Then
if
U''
W' "
...,
",
an integral polynomial in
^r is
kind,
first
of degree
x,
jjt,.. 5...
^ ^^^_ j_
jji^.,, *.,
if
now
^^,^
J_
w^'
",
f/',= [2w,.,i]w,
. .
.
= Lr,i [2Q>'i,i] +
1, 2,
;,
. .
.
so that [<V,t]
[<o\,
. .
w"' ",
+ Zr, p [2&)'p,
(r, s
>]'
1, 2,
p),
@ (w
is
purely
also
is
= 0),
Wr = Lr^iUi+ ...+Lr,pUp,
...-|-[2tD,,p]M;j,,
(i
|-^,^ ._^-,_
imaginary
given
t'),
335) by
where K, K' are obtainable from Q, Q' respectively by reversing the order
in which
of the p elements, and A is the constant Vz'/Ai ViAj/Aj \fi^^/A3
=
find
immediately
that
=
etc.
We
A T,,i, As Ti,iT2,2 r\2,
. .
f^fc<-.,
(i
= 0,
= _ [^^_ _
.]
1, ..., ja),
[;a,^_
^j
+ [;
.]
= _ [,.
tr^.
.^j
ftjt-i, 621
.
.
_ i;^^
^j
+ [-;_. ]^
even characteristics
,.,
/0...0 1
ON
0...
./Oy-wi\f
{2^}=no...oo-i...-i)=noj
*
The
by tt'= -
quantities
1)
tj,
of magnitude.
when
c,
Oj,
...,
o\
l)
(o)(-i>
ia
'
and
t'j.
of 342
in ascending order
WEIEKSTRASS'S RULE.
570
[342
and may denote the theta functions with these characteristics respectively by
@ji_i(w; t'), @jt(w; t'); if 6*, hi, 6^, .... be any of the places h^, ..., 6, not
more than p in number, and if, with > g, < 2,
> 5/ < 2, we have
[r**.
TJl'--^
= -g, [wr,,] -
...
...
is
"
[
may
be denoted by
k,i,m...{w,T').
This function
characteristic i
We
equal
is
to,
for the
2^
is
of,
even or odd.
half-integer characteristics*,
For the numerical definition of the half-periods, which are given by the rule
at the bottom of p. 297, precise conventions are necessary as to the allocation
of the signs of the single valued functions \/x
(cf.
br
Chap. XXII.).
p = 2,
table of 204 are supposed to consist of positive elements less than unity
Qi!
Qi ^^^ 6ach
."^-^
e [.
|.'
|>^.-^-' e [.
the
when
Q^, Q^i
either
.'
4;, 4.,]
number notations for the transformed characteristics are then immediately given by
The result is that the numbers
204.
the table of
1,
14, 4, 34,
03
by
* For convenience in the comparison of results in the analytical theory of theta functions,
it
it
344]
CHAPTER
On
343.
which
it
XIX.
The
desirable to have in
is
mind
is
it
possible to build
as
up functions
for instance
344.
If
variables Mj,
is
an
...,
unaltered by
uniform
analytical
function
of
independent complex
...,
m^,
lij,
of the
...,
Mp,
it
constants
Pp, then Pi, ..., Pp are said to constitute a period column for the
Such a cohimn will be denoted by a single letter, P, and P^ will
function.
denote any one of Pi, ...,Pp. It is clear that if each of P, Q, P, ... be
Pi,
...,
period columns for the function, and \, /i, v, ... be any definite integers,
is
independent of k, then the column of quantities XPk + fiQii + vRk+
this
column
by
denote
we
shall
function;
the
for
also a period column
\P + /iQ + vR+
P, Q, R,
...,
...,
the coefficients \,
fi,
it
is
v,
...,
572
The
More
imaginary parts.
[344
when
generally,
the
fi,,v, ...,
it
. .
clear
is
known
it is
that
or more,
1,
is
is
a linear
not zero.
positions in the
column
equations in
all,
with
real coefficients
can always be
for,
in order to such
satisfied.
m'
= at^+i(a{,
are connected
by an
equation
tQ.-\-xa+y(o=Q,
in
which however,
Thus,
for
if <jo)2'-<i>2<ii'i' = 0,
also
= 0.
exists
and 2p
a least number,
r,
of period
1,
equations
Qr=x,p'''+
+x,pr,
{k=\,%...,p),
wherein \, ...,
are independent of k, and are real and not infinite.
It is
purpose of what-\ follows to shew, in the ca^e of an uniform atialytical
function of the independent complex variables ih, .-..Up, (I.) that unless the
function can be expressed in terms of less than p variables which are linear
the
Two lemmas
(a)
column
If
are
coefficients.
of infinitesimal periods,
,,..., Wp
have a
t These
it is
;
it is the number r which is unique
Abhandlungen aus der Functionenlehre
is infinite
HAVE INFINITESIMAL
345]
than
PERIODS.
be expressible as a function of
573
Wi, ...,
Up, it has
less
columns of
infinitesimal periods.
Of periods
(/&)
Wi, ...,Mp,
there
is
To prove the
345.
when
first
lemma
part of
...,
...,
Up, then
it
periods.
We
Ui, ..., Up', such that, for absolute values of the differences
which
/(mi,
itj
u^,
...,Up
Up
function,
Then
Ml,
if
Up, there
...,
Ci,
linear functions of
...,
Cp
function
<.:*
du^'
vanishes for
require
to
Hence there
all
8/
'"
+ ^*&M,
^--V
'
be a function of the
p\
variables
Ci;ui
for this
c^Ui (i = 2,
would
. .
. ,
p).
exist sets of ordinary values such that not all the differential
respectively of
',",
. . . ,
v}^,
iti, ...,
!'',
...,Mp'
be such an ordinary
all
the partial
Then, similarly, the determinants of two rows and columns formed from
the array
df
du^r
df
df_
INFINITESIMAL PERIODS.
574
i, ...,
[345
...,
m^
the
Proceeding step by step in the way thus indicated we infer that there exist
sets of ordinary values of the variables, (v^
...,Up),
...,
{u'
...,!//),
such
element of
A:-th
from zero
different
ty
Vi
...,
Vp
if (for
where
v^p
r=l,
p) the set ul
be replaced
Up
...,
is
(r)
is
infinitesimal periods.
=/(';'
= ^
where
d^,
H,^
[u':'
+eiHi,...,
+ 0m H^
for mJ^
these
we
is
...,
Up
are (for
p equations, by
. .
p),
:^1,
m=
1,
and the
.., p) to
A by slight
we have seen above that such a determinant
+ e^pi
0p are quantities
...,
wlr'
(r
...,<),
is
obtainable from
not zero.
wherein
m< p.
Uj, ...,Up
is
= ct^,jih+
The
of
...,
by
ak,pUp,
conditions
taking Qi,
lemma
expressible
that
(a)
we may proceed
arguments
(k=l,
...,
+a*,pQp
= 0,
Mp
as follows.
Vm given by
m),
Vi,...,Vm remain
% + *,,
by
...,
Vi, ...,
+ Qp
unaltered
when
are satisfied by
Qp so that
a*.iQi+
{h=\
m),
and since m,<p these conditions can be satisfied by finite values of Q,, ...
which are not all zero. The additions of the quantities iQi, ..., <Q
not altering
Ml, ...,Up,
Hence by supposing
to
Dm. will not alter the value of the function/.
taken infinitesimally small, the function has a column
i;,,
...,
of infinitesimal periods.
lemma
P* = p*
As
periods, (^
to
1,
. .
jo),
(/3), let
wherein
347]
is
aa assignable
is
/*i6
2p quantities
be
specified
vt
positive integers,
2p
at most one column of periods satisfying the conditions
than
there
575
Pt l< (m*
e.
1)
Then
6,
if
"ite
fi^.,
:1'
o-i
I
<
(i-fc
1)
{k=l,
6,
...,p);
wherein \p,i\, \a-^\ are the numerical values of p^, o-j,; for if
pt' + tVi' were
one period of another column also satisfying these conditions, the quantities
Pk Pk + i (o^k o'k) would form a period column wherein every one of the
2p quantities
Hence,
finite
pic'
pk,
since, if
number
v^
/i.^,
is
Pk, <rt is
347.
numerically
We
return
less
than
now
And
g.
this is the
it
/by
only a
io-*,
is
most general
Q = X,Pi" +
here the suffix
is
Consider,
<
and
is
is
Xi
first,
:^ 1.
omitted, and
+X.P"-;
linear combinations of ,,
Q from which
X^
...,
Up.
'\^=
...
\,.
is
is
only a finite
number
an existing period
of
which
and Q'''
for
X,P
+XmP'"",
{initr),
wherein
Oi^X^lf-l,
:!
x_,
:f 1.
0<x:|'l.
Since there are no infinitesimal periods, there is a lower limit to the values
of \m corresponding to existing period columns satisfying these conditions;
since there is only a finite number of period columns of wholly finite periods,
there
is
column Q
for
which Xm
is
576
X by
there
P""
may
X.,,,, ...,
Xw-i,m values
by
Q'' given
exist
[347
which the
in
coefficient of
is X,,.
m = 1,
Thus, taking
we
..., r,
2,
wherein N^,
Nr
...,
...
iV^Qi'-)
+ ^,pw +
/x.i,
...
/^r
and
less
than
X,_,.,
where Nr
we have
A'r.
...
an integer and
is
Q = \,Pw + ...+VP<'-
= VP<" + ... + W-iPi--' +
Q'""'.
which
in
/i^pw,
i,
...,
V^^.
fir, if
iV.Q"-'
not zero,
positive
is
+MrP"'',
where
Xi
= Xi JS r^-i,r,
Q'
= Xi'P"* +
X r\
...
= Xy_i iVyX,_i
,.
X'^-jP"'"''
q = X/'Pi" +
and
so on.
But now,
it
. . .
if iV^iQw
follows, as iVi,
...,
Nr
...
+ iV^Q""' +MiP'" +
...
+/i.,P"-'
be a period column,
+ ... + firP'^^
is
a period
column; and this in fact is only possible when each of /*,, ..., fir is zero.
For, by our definition of Q""', the coefficient fir is zero then, by the definition
of Q"""", the coefficient fir-i is zero and so on.
On the whole we have the proposition (II., 344) if
;
Q=X,,^Pw + ...+X,.Pw
be that real linear combination of the first
coefficient
Xm,
m has
(m=l.
...,r),
...,
PM
in
and not greater than unity, or be one such combination for which \n,m satisfies
same condition, then every period column is expressible as a linear combina-
the
tion of the
columns
with integral
coefficients.
(i),
..., *) are not connected by any linear equation
with
same would be true of Pl^), ..., P*""). And it should be borne
mind that the expression of any period column by means of integral coefficients
It
in
in terms of
348]
577
.^^*^^'
^mbZ"'
Omn^n'
y"' in'-L^r
V
the form
^' ^^P'"''"^
the period
terms of
.,
'.
coefficients,
any
coefficients is a linear
^^^^^^
combination of p,
...,
pw
coemcients.
...,
with rational-number
....
this property.
We
348.
consider briefly the application of the
foregoing theory to the case of uniform
analytical functions of a single variable which
do not possess any infinitesimal periods. It
will be sufScent to take the case when the function
has two periods which have not a real
ratio
this is equivalent to excluding singly periodic
functions.
If 2a,i, 2a,2 be two periods of the function, whose
;
period,
It IS
possible to find
two
real quantities
^^ such that
Q=X,<Bi+X2Ci)2;
0,=;iiG),.
there is a finite
fi
Xi
- iVjX = iVj/i, +
'i</'i. '2</^ is
+ 2
1-,
is
Q = 2iViQ,+ 2^^202,
where
iT,, TTj
are integers.
Cf. Forsyth,
B.
37
578
by
[348
2a>i, 2<2
coefficients.
Ex.
349.
If r quantities
i.
If -P<i),
ii.
a column
NiP^'^)
constituents
Chap. IX.,
Ex. iii.
. . . ,
are
166,
...
within
and Clebsch
r\
u.
p.
(cf.
135).
An
B"), each of
...,
The
result
is itself
text.
Ex.
iv.
Riemann
p'r,i,
If Wj'",
surface,
+ i(T'r,t,
...,
iC'''^
and the
first
kind, on a
shew that the vanishing of the determinant of 2p rows and columns which
is
denoted by
Pr.lJ -J Pr.pi
Pr.lj
"M
Pr,p
<^r,
1 )
*^r,
p)
...
-I-
*'"
r, 1
""
r,
'
Jfi
ii\^iX' " +
(ifp
- tiVp) M^'
"
= constant,
where M^, N-^, ..., M^, Np are the minors of the elements of the first column of this
determinant and are supposed not all zero.
The vanishing of this determinant is the condition that the period columns of the
integrals should be connected by a homogeneous linear relation with real coefficient's.
350.
The argument of the text has important bearings on the theory of the Inversion
Problem discussed in Chap. IX. The functions by which the solution of that problem is
expressed have 2j columns of periods in terms of which all other period columns can be
expressed linearly with integral coefficients these 2p columns are not connected by any
linear equation with integral coefficients ( 165), and, therefore, are not connected by any
;
remarked
( 174,
351]
579
whereby
it appears that the most general 2^-fold periodic functions that are possible can
be supposed to arise in the solution of a generalised Inversion Problem ; this Inversion
Problem
differs
functions*;
351.
t.
We
necessary to
viii.
The reader
(Berlin,
1886),
p.
is
referred to Weierstrass,
177,
(1886).
pass
make
now
clear
is
diflFerential
coefficients of the
all
must
necessarily hold
among the
periods,
functions.
[351
NECESSARY LIMITATIONS
580
is
periods, 2a)^K
consequence of the two suppositions is that in the matrix of ocolumns and 2p rows, of which the (2i - l)th and 2i-th rows are formed
respectively by the real and imaginary parts of the row a^", ...,aY\ not
every determinant of
variables
real
iCj,
...,,
we form 2p
For
with
cr
arbitrary
l)th
vanish.
if
and
that
all
cr
And
is
is zero.
case
352.
(I.).
We
uniform function,
it is
<^
(m) is
an
rows and
a-
columns which occur in the scheme of 351, that the matrix of a rows and
columns^, expressed by
ab ba,
(A),
should be a skew symmetrical one of which each element
Chap.
When
conditions
is
rational
IX., 166.
<r
= 2p,
(of.
the hypothesis of
no
345).
X The notation already used for square matrices can be extended to rectangular matrices.
ii., at the end of this volume ( 406).
353]
581
integer.
than
ax
of the elements of
a;
(B),
We
shew that
a condition only
fulfilled
by putting each
equations
= 0,
when (r<p +
l.
353.
Before giving the proof it may be well to illustrate these results by shewing that
they hold for the particular case of the theta functions for which (cf. 284, Chap. XV.)
fr
= 2p,
a=|2,
27n'6
2<'|,
217, 2ij'
and therefore
JiTTX
where JT
is
y=
T and
Y'
is
Let
then*
\T'\
by = {db-ba)xy=kxy.
but
i,J
Jcxy=XY'-X'Y=(xy,
namely, the matrix k is in the case of the theta functions the matrix
columns, which has already been employed (Chap. XVIII., 322).
.-,
of
2/>
rows and
Next
if a, b,
hQx=7riX+hX',
Hence the equations ax=0
namely h.ax=iriX+hX'.
give
X=--.bA"'-
If X^, X-l
Xi= ^bjZi'.
Hence ikxx^ = Uxx^ = i{XX{ - X'Z,) =-\ [bX'Z/ + b, Z^'Z'] = - ^ (b +b,) X'X{, since
b=b
and
b,
= bi.
be negative
is
Thus
if
is zero,
t Chap.
582
354.
we
[354
352,
(I.),
'''
+ i"'] + 27r.ca)^(^
^ ^,)
Zi2
equation
= 7ri [feWa^ -
=-
6<2)a<"],
hence L^Jiri
is
a rational integer,
L^i.
symmetrical in regard to
is
= k^
Oj
say,
and
a^,
e^"
such that
k^^
= ^a-
Obviously, in
ing the
first
the equation
<^
(m + ai"
ii^i
?wi
[W" !,
-I-
a'-'
ms+
+ a^ m^)/<j>
...
(u)
. . .
+ d-'l jk^) +
ttiI,
where
X=
ka^m^m^,
a=l
P = 2,
and
TJZi,
...,
m
,
<^
In
fact,
are integers
(m
.,
2fft6m
....
cr
this equation
+ am) = <f){u)e
,
cr
may be
represented* by
rU + <"! +2in'n
/?)!
+ Tt."S^,
S *-=
-Sp
p
L
2 J
.
(II.) to
m,, we
iaft)]
iirib"^*
...,
(I.),
see
Appendix
lim
11.
or
+ bi^m] u^
= (Sftiittj)
77!,
+ (Sfci,U;) 77V
+ {Sb'/kf) 771,
= (Zftl^Uj)
777,
= !)'"u.r(!i+
+6''''u.7,
= 6'''77li.+
H-i'^'TTt^.U.
S k^m^m.
k^ntamp + JriR
thus
u = S [buMi +
iri
j,
/^\
/\
OF THESE LIMITATIONS.
355]
where
R is equal
lb<^\af>vi,
...
to 6<"
+ al'^m,] - S
[6]'
V+
...
583
to
so that
. . .
. . .
A;,2
. . .
. .
. . .
. . .
hence
wiR + iri 2 k^m^nig
ri
1,
nh,
J.a'oi'"'fl'
where
m/] =
[wi'
[mi
...,
m]
therefore
<I>[U
^i^^^k^m^'mp'^
+ am] = e^^ibm'iu + iam'HiHcm' +
^^^
Similarly
(I.)
<t,(v-
where v
ao-))
+ am
a*''),
u + a'JK
Zoo.
character,
all
values of
the variable.
wherein
fi, ...,
and
and
m^
between
...,
v is finite,
let
is
Now when
unity.
L (^)
is finite.
(I.),
and
is
f, are each
Its modulus is therefore
^i, ...,
f
can be determined by considering values of ^ between
and 1. Let now
Xj, ...,x be such that aa; = 0, and let a;, denote the column of quantities
which are the conjugate complexes of the elements of the column x. Put
;
= a; +
aji,
so that af
= omJi.
Then
<f>(v
+ aa;^) =
(j)(v
+ a^) = e^i-f
f+2"(+')
may be
is
L (^),
a positive quantity
G whose
other solution of
of
[355
584
i (f) when
cm;
it is
and
between
is
1.
We have
= hx
(c
say,
w=c +
where
6v
axi
= Icxx^ + ahx^,
therefore
giri6. of +2 (c+M (
same
J[j(^)
(^
(d
= girrkxx,+mabx,''+'in iw-w,) x,
(m+w,a:,) Z,
garri
(^)
as
aXi)
e'K,
where
has the same modulus as
and where
(f ), less than G,
= iTrkxXi
=
- yiZj) = 2irkyz,
iirlikij (yjZi
is
+ izi,
liirlikiji
yi-izi
iz).
Now
solution,
if
fj,
K having also
modulus <
its
Herein the
left
side, if
G.
not independent of
is,
/i,
of
Hence p must be
/A.
the real
that
values of
Hence
fi.
positive, viz.,
quantity ikxx^
/t which is
must be infinite for infinite values
valiies of x such that ax=0 are such
it
necessarily positive
is
pression
is
not independent of
Now
if this
obtained
when
p..
expression be independent of
p.
= 0, and
/i,
it is
equal to
^ (v),
the value
therefore
g-inu.-'abx,' "^
(^
p.aX i)
_ ^^^ ^j^_^_j ^^
<\,{v)
here the
is
(i)
zero
its
left side
value
Suppose
is
is
unity
first cue, is
not zero,
cmc,
FURTHER DEDUCTIONS.
355]
585
then
Wi)
(w
must, like the
left side,
a?!
Vy
^j.(5i;+(7), so that
^{v +
hence
iMOXi)
In
this case
A = iTrahx^^
where
column of periods*
fiaxi represents a
(v) e^'"+-B/x+Q*^
<f)
<^
{v)
and
this
fi.
however
{v)
<^
Hence p must be
is
zero
and
\<h
Hence
ikxxi
ax
satisfying
is
= 0;
of x, other than
and
this is the
theorem
zero,
(B).
From
Remark
i.
which ah ha
of
a-
is
vanish
and
by rational integers
columns
o-
of qiiantities
x^^\ ...,
For then,
:c<<').
equations
also,
o-
...-(-aio-^i')=0 satisfied
Remark ii. In the matrix i, if <r be not less than p, all determinants of 2 (o- -p) rows
and columns cannot be zero. In the matrix a, not all determinants of \(t or ^(<j--|-l) rows
and columns can be zero. In particular when <T=2p, for the matrix k, the determinant is
not zero for the matrix a, not all determinants of ^ rows and columns can be zero.
Let I,
be columns each of o- quantities. Then the coexistence of the 3 sets of
;
ij
equations
af=0,
is
i(|+,)=0
a^r,=0,
( 352),
The second
-^
and therefore
ai)i
= 0,
*
ij.
of
also
/tijiij
(f
-7)
"Ji
6ijj
6|
aiji is zero,
and
also
But by condition
is zero.
kri^rf
when, as here,
We use the
word period
(I.).
586
[355
Similarly
zero
simultaneously
^, ij, the chosen independent functions could be made to vanish
than zero values of the variables, and then aU the linear functions dependent on
of variables
for other
must
Hence
these
also vanish.
2v
+ 2k>2(t
or
+ K>a:
Now
1/
5 p,
2ic
(T
hence
= itr,
2k
= 2 (o- -p).
Remark iii. It follows from (ii) that if k=0, then v = a- and <T=p. Also that a function
oip variables which is everywhere finite, continuous and one- valued for finite values of the
and has no infinitesimal periods cannot be properly periodic (without exponential
more than p columns of independent periods in every set of n independent
periods of such a function the determinants of a rows and columns are not all zero. The
variables
factors) for
proof
is left
Remark
to the reader.
iv.
When o-=2p we
is
zero.
2(a'|,
When we
a=|
2<i),
2w'
we
shall
2< is
always
356.
Ex. i. Prove that the exponential of any quadric function of j, ..., Mp is a
Jacobian function of the kind here considered, for which the matrix k is zero.
Ex. ii. Prove that the product of any two or more Jacobian functions, 0, with the
same number of variables and the same value for <r, is a function of the same character,
and that the matrix k of the product is the sum of the matrices k of the separate factors.
Ex. iii. If <j) be considered as a function of other variables v than u, obtained from
them by linear equations of the form u = fi + cv (ji being any column otp quantities, and c
any matrix of p rows and columns), prove that the matrix k of the function 0, regarded
as a function of
v, is
unaltered.
a,
b,
(Cf.
Ex.
i.,
190,
Chap. X.)
Ex.
iv.
a we use
a'
= ag,
where ^
is
rows and columns, prove that <^{u + a'm) is of the form e2'rf*'"'(''+Ja'm)-l-2'*m0(^)^ a^j
that V=gkg ; and also that kxy becomes changed to Vjiyi by the linear equations x=g3f,
y 9y'- 1 such case the form k!x!i/ is said to be contained in kxy. When the relation is
reciprocal, or ^^=1, the forms are said to be equivalent.
Thus to any function ^ there
corresponds a class of equivalent forms k. (Cf. Chap. XVIII., 324, Ex. i.)
Examples iii. and iv. contain an important result which may briefly be summarised by
a-
p. 242.
is
a.
known
356]
587
no theory of transfor-
is
A transformed
( 324) are
theta function
is
in order that, for this Jacobian function, the matrix h should be the matrix
or
e,
353).
(of.
first
AcA^k.
In fact
if
^ = Ax,
^ = Ax\ then
hx'x = ax bx' ax" .bx = S
.
= (Ax
Hence
when
also
which in that
tr
= 2/>
Ax' = AeA
the determinant of
- ij' ij + p] = f ^|'
[^^ |'i + p
x'x.
is
case, being a
order,
is
/,
a perfect
square.
Ex.
Shew
vi.
when
that
o-
= 2p and
2o)'|,
27ri6=|2i;,
2i)'|,
that
AiA = .
TTl
mrj
TJ'ta^
ia'rj'
^'o>'
(Chap. VII.,
Ex.
of Ex.
v.
2/o
name
two
( 140,
difiFerent
and
p.
Riemann.
We
can give a
corresponding transformation of the equations (A), (B) ( 352) in this case, provided
the determinant of the matrix k not being zero.
As
to the equation
(A), writing
it
in the equivalent
immediately deduce
is
As
= 2jo,
v.,
we
(A'),
the column
satisfied.
(t
_
Ak--^A=(,
which
in the
140).
vii.
Thus
z,
of
<r
= 2p
.r
be a column of
o-
'2p
= 2^
arbitrary quantities,
and determine
Then
dx=dbz={abba)z = kz,
ji.,
thus
ikzzj
ik~^
therefore, the
liilt
= ik~''-ajXi
Ax=iak~^diXiX
bz
form
iak
is
column
x,
^d^XjX
(B'),
This
of equations (B).
Ex.
viii.
When a = \2io,
AeA =
2to\, 6
= ^.|2i;,
2i;'|,
cr=2^,
we have
is
588
[356
Hence when k=, the equation (A') of Ex. vii., equivalent to AeA= f, expresses the
Riemann equations for the periods (Chap. VII., 140). In the same case the equation
(B'), of
Ex.
vii.,
expresses that
^Vi^=
is
negative for
Ex.
all
values of
When^ = l,
ix.
iejjA'j
ft,
^ (a,)
x^ {Xi\
zero.
2('|, if
-a
^a^
[(a,)
X=l
x other than
= positive
v=\
for
ax=0,
(B'),
or
arbitrary x,
x and
x',
and
\i
(eacoi)"' (o)Q)i'
ap + ia;
a>i= p
cDOJi)
J;'.^'l'
= positive,
357.
i (<bj(u'
p'o->0,
o-
= 2p we
;
Two
proceed
arithmetical results,
utilised,
kj,
-k,,
(i)
357]
Ex.
(cf.
ii.
If
fi
3hap. XVIII.),
XVII
322, Chap.
we have,
if
h=geg, also
2jo
589
k=gii~^eii.-'^g,
fiT^g.
(ii)
If 5^ be a given matrix of integers, of 2p rows and columns, and x be
a column of 2p elements, the conditions, for x, that the 1p elements gx
should be prescribed integers cannot always be satisfied, however the elements
unity, is finite.
g.
is
9ii ffu
there are
than unity.
5rf22
- gr^g^
integer pairs
+ g^^x^,
when g =
1 2
590
variation of z
tity,
of
=M
jj.,&sM changes,
M,
[357
its values,
is
''
...{Ndz,)...{Ndz,),
gx = z,a, value
Since this
358.
is
equal to tN",
it
all
Now
than unity.
less
this integral is
follows that
is
g,
equal to
|^|, as
was
stated.
when a = 2p. The determinant of k not being zero, the determinant of g is not zero.
Put K=ag~^, so that .ff" is a matrix of p rows and 2p columns, and
a = Kg put similarly b = Lg; also, take a row of 2p quantities denoted by
G, such that c = gC + ^[g], where c is the parameter ( 351) of the Jacobian
function, and [g] is a row of 2p quantities of which one element is
[5']=
c
take
X, x',
X = gx,
(a=l,
g,,agp+,,a,
X'=gx',
so that
ax = Kgx
= KX,
bx
= LX,
then as
kx'x,
is
...,2p);
=l
= ax.bx' ax'
ax'
= KX',
bx
= LX';
bx,
also equal to
gegx'x = egx' gx
.
we have
= eX'X,
KL-LK=e.
(C),
so that
_
_
1
KxLx' Kx'Lx = (KL LK) x'x = exx = 2
^
'.
further, as ikxx^
is
positive for
ieXXi =
thus,
if
= 0, we
positive
K\
^
I
ax
if
^ be a
(a;.;a'i+p
Xj
Xj^p)
have
when
KX = 0,
AeA^-AeA =
and,
(D)
(C),
(E),
X be a row of
2p quantities
358]
KX=0, LX = z,
such that
eKz = - X,
therefore
so that
= eXi, we
K^z^
591
and
have
iK^eKzZi
= positive,
iK^eKzzi
than
(F)
zero,
for
it
Remark
iv. (
\2r], 27;'|,
of the matrix to is not zero, then the equation (E) shews (cf Ex. viii., 356)
= r)a)~''o)' ^7rwi)~^, so that
that the matrices a, b are symmetrical, and that
17'
we can
also write
7)
also,
7)'
2a(ii),
2a,a)'
2hco
h',
= Tri,
2hto'
=b
by actual expansion,
iK^eK =
Q)'aJ~'] w
4iQ)i [&)j""'a)i'
Wj [bi
+ b] oj =
(Ui
TT
WjCft),
if
b=
+ id
[bi
b]
55
TT
TT
thus
iKieKzZi
ctit,
where
= wz, z
and
TT
and
by the equation (F), for real values ot n^, ..., np other than
form bw'' has its real part essentially negative.
we
define
Hence
can
a theta function by the equation
therefore,
wherein
Gpj^
7, 7' are
= 7,,
for
'
-yJ
rows o{
r<p+\.
quantities given
Denoting
this function
is
immediately seen
^e^ =
we
XVIII.).
322, Chap.
592
359.
where
= ni,
Vi
aN = KgN,
Vi+p
let
<f),
= ni,
(j)
(m
v)
i<p + l.
for
Then, since a
v)
= (p(u+ KgN,
fi
denote
= Kg,
and therefore'
we have
^ (m + aN,
where
[359
(w
ft,
v)
= <f>(u +
aN=Kfi,
a column of integers
K/j., v),
N being
thus
</>(
(1),
a column of Ip
+ aN,
v) is equal to
where
R = liribN {u + Kv +
ii
aN) + 2incN +
m 'I
K^ iV N^
N being
(fi,
a column of integers
thus
<}){u
+ aN,
v) is
equal to
,y)e^"''''^(
<t>(u
therefore
= mi,
/tf
yitj+p
m/,
to consist of integers,
<l>(u+
aN,
v)
etc. for
it will
=
<f>
i<p + 1.
e/tv
If then
= mft' m'n,
we take
if,
as well as
fi,
follow that
(m,
i;)
e^'''''^
k^N^Np^
4>
(u)
4> {u,
p)
Next
(j)
(u,
and
fj,
+ v) =
(f)
[u
{n-^-n')
t^)
this
where
ilf
= 27riZi' {u + Kti + ^Kv) + 27riCi; - irinn' - ltd {L(i + Lv) (m + ^Z/t + i^Kv)
- 2in (Gfj, + Cv) + 7ri{m + m) (n + n')
therefore
<j){u,fi
v)
=; ^iiTiLft {iKv)~^viLv (Jjf^) +in;nm'-f irinn'ffi (TTM-TO'Kn+n')
359]
n-i
so that, since
fi,
593
is
is
unity.
Hence we have
^{u,
We suppose
now
(m)
<})
(f)
and therefore
values
+ v)
fl
be carefully noticed that this equation does not require /4=0 (mod.
It is to
= bN,
Kfi
^ (m,
(u, v)
(f>{u,
/j-
+ v) =
<})
/i,
{u,
Then
g).
= aN,
2 K^NaN^ = GiJ.-^mm'
I/)
v)
/J,
cN-ir^
^r).
gN,
is
matrix modulus
The
than unity
2p integers
And
g.
set oi\g\ integers gr, wherein r has all rational fractional values less
will
now be denoted by
in particular r
v,
/^
(mod.
g).
shall
r'
less
than
1.
With the
let
v,
v/r(M)
= 2e-2"'^''^ (,!/);
then
yjr
(m
+ KX) = 'Ze-^'^'-'"
<f>
{u
+K\,
v)
for
any
set
i<p+\,
of integers \, as
\i-=li, \i+p
for
k').
1,
p')^
g2irir'A+li7rii/A
'^gi'rir'i' J,
/^
j,\
B.
That
<t>
(u, v) is
unaltered
when
to k
is
added a column
(mod.
g).
38
594
284,
Now
h derived
a, b,
%^fr (m)
v'
since
= gr, we
v'
have
"'
[359
v'r
v)
then
f'
= gr'r = grr' =
vr'
thus
v'
v'
this
J/
have
\k\
(\g\y, so that
\g\
some ot
zero, as is easy to
Hence we have
g,
equal to the
Since k=geg, we
=\g\-
= J\F\.
so that
g),
the following
^/\k\
(j>
is
360).
fundamental equation:
(u)
= 2^ , ^ (m, C+v),
Thus between
there exists
Ex.
into
is
(p
<^ is
a function
<^
coefficients*
for
which
Vl'^l is
changed
ii.
And
necting any
p+2
when n
homogeneous polynomial relation conof the same periods and parameters.
(Cf. Chap. XV., 284,
same
of the
functions
large enough.
Ex.
fxmctions
iP '\l\k\.
'i
i.
Ex.
p+
'J\k\
functions
(f>
V.)
We
360.
If
(a)
|i
(mod. g) or
cN +
results assumed.
fi = gN,
ka^N^Nf,
where
Gfji
^mm'
(mod. unity).
For
kafi
ip
(geg\fi
= 2 (g)^y(eg)y^ = S
P,
= 2 g^y, Z
[eyxfiTxp
[ey.xS'A,?
ey,^+pgK+p,fi]
\=1
y=l
_
{g)ay
ey,K+pg>.+p,i\
Igy+p^a.
y=l
y=l
7=1
[ey+p,\g\,fi
ey+p,K+pg\+p,fi]
=2
[gy+p,o.gy,P
9y,<^9y+P,?]
>
r=i
* 'Weierstrass,
Compt. Rendus,
Picard, Poinear^,
360]
595
therefore
=^ -
[>Y+i>,.-^
5'Y.^-^^
+ 5'Y,-^a
(mod.
gy+p.^N'^],
2),
=2
where the
22
^'S.g^+p^^Na gy,Nfi
<<?
(mod.
y=l
= /3
summation extends
to every pair
2),
o,
/3
thus
2p
y=) o = l
=2
[5r^,iiV^,
=Z
fiy./iy+p
+5'y,2p^i>p] [i'w.iiV',
+ g^+p_y,N-y,]
= 171711',
(mod. 2);
y=l
therefore, since
unity),
and therefore
i ^gy+p,aJ!^a.gy,aN^ = ^
{l]
N,
we have
ci\r
+^ 2
A:,3iV,iVp
cN-
+ ^mm' -^[g]N=
{gC
^[g]} JV
+ ^mm' -
(b)
and not
(v'l,
ri,
1),
as required.
than unity
all
,
be any
ry,
N,
i[g]
(mod.
zero
the form gr' for values of r' less than unity, then
(e~2>,^i'',
^g-2mr,y,
(e-2"'-:rt,)i.'
v'
Since, as
is zero.
remarked
359), the
2 (e~^'"'''Y'
sum can
also be written
(e"2"'''2p)''3',
r'
wherein
v^,
..., Py,
each
of
any
all sets, v, of
set
sum
is
is
incongruent columns
of incongruent columns v
(for
namely
it
the modulus
has
g).
an integer, we
shall
of in-
congruent columns.
382
596
Suppose then
in the
above sum
r^
the column
t'-th is
zero.
is
fractional.
(0, 0,
. .
1, 0,
is
. .
Add
to every one of
0),
of which every
is
unaffected
Hence the
e"^'"'"*, which are multiplied by e"^"'"'.
therefore
be
zero.
and
must
unaffected when multiplied by e"^"""',
sum
[360
We put down
^ = 1. ^=li 2p
then 5fr=(4ri + 5>'2, r-^\%r^ and the equations
ri
Cj, r^
and
+ 2r2 = i/J
v,, ^2
'l.
are given
I-I
gr=v
give
i/,
361]
OF A JACOBIAN FUNCTION.
ab-ba = k,
the conditions
ab'
and geg
= k,
~ 3,
-a'b =
if
= {a,
597
= (b,
a), b
- g^g^ = -
gug^x
b'),
become
we have,
if
ikxx^
where a
(x, x'),
x-^
= (a\,
and ax
a-/),
'
+ a!x = 0,
= o + i/3,
> a'^-
a^'
a;
4.
The
a'
a'
+ i/3'.
quantities
a, b,
a,
= ^^' (a/i?' -
-ab = - 3, we
a'/3),
must have
b'
= Lg
cm/)
the equation
give
= {4:K + K, oK +
{a, a')
2K')
there-
fore
3K =2a -a
3K'
further the equation
(c, c')
15
= 4a' -
c=gC + ^[g]
(C, C")
3L =2b -b'
3L'
5a,
46'
- 56
gives
+ i (4,
10)
(4(7
C + 2,
5C + 2C" +
5),
so that
30=2c-c'+l,
K=\2to,
Also, from
2oi
,
\
2-TnL
=^
2<u
=
=
3C"=4c'-5c-10.
|
2t], 2i?'
,
|
^", b
with
2ha,',
2cu
we obtain
a
= 7n(26-6')/(2a-a'),
^ (it
If then
h.
= '6Tri\{2a- a!).
^,
given by
(if C)
= 2ea"+2hK(+C)+b(+0-2WC'(+C)
a,
6 as periods,
and
c as
parameter,
is
given
by
S<f){u)=tA,.'$,(ir,
C+r),
V2/3;
?'
is
in turn equal to
11,
/2/3\
'
The
U/sy
function
third order
its
<^
(u)
may
the matrix g, are transformations of one another, in the sense of Chap. XVIII.
and XX.
598
is
which has
and that we have
The theory
862.
been given
[362
2p.
when
o-
= 2p.
Suppose
o-
<
2p,
AB
matrix of integers
a;=0
AiX
or
There
for
which iKxXi
is
positive
when Ax =
0,
except
when
= 0.
will
& function
<J>,
involving
VlKl
The
is
function
<f>
which
(w),
is
B are
</>
<I>
(w), in
are satisfied.
and
which
Crelle,
363]
CHAPTER
XX.
363.
in the argument.s u, with characteristic {Q, Q'), say ^ (u, Q), may
be regarded as a theta function of the r-th order in the arguments w, with
first order,
Let
satisfied.
(I), (II),
be denoted by
K).
IT (w,
We
confine
Q)
even
is
therefore, since
Then the
= Mw,
is
when
odd or
K) is an odd or even
have shewn, in Chap. XV. ( 287),
Now we
function ^{u, Q)
r, is
expressible as a
y\rr
K,K' + n) = ^ Inv
(w;
+ e^\-rw;
where
e is
1,
{K'
+ fi)/r
K
|
is
^ ^^^''
K
is
^^
^ (w
364370)
The theory
e is e'"i',
I
Ql
(mod. 2)
is
odd,
.K"!,
i
is
yjtr
(w
K,K' +
fl)
r,
GENERAL STATEMENT.
600
which
in
may
not be
When
e"*'-^'.
even,
is
e,
though
[363
equal to
still
the transformation
the value of
also,
e'"'''2i,
may
or
which determines
in the functions
i/r^
(w
the 2P functions
'^
(w
the
coefificients arising in
these
polynomials are theta functions whose arguments are r-th parts of periods,
The completion of the theory of the transof the form (2vm + 2v'm')/r.
formation requires that these
coefificients
characteristics
half integer
( 373).
Further the theory requires that the coefficients in the expression of the
K) by the functions i|r,(w; K, ^' + /i) should be assigned
function 11 (w;
in general.
is
by the juxtaposition of
linear transformations
(i 371-2)*
364.
is
prime number.
two
cases
when
= 2, and when
is
We
former case, and shew that the transformed theta function can be expressed
as a quadric polynomial in 2^ theta functions belonging to a special Gopel
system.
is
given later
( 370).
For the transformation of theta functions, and of Abelian functions, the following may be
Jacobi, Crelle, viii. (1832), p. 416 Eichelot, Crelle, xii. (1834), p. 181, and Crelle,
XTi. (1837), p. 221
Eosenhain, Crelle, xl. (1850), p. 338, and Mem. par divers Savants, t. xi.
Henuite, Liouville, Ser. 2, t. iii. (1858), p. 26, and Comptes Bendus, t. xl.
(1851), pp. 396, 402
*
consulted.
(1855); Konigsberger, Crelle, lxiv. (1865), p. 17, Crelle, Lxv. (1866), p. 835, CrelU, Lxvii. (1867),
p.
58
Weber,
xii.
H.
69,
(1867),
ix.
;
(1878)
Thomae,
Kronecker, Berlin.
Smith, Beport on the Theory of Numbers, British AssociaWeber, Acta Math., vi. (1885), p. 342; Weber, Elliptische
Functionen (1891), p. 103; Dirichlet, in Biemann's Werke (1876), p. 438; Cauchy, Liouville, v.
(1841), and Exer. de Math., :i., p. 118; Gauss, Werke (1863), t. ii., p. 11 (1808), etc. Kronecker,
;
J. S.
(cf.
Frobenius, Crelle, lxxxix. (1880), p. 40, Crelle, xcvii. (1884), pp. 16,
35 Wiltheiss, Crelle, xovi. (1884), p. 21 the books of Krause, Die
;
Transfovnation der Hyperelliptischen Functionen (1886), (and the bibliography there given),
Prym u. Erazer, Neue Grundlagen einer
Theorie der Doppeltperiodischen Functionen (1895)
;
364]
By means
with
of the equations u =
half-integer
(w
characteristic
K, K'), of order
2,
2ft)a
2ft)'a',
becomes
2a,
2Jlfu'
function in w,
theta
2&>/3
2F(7;a +
4- 2ft)'/3',
2rj'
2ft)', 2>j,
and
7;'a')
4?',
- K = $Q' -^'Q-^d{m,
Z' = 5Q'-'Q-4d(aa'),
+ V^) = 4?',
2Jtf(^/3
a function
2Mv =
Mw,
601
and
5a'
Vr^
(w
-B:.
+ /*) = ^ \rw
'
(^'
2v, 2rv', 2?/r, 2?'
+ 6^ T-rw;
r being equal to 2
here
e,
= e*"^'^',
is
1,
equal to
symbols
e-^'^""'^->/rr(';
^('|J)'(^iJ)
K, K'
respectively by
is
even or odd.
For brevity
^^^^^^ respectively
final
formula of
v, v'
f, ?',
K'),
+ fi).
^[;;2v,2i;',2?,2r|j],
Taking the
+ m)/*-"
K
is
463, be expressed as
K,
is, according as the function II (w
we put w = 2vW, xn' = v, and denoting by
we consider the function
that
which
p.
i(^')
(t^;^'|J')-
i(^')
+ (f )
to,
w',
r,,
V,
(^ )
'
or 1, and
each either
equation by e"^^-^-'"', where /* is a row of integers
each
of its
which
in
values
all
adding the" 2P equations obtainable by giving a
elements
is
or
1,
we obtain
602
@ V-U;
2gii<i(|u.-X')
VeTt('+M
2@
t'
2F:
@
e 2U;
2t'
2t'
A'
A'
U respectively
2W-2t'
2^@
=
may be
This
we
Se"'"''*''
^'^
e[':.-J4(^')]e[ir,.-l4() + (f);
'
(if,
K')
is
= 2~-P2e''"^= W:
2t'\^^
it.
we
obtain
iOy^b'^V'T'i
put
0,
When
2F;
by W,
[364
a',
which
for simplicity
may be
= 0.
We
and a has
(ii)
'2W;
in which
a' is
When K' = 0,
2r'
))
it
^[^
K=^n
is
an arhitrary row of
which
its
integers {each
or
1.
365]
f-^
It can
j.
603
quadric transformation
is
(cf.
As
it
415 below)
may
so that
often be a matter
it is
functions
-2TF:
2t'
by
this
case.
When K'
(iii)
is
not zero,
sign of
W in the
2^0
= 2e,i,^-ir-.
(70 r
t'
i (^')
I
ok'
= k (k' + a.')
= 0.
In either case
it
right side of the equation contains only 2^' terms, and contains only 2?'
theta functions whose characteristics are a special Gopel system.
It is easy to see that the results of cases (ii) and (iii) can be summarised
when the characteristic (K, K') is not zero the tratisformed function
a linear aggregate of 2^"^ products of the form ^ [w ^, PJ ^ [w A, K, Pi]
as follows
is
A,
iLj. ^^\_
j^)'
'^'^
e.
These results are in accordance with 288, Chap. XV.; there being
2P-1 (1 + e) linearly independent theta functions of the second order with
zero characteristic and of character e, namely 2? such even functions and no
odd functions, and there being 2*^' linearly independent theta functions of
the second order with characteristic other than zero.
Ex.
365.
e\W;
r'l
When p = l,
i.
_f)
,^
as
is
results
of
case
(i),
if
we put Qg^iW
usual, are
eoo{2W; 2t) =
,
the
eL(Tf;
r')
+ el{W;
2eoo(2T')
r')
e\,{W;
r')
+ el{W;
2ei(2r')
r')
t')
for
604
[365
and
e,o (2
where 6
(2t')
(r
denotes
Jr
_
207(2/)
e (0
If
2t').
vx'=l4;^>
600(1")
OooCr')
^^-VIeo,(2>r;20' ^'-^*eo,(2W^;2r')'
Vleoi(2^;2r')'
^^-^
^''-Vxe,(ir;r')'
''^^Vxeo.e'^;-'!'
find
(2r')
notations
vx=1^!^;!.
^
000 (2t')
eoo(2r')
we
we introduce the
then
n/f=|24|4-!.
v^=?4?!;)
"^
2900
eo.(Tr;r')'
e;(F; r')-el,{W;
= eJ(F; r)-e\,{W;
r')
r'),
while,
/,
1-X'
-,+x'r^
2t')
^'
v-x'^-^
'
2V^
and eio(2W^;
1^=0, we obtain
Ex.
The equations
ii.
eo.(2.r;2/)=
of case
these
(ii),
also for
we have by
Ex.
20^ ^^o(^r;
/)e(ir; /)
.
01 (2t
division
1,
we
find
= [l-(l+X')^]/Vr=X^,
When p= 1,
iii.
= \k'^x.
give
e(2Pr;
(''^
p = 1,
^o(>r;.Oeo.(T^;0
001
From
\/F=[l-(l-X')|]/\/l^X^.
r^nWi
^0:
when
2 is
increased by a period,
(-)<,.
w=0, we
Uoi(')J
we immediately
(
find that it is
is
W;
(iii)
of
above, the
='u (r)
a theta function in
r')
e^{W;
rye^iW;
r').
t')]
365]
THE ELLIPTIC
IN
which
is
CASE.
605
^^
e'(T')eoo(T')
01 (r) e,o
[We may
286, p.
(t')
d^
rt
^/4f
(1
_ I) (1 _ X2^)
obtain the theta relation, here deduced, from the addition formula of Ex.
^Pii(i) ^e immediately
a2=j(_J),
^-=0,
By
iv.
we
i.,
ii., putting
W=0, and putting
the same example and in the second value for e^{2W; 2t') in
obtain
e'n(2rO
e'liCr-)
6^ (rO
we
= CO
last
please.
,
is
Hence we immediately
example be denoted by
J we
(r')
e, (r')
'
unaltered by replacing r by
find
from the
equal to
is
e,
w.
Hence
2"r',
n being as large
by putting
and k
coefficients of v
?=iQ).
5ii(+w)-5,i(K+i>)5(M-i))];
if,
Ex.
i.,
derive
>
5loM9ooM5ll(2^')
m = j(_J), ai=i(_;),
way we
by the substitution of x
find
for ^
Hence
for X.
/M=2e^(2r')/e^(T'),=i+x',
as follows from the
From
first result
these results
we
of Ex.
sn|^(l+X)^,
and
1.
dn
r+X'J~
(J, X)
ii.
in the
form
'
JKr. V.
(f^
Prove when
is
p= l
e =eoo(2TF;
2r'),
G2 = e,o(21F;
e = ei(2lF; 2/),
2r'),
e, = e(2r;2.'p/^)+e(2If;2x'|-^/^),
e,=e(2Tf
(2 If;
2/| ;g)+.-e
2.'
|-;/^),
e, = e(2Tf
G,
= e(2>f;
2r'),
2/^/^)- e(2H^;
2.'|-^/^)
e,=e(2lf 2r'\\/';)-is(2W;
2.'|-;/^)
the matrix
(n o
606
CASE.
[365
q o
^'"^
Obtain from the formulae of the test the expressions of the functions 64, e^, 65, e, of
the form
C'4,
Cj,
C'g,
C, are constants.
E.r. vi.
appear subsequently
415)
by combination
the matrix
[(!+/!;) , 2
it
(f,
which
is
supple-
(. ^)/[l
+* sn^ {u,
,
'^e
*)]
tion formula
e(2Tf;
where
r')
= 4ei,(Tf;
a constant (c Ex.
is
The matrix
r\
vii.,
associated with
for instance
we have
(rj)G?)(-i)=c)
with the corresponding equations
V=tW^,
W^ = iW^,
W^^-r^W^-
r,= -l/r,
r2
= r,/2,
T3=-l/r-,,
-iHlP
=e
(E,
P being
from those
il'Soo
rj),
for &^^ (2
366.
We pass now to the case when the order of transformation is any
odd number, dealing with the matter in a general way. Simplifications that
can theoretically be always introduced by means of linear transformations are
considered later
372).
366]
We
first
^ Fto;
607
^^'
"^'^^^""l
Let
o",
by
and
o-'^'i
matrix
o-''^',
2i;o-
let
the
of
j'-th
+ 2i'cr',
the
Z^
given by
is
rows 2vK+2v'K',
2^K+2^'K'; and
XT
sum
the
r,
s = (s^^\
(w
K, K')
..., s'"")
2i;o-'-"
of
any
let
let
finally, let
T, denote the
I'-th column
also,
be denoted by T^
1, 2, ...,
denote the
= Zk{w + ^Tk) -
be a column of
iriKK'
so that
i
^ (w)
for
{iv
2f'
and
For when the arguments w are increased by the elements of the row
where N, N' are rows oi p integers, the function
T^r,
is,
2v', 2f,
p. ( ,+T-:iI^) + Tfl
is
"
+ 2?'iV'5'^'']
"^^
'
-^j is
equal to
'^'
-iri{Ns^^]{N's^^],
that
is
[s^^Y
the
sum
of the
-f,-
[^N (^ +
'^^^ + iT;,) -
m values of -^j
is
-rriNN']^
+ Z^rf
s^i>
;
given by
Z^vT^s
+ Z^T^s
Mim. par
divers Savants,
t.
xi. (1851), p.
402.
608
when w
also,
by - ZjfTjv-
increased by T^f,
is
function
tl^e
{w
r-cr
Kjr, K'jr)
is
is
increased
is
equal to
is
N, N')
- rvr [w
thus
[366
284,
p.
n(w) = 24^^ rw
where /4
and the
is
!(if'
2f
+ /i)/r"
K
a row of
coefficients
replacing
by
w + 2vK/r + 2v'K'/r.
K by K +
and
less
M,
where A
h,
than r
is
in the
ofp
integers, each
form
^{u; q + M) = e^"^'''^{u;
we
a row
q),
find
- 2x! (K' +
K'jr']
hjr
e)
,.i)(,
= 2.4e2'''0'-)ft/'-Qf rw
where
than r
is
;
ascribing
now
2i;,
+ MI-zi>UT</1
A{K'
+fM)M
^)lr
K
J'
\^ or
and
0,
according as
e=
/^
C^b \rw
or
2v, 2rv'
^ 0,
(mod.
r),
^^'
,
2^lr,
2f
'
we
infer,
by
+ ^^Z"'^]
where
yjr
and C^,
= rPA^,
independent of
*y
K'/r]
less
fact that
367]
We
367.
(a)
if
integers sW,
...,
G^^ \rw
609
s^
2i.,
cr,
o-'
be unity, so that
consist of zeros,
m = r,
we obtain
^^' ^''^/^
1
2u/n-Tjf1
^']-
W,
t')
g-lwiK'lW+WK'/rJ-iiriKK'lr
v,
v,
(rW' TT
f,
loss
i^'
of generality, by
means
of
"^ '*^'
*'
^^k^K+r'K'
(I)
A
where C^
is
independent of
This equation
is
it is
is
Ex.
i.
Ex.
ii.
When r = 2, ^ = 2, we
If \,
fi,
hhe rows
of a different
317,
@(W,
Ex.
vii.,
t) with
have
r,
independent of W.
(/9)
if
each of the
if
integers
s ",..., s""'
"'
will
^l:ii
r'
r'
'
its
i-th row,
elements
Hi
"'
B.
20;
T'
(r-l)n
,.
39
the
THE FIRST
610
where
fi;
the
is
sum
[367
STEP.
row
so that
(fli, ...,
%)
rw, 2v,
2rt;',
by
XI,
T,s
will
2f/r, 2^'
the theorem
where
'
.7
-|r
is
+ Tx- /j
n ^ - + --^r-^+^-
-^)
2r j
^ r
=
2u/t
Se*
be
+ m)H
(A-l''
C3
a-,
2r
r-1
(r-l)fii
And
366.
K
h + K + t'K'
A
[/"=
where
the
J;
t')
u,
^L + ^^;
r')
number
for if
(11)
= (hj,
(mod.
Ex.
r),
i.
. .
ii.
r= 2, we
JCoe(2Tr,2r') = e(ir,-J,
Ex.
?7 H
obtain
For ^ = 2, r=3, we
on the right
w^-\)e{W^+l,
side,
ir^+i)
THE SECOND
368]
We
368.
now the
consider
%\rW;
terms of functions
in
611
STEP.
tt'
+ '")/'' H-eef-rTf;
+ 0
TT;
r-r^^'^'
^^f>^''
when
in the case
is
We
odd.
e is also
equal to
e"'
^
'
It is immediately seen that equation (I) ( 367) can be put into the form
-^
t'A-'"1
K
h-(r-l){K + T'K')
'
i{K'+ff
= 2e
%r
w+
K'
I
+ ^]
C,e''''^'"'''-''-'''"''"'^r[W;K.K'
Se"^""
\^'^'')
''
|e-2'"'
'-1'
[h-{r-
1)
^'
^ '
K'
TT+i
^g2iri(r-l)X'ir@r
^"^-"U-']
ee-*"''^*^', = e-^''i*'', by
{K + T'K')]lr, which
unity,
an r-th part
is
We
of a period.
@r
W+a
g2iri|r
-l)K'W@r\-^_a'
-^"1
functions
can he expressed as an integral polynomial of the r-th degree in
system
Gopel
any
of
characteristics
the
of
are
S^IW; T'\APi], where APi
half-integer characteristics whereof {K, K')
From
p.
obtain, if
one characteristic.
513, putting
/P,A
is
e.-,
rP
(^) ej
C =0,
A'
respectively
= A, B = P = ^(J^y
by W,
a, b,
eu
P. = if*'),
_
as in 287, Chap. XV.,
m being integral,
2jriA:(rm-2M)_j2TiXm_^4inA:
and
tm/2'1
392
we
612
2P@(Tf+a,;
[368
A + P){W + b; A)
where
^+P+P.)0[; A+Pa];
X{u,v; P, 6)=2e.e-i"^a@(w;
a
terms
we may
element being
characteristics, each
or J.
Hence we get
X^^' 'l'^'
2P@'{W + a; A) = X
is
^^
4+P.)@(F; A+P^),
Se, (F+2a;
equal to
e'
where
o-
^ X (^.
'
0. g)
rr
X(2a,0;0,e)' ^^
we obtain
2''"-"?
"(
= 2^,S@,2ff,2e,... 2 ir._,x(F+ra,
<1
"l
<4
'!
where each of
W+ a
P., e)
P.,0'
^)
TT; P.,
...
+P.,_,;
e,_,),
ej,
e^,
becomes
...
...
in turn all
relate respectively to
+P..., e^]-
^[(m +
l)a, 0; P..
...
+ P._,,
(m=l,
0^ = e.y"
F ^ + P.J,
;
\ = -
i,rt
(^',,
..
The equation
is
teristics
functions
by
of the (r
PF
^)
[W + ra; A +P.].
...,r-l),
...,r-2).
as an integral polynomial
@{W; A + Pa),
ej,
+ 9'..,) ?.,
(m=l,
which
(HI)
and further
Pa,, Pa,
H,^
(3a,
%_,
P.,, P^,
_ y (2a,
is
of the
whose charac-
first
degree in
^\
368]
which
is
613
W+ ^) @^(Tf
J),
which
is
The reason
is
308, p.
-w;
that ;;^(-M,
(P),
efficients
Hm
P,
e) =e''''^'+"i^--Pi
characteristic .4
/K'\
we
2('-i)p [e-^rfc-DJC'W''-
is
by change
are unaltered
infer,
x ("-
v;
(F+ a
A) +
P,
e),
of the sign of
a,
so
(III), that
e^'i'-"*^''^'-
(Tf- a
A)]
equal to
SffiS
where
P denotes P.^ +
[h (r-
equal to
1)
Pa^
(K + rif ')]/r,
U=Hr-,[e-'^^^'^^'^xi'^+^<^'
and
it
'^''
P,fr-i)
+ e^"-"^''^x(^-'''<^;
if it is
"^rV^
@{W; A Pa),
K, K'
/j-]
is
P-e^-i)],
is zero.
The
expressible as a
polynomial involving only the functions (W; AP^)For the fundamental formula of 309, p. 510, immediately gives*,
any values of
for
a, h,
x{W + a, W + h;
P, e)x{a +
b,
0;
P,
e)
= x(a,
h;
P,
e)xiW+a + b, W;
P,
e),
U is equal to
{r-l)a; A + P + Pa]
2e^e-4"9'" {e--<('--K'r@ (^W+a; A +Pa)[W +
+ ^(r-i)K'w@(^W-a; A+Pa)[W-{r-l)a; A + P + P^]},
where
P,=i(^\
e,_i
is
simply by
used
for
e,
the expression
Pa^+
P.,.,
and e...
for
(e^_Oi,
h-{r-l){K+T'K'),
Replacing ra in this expression by the period
h, K, K' and P, it
r,
on
only
depending
and omitting an exponential factor
(er-i%,
becomes
2S.e-*"'"{0[^ + a;
^+P + P.]},
We
when
G14
A
is
fl
and
('"^')
f.
[368
e,et*-''-"J>^]'+"^<'-'-^'-;
and
this
^P
(t)
e-i^l-P
i)
2r.e-i-a@
{W + a; A + P^) {W - a; A + P + Pa),
when
?'p=
e*'"'-'''
p)
that
Now
@[ra;
is
A + P + Pa][0; A + P.],
where
denotes Pj
still
. . .
+ P.^^
and
is
e,._2
equal to
sr.no; APa),
(l
?p
according as
group of
the group
fp
is
P=
2*""'
(0,
(p)
[0;
A+P + Pp][0;
P^
is
to describe a
P)
not equal to
e^ ^
'
'
p)
>
We
shall
A+Pp],
assume
group with
that,
when
t'.
^r(W
Let any
2*'
functions
^ (m
2ft), 2ft)',
2tj, Irj
369]
015
^(w
2)
=3
is
at most,
p=
the reader
and
may
for
d.
Mat.
2' Ser.,
ix.
t.
(1878),
126.
p.
369.
^\
(^
then
]^;
or 5i
(m), is
equal to
^m (3w
2,i,,
26.',
M=Z,
the function
2,'|i(_^)],
2,,
0).
Now we
have,
witha = (A + l)/3,
^0*3 ( ^; - i,
also e^j
if
TT+a)
0)
=2
[e^,
W+ a) + el, W- a)]
{
equal to
is
to be i (
V ^ Qj, so that P, = i
eoi(a)+tieio(a)
,^.
etc.,
this is equal to
eoi(3a)ei+.i'e.o(3a)e,
- u^ O qi (2a) 6, Ja)
ei(3a)eo:-tVeoi(3a) e,
e^,(a) + f,e^(a)
e'
^.
^'
and
i5=ei(ir+3a)e,(TF)+e,'ei(ir+3a)e,(lf),
^, = eio(W'+3a)e,(Tf)-Wi'eoi('l'+3a)ei(').
Now,
- u.'epi (2g)
Oil)
(g)
ei(3a)eoi-wi'ei(3o)e,o
vanishes
^
when
,^-=
_,i^^(^)-^('^-^A')<+^.J.'.
in fact, putting
(J) =i (J)^.!,
a=(A+l+^)/3,
amely, as
t.
*.
when
616
when
tfi'=e'"^
',
functions, being
is
argument
for zero
We
[369
ii'
equal to
.'[e.,(Hl)e,.('l)-e.,.('i)e.('')],,
in
"
while the expression e,
and
H'- 3a)
of this.
[e'j
e',
ei
w) e,, ( W) (
If )
e'oi
equal to
If) e,
W-\- a) can
is
W)] x,
W)
the negative
If) is equal to
If),
Om ( W),
W+ a) + ^j W a)
(
only.
(g)
Oio (3a) e,
,'
4,- (ft+1)
Ex.
i.
\^J
e,
eo.tt)eo.-eL(i)e
1^01 \i) 01
-e" '''+",
^2
'
namely,
W) e,
If).
equal to
Boi
- <,
(i) e?]
{y>) 5? (')}
e^o(i)eg.(i)[e.o(i)e.+e,.(i)eJ
,
where
ei, Sjo
Ex.
eoi
ii.
(0)
and e^
(0) respectively.
Prove that
W- 1) e,o
denote e^^
Tf + i)
- e, (
W- 1) e,
e'lo (4)
01
if + j)
-K(H')e.<,(if)-e,(lf)e,(lf)].
The
When
{K, K')
is zero,
2*-
370] GENERAL
When
the characteristics Pi are those of any Gopel group whereof the characteristic K, = {K, K'), is one constituent, and 4 is a characteristic such that
\A,K\ = \K\,ov\A,K\ = \K\ + l (mod. 2), according as the function to be
expressed is even or odd*.
When {K,K')
is
1,
W by
i^iir,
where
/i
is
gives
= 0.u=W+a.b = a = h/2, ^ = ^
we
= .K" = /i = 0,
(j)
P,-
=^
(p.
,
and
(^^^
(^^)
e^^^Ui
Ki,
obtain
2PCef2lf;
2t'
'^I^')
Pi
s(^^).-.-f.0=(O;r'ig)+P.
i
where
C is
independent of /x.
It
is
t'
APi)
is
2*"
f;.
This formula suffices to express any theta function of the second order with
zero characteristic.
1,
where
a= ^X-|- Jt'V,
We
When
= l,
288, p. 463),
when
and therefore
so, as is
|.ff|
When r is 2, it
which o = 2, ^=0,
r is odd.
always
/S'
have seen
= |Q|.
is
not
a'
= 0,
618
and
= e"IJ*^l + "M'^l,
when X
that 2pC'^,{W; K,
K' +
[370
h') is
emK-^K+ir'K')
{W + x\
A, K, Fi)
+ {W-x\A,K,Pc)},
M
where ^i=(
e"(*'i-'''i'ej,
^^"
and
'
(0
^, Pi)
be proved
It can easily
vanishes, for
0, for
e*"''-^!,
P,)
+ %{W-x\A,
K, Pi)]
the summation
is
=+
the
fj,
sum multiplying
is
may
It
e^"''-^'.
{ jt-]
sets of values
for
a;
by
Xeie'iU; Pi)
i
fourth roots
of
6;
(cf
(/S),
p.
514).
X.
Ex.
e(2Tr;
for
which {K,
A'')
-4,
0)
and
A'
2T')
= l, we
is
^= 1
= i*2(H^;
-i,
to the function
1),
where A =i
fi= - 1
is easily
found to be
P^ = K=^ (
=1
^^
the value of
we
E. for
372]
^j=e-"*-*
of which
formula
[e^o
(i-^-i)-
(i^+i)]-e,,
e'^0
x=0.
for
619
e, (O)
The
final result of
the
is
^')/e;, (0
/) e^^ (0
r').
Ce(2TF;
and
(of.
2r')
= 2e^a;
e, (0
r'),
^') e'lo (i
;'')=- 1 ef (O
^(0;
= ie^(0;
'')
e2o(i;
r')
e^ (^
r'),
r')e,(0; r')[e^(0;
r')
+ e^,(0;
r')],
= i[e;(0;r') + e^,(0:r')],
2r')
C=V2[e^(0;r') + e^,(0;r')].
The preceding
371.
all
^(m;
"^'^
or
or
S^
Q/
when expressed
In
many
^ (w
in terms of functions
particular cases
it
is
'2'^
(u\
\
\Q
[w
For
two stages,
function
this purpose
we should
But
it
consider
(i)
by means of functions
i/r,.
(w
K, K' +
yi), (ii)
the determi-
^ (w
by means of functions
\.
'
i/r,.
(w
K,K' +
/it)
of this chapter
but this
is
by no means
to
be regarded
Dealing
372.
(m
^
j
first
by functions
<^r
(w
K, K'
+ fi),
there
difficulty arises,
matrix
then
the row
(^ J)
K' being
&
(u
^') is equal to
^ (rw
is
is
namely ^ (u
is ^i/r,
{w
^'jj"^
,
K, K').
620
Now
THE
[372
it
transformation of order
square factors,
that
IN
n^ =
^'.
r,
il'
have been
upon the
given function
/r 0\
(
^(mI
);
...
as
an integral polynomial of
not a prime
order
We
in the
Appendix
II. (
It
prime
to successive transformations of
make a statement
can however
shewn
is
_,
where
D,
ways
all
is
all.
cases in which this has been done the required coefficients are given
In
by
the equation
-=;^(m;
V|
2m,2co',2r,,2r,'
lo
I
r^,,
^-^,^.+^^.^^.ZJ,,.^.^
rw
^^
''"'"-*''''
'^\M\\v\\B'\
u=Mw,
|ilf|
is
quantities
B'fj.
7 denotes the
IX.
(18789).
We
and for details in regard to the case j)=3, Weber, Ann. d. Mat., Ser. 2,
have shewn (Chap. XVIII., 324, Ex. i.) that the determinant of the
matrix of transformation
is
+rP.
is t>.
From
it
372]
denote
value
its
when the
characteristic is given
where
fi
= (^, ^,)
To prove
^,
is zero, its
621
by
and
Q,'
this formula,
first (
the equation
^,^M.;
where u
have
f^,
\p
2,2-,2,,
= MiUi,
^(m,;
ilfioi
V|').-;,^=|M".;
= wp + co'p',
Writing
etc.
'^)
,
a-
2.,2,-.2,2,.'|*'),
tt,
= 2<ai
fT',
<b,'
= (1)iTi, we
and the equations Ui = M,^w, MiV = tOiA, M^v = ta^B -\- m^B' give, if w = 2vW,
v = VT, and in virtue of AB' = r, the equations Z7i = ^ TT, rr. = .4t'Z BA,
while, by the equation rf= JlfjT/jJ., we find 7;i&)i-'
= r-?i;~'w''. Now it is
,
(Ui;
t,
)
gives
27rif/,n
iirTiV.^
= 2-7rirW (m + -j + mrr (m +
-j
+ md im + -)
y/jt.",
equivalent to n
unique
m and
B'm = -B'ii\
an unique
for
ix
corresponding to every n
which
is
integral
it
determines an
corresponding to any
TV
assigned
/i
for
which
is
integral,
determines an unique
n.
B r- m
tO
= 2e
'
+ dm
is
we have
an integer,
integ
is3
"'an
~'^"^
nidu.
matt,
{U,,
vi
iVd^
\rW;
~\
^^
rr'
^1 /
CT,
by
Q,
TiQ/,
we hence deduce
''J
for
622
where
of w.
[372
that (K,
so
K')
the
is
therefore |if
J5'
1;
= r^ |il/i 1,|,
= e^^^"'^"'', the
|
e^""'"'
"'^
When^=l,
i.
0{W
the function
^t)
we have
ir') = e{3W;
e(W;
+ e(3>r;
3r')
Other simple examples have already occurred for the quadric transformations
Ex.
( 365).
ii.
odd) for
(r
which
that
= J,|r(0, 0)+
ru; rr
'
'
+e
2
n=l
<
-njr,^-njry
(
i/.(ra,
-n^),
(^ji^heiss,
373.
w
the problem that arises
is
The theory
functions.
present volume.
for instance
of this determination
is
odd
periods 2v,
functions
'^
2v',
317,
2v,
2v',
for
^ (0
vii.,
(rw; 2v,
^ (w
By means
2v'.
functions
2f,
By means
2f' J
are
2v, 2v', 2f, 2^') or this determination may arise by elimination from
the original equations of transformation, without use of the multiplication
;
equations.
the functions
which
is
373]
IN
623
in terms of functions
3 {w
{u
Ex.
In case
i.
fiinctions
e^{W,
respectively
ig^
r'),
K-l!
-l)' *(-!,' o)
polynomial of the
characteristics are
*''^*^ ^'"'"
^^^
^J'^*^"'-
of these functions which are theta functions of the third order and of
zero characteristic are those contained in the equation
e, (3 W, Zr')^A<t.l+B<l,^i>l+C^^cl,l +D4,^<pl+E,p^ct,,<p^,
where (f)^ = ei,{W, t'), etc. this equation contains the right number ^(7-p +
on the right side. Putting instead of the arguments W^, W^ respectively
;
we obtain
1)
=5
of terms
e,,,
(3
in turn
e^(3W,
3r')=
A<t>l,+B4,^<t>l+C<t>^4>l+ii<t>u'l'l+<l>,<P,<l>^,
i>l
+ 34.^4,1+ E<t>^i,^^^,
^=
where S^ = &^(0
3t'),
etc.,
66~3434
4
Tj
'
and 65 = 05(0;
"^
r'),
e^ (3 IT
for C, D,
Ho', o)'
;
etc.;
4"^
'
H-i!o)'
{3W;
3r')
= A<t>l
9,2(3 IF;
3r')
= A<t>l, + B4>^^<t>l
e,
-B4>,<t,l
3r')
^3465-66^34
periods to the arguments, for instance, those associated with the characteristics
Ho!o)'
II',
-<?</>4< + i)</.4*?3-'S^<^3'^.A3.
+C4>^,<Pl +I)<l>^^<t>l
+ ^4>,<t>2't'm'
half-
PARTICULAR EXAMPLE.
624
[373
putting herein
of the expression of the zero values of the ten even functions in terms of four independent
quantities
Ex.
(cf.
iv.,
Ex.
Ex.
Denoting qj
ii.
iii.,
292
etc.,
(0
3r') qj
(0
t')
by
C^i, etc.,
Co,
(0
(0
3r'), (for
result of
(p.
t'),
By
and the
34(0; 3r')/6(0,
12(0; 3r')/6(0;
3r'),
3t'),
(0;
3r')/5 (0
3r')
634(0; r')/6(0;
By
r'),
12
we
(0, r')/5 (0
r'),
(0
r')/^ (0
r').
obtain
iii.,
292, for
= (2P-1)(0;
26[o; 3r'|i(*')][0;r'|i
where the summation on the
characteristic
for instance,
0, (0
namely
(cf.
Ex.
i.,
3r') 0, (0
left
extends to
whenp = l,
;
r')
+ , (0
is
r').
all
3r') i (0
t')
= (0
3t')
B^
(0
r'),
p. 188),
^/lk\+^/kOi'=l.
result of Ex.
iii.,
292,
Ex.
iii.
of 364
B.e[o,
where
and k,
is
,-!i(/;,)],
a row of p quantities each either or 1, so that the right side contains 2p terms,
are any rows of ^ quantities each either
or 1.
k', a*
374.
argument
is
follows
the
374]
M,
relations expressed
rM-'
2u,
2u'
2f,
2?'
'
Ex.
(cf.
2a),
2a)'
277,
27,'
ii.,
/3
),
0'
a',
625
If
of 324 (Chap.
now
X,
matrices
a,
a',
/3,
yS'
satisfy
the relation
XVIIL).
M,
Ivx,
Ivy
=(
2a),
2a)'
217,
2?/'
ax,
/Sy );
a'a;,
yS'y
'
rU-"^
'
2fa;,
2^'^/
we suppose that x, y are such matrices of integers that cue, ^y, ax, yS'y are
matrices of integers, and, at the same time, such that rx is a matrix of integers
such matrices x, y can be determined in an infinite number of ways.
Mw
Let
u,
integers, the
= oucm + ^ym,
where n
n'
S-
= OLorni + ^ym'
{u
R is given
where,
2a)r?
The
+ 2a)'w',
resulting
H^ = 277a +
if
277'a',
by
R = Hn (u + in) - Trinn
= (Haocm + H^ym') {Mw + Mvxm + Mv'ym') - Trinn
= {MHaXm + MH^ym') (w + vxm + v'ym') irinn
= r {2^xm + 2^'ym') {w + vxm + v'ym') - ninn
y8'a = r + y8a', and because ax, /3y, a'x, 0y, rx are
;
now, since
integers,
matrices of
we have
nn
/,
Thus,
d>(w
if
we denote
2a), 2u>,
2v, ^v')
by
<i>
(w),
we have
^,-)
40
628
Further
respectively
[374
if a, b
rows, given
by
= (2vx,
2v'y),
^irih
= (2r?a;,
2ri;'y),
we have
\'^{ab-ha) = {xv
){^x,
){vx, v'y)
K'y)-{^-l
\y'"\
\yl'
y{^'K-K'^)<, y{'"'K'-K'v')y
= |7n(
-xy
0,
);
yx,
\
so that
ab ba = k,
say, is a
ab
ba = k = (
rxy
of integers given
by
),
ryx,
I
and we have
ka fimainf,'
= S
a,
Finally, let X,
be rows of
fj,
quantities a (X,
/i)
x\ =
and
cr'
(a, y8
1,
p).
X^,
/Xj,
and
let X, fi
complex
consists of zeros, or
a
so that
= ryxmm',
( rxy)^ pm^m^'
T'yfi,
(X,
where*
t'
fj.)
= 2vx\ +
=v~^v',
is
0,
2v'yfi
+ ia',
say,
then by
we have
tfc
in
which v
form
/x,)
= yp,,
r<r'v}\ is
On
(II),
is
positive, the
a
*
= {2vx,
The determinant
2v'y),
2-,rib
{2r^x, 2r?'y).
= (i/,
hf')-
374]
IS
To
this function
express
by
there denoted
rx,
0,
we now apply
by theta functions of
it
g,
-1
0,
_
namely geg
1,
I
The condition
lu.
= k,
ra,
by
is satisfied
=(
0, y
627
ra,
^r
(rx 0\
0,
for
'
for
-y )=(
,-rxy);
ryx,
ra,
Q, ^
2r-f 'y) T
)=(2u/r,
-x-\
Hence, as our
;
functions*
^ (w
by
final result,
2f )
2r^').
sum of
can be expressed as a
= {^^,
2/->
,
I
^ (m
2v'),
</>
(w), or
constant multiples of
y\K = rf\x\
such
being
most
at
x,
y.
This
is
be the characteristic of the function ^ (it 2a), 2a)', 217, 27)'). The generalisation is obtained quite differently by Prym and Krazer, Neive Gruoidlagen
;
Ex.
other than those in the diagonal, which are each unity, and taking for the matrices
a,
respectively - E, 0,
0,
E,
factor,
zero,
a,
|S,
we have
the formula
/n
11^
ms/n\
or numerically
r, s are rows ofp positive integers, in which every element of r is
than m, and every element of s is or numerically less than n. This formula includes
that of 284, Ex. iii. (Chap. XV.) ; it is a particular case of a formula given by Prym and
wherein
less
Krazer
To obtain a
verification
That
is,
functions
2f
of.
is e'^,
where
284, p. 448.
402
628
hence
2e'^
r
unless
N/m
is
integral;
when N/m
is integral,
[374
'S.S'
='nP e^
where
K,
=nM+s,
obtaining
all
integral values
The theory
375.
when
M takes
all
less
integral values
than
and
takes
all
n.
may be
said to
have arisen in the problem of the algebraical transformation of the hyperelliptic theta quotients considered in Chap. XI. of this volume. To practically
utilise the results of this chapter for that problem it is necessary to adopt
conventions sufficient to determine
(cf.
The reader
is
(Leipzig, 1895).
* Cf. Eosenhain,
(1865), etc.
Mem.
416
ff.;
377]
CHAPTEK
XXI.
376.
former
is
the latter
is
theta functions.
than
is
Riemann
is
given
is
377.
In the transformation of the function
(u r), to a function of the
arguments w, with period t' ( 324, Chap. XVIII.), the following equations
have arisen
u = Mw, if = a + to', Mt' = /9 + t/8'
;
t,
in
which
t'
equal to
is
t.
We
we do
not enter
Suppose
that, corresponding to
a matrix
^=
oi 2p rows
and
where r
is
= /3a,
a'j8'
is
a;8'
which
= yS'a',
such that,
+ ra') t = /3 + t^,
is positive.
j)
=l
383).
146;
p, see Kronecker, Berlin. Monatsber. 1866, p. 597, or Werke, Bd. i. (Leipzig, 1895), p.
Weber, Ann. d. Mat., Ser. 2, t. ix. (18789), p. 140; Frobenius, CreUe, xcv. (1883), p. 281,
Wiltheiss, Bestimmung Abelscher Funktionen wit zwei
where other references are given
Habilitationsschrift, Halle, 1881 (E. Karras), and Math. Annal. xxvi.
Argumenten u. . w.
;
(1886), p. 130.
630
when multiplied by a
@ [(a + ra') w
t],
expressible as an
e"*^, is
2^ functions
[377
[w
t]
on this
The equation
tS') t = yS + to
equivalent to (^'
matrix
(a
+ to!) t = ;8+Ty8'
supplementary
_
/8'
_
-/8N
--(-?-?).
just as the former equation arises from
we denote by A
put
M=a + ra, N =
A \E,
A.
A we
where
/S'
is
ra
the
Then we prove
to
first,
X of order p
given by
iV^
Jlf
X = 0,
X =
|
there
that the 2p
modulus
roots are of
\/r.
be
shewn
to be all
roots of unity.
The equations
378.
r,
and
its
supplementary
transformation, namely
M=a +
Ta',
i/7-'
= /3+T/3',
Nt= -^ + T'd,
N=ff-T'a',
give
(a
hence,
if r
= rj
itj ,
where
tj
and
t.^
suppose t'=t
=f+ig aaAf-ig,
= j-/
tj
is
rj .A^= Tj (/^
say.
TiN=f- ig,
Mq be
not zero,
\M,-\\=\N-\\,
*
ir^,
then from
and hence, as
and similarly t
two matrices
Mt2
Now
parts
(a
therefore the
+ ra')T-'=/3 + rj3';
378]
631
which shews that to any root of the equation lM-\[=0 there corresponds a conjugate
complex root of the equation A'' - X = 0. Further we have, if t=t.^- ir^
|
and writing
r \ /a (3\
ll
rj
Mr
/JT
\
_ {M
~\Q mJ
\M, M,tJ
\a' P'J
/I r
\
'
\1 Toj
form
where
^=Clr)'
it
''^(ojSo)'
single quantity,
is
if
X be any
we have
so that
Thus the
of the equation A - X =0
roots
jiV-Xl = 0.
\M-\\=0,
Further, from a result previously obtained (Chap. XVIII., 325, Ex.), when, as
t'=t and 2<o = l, 2u = l, we have
here,
shape m,2 +
TO
+ i
Jij, ...,
form r^n^
np, the
a positive form,
it
=Ein^, say, .S being the matrix unity of p rows and columns, and
m = Sn,
is
where
/? is
a matrix of
we
for distinctness
real elements
the
shall write
then
:ii=Z-WZA'A7'-Z-Wr2^o^"'' = 7-2'-ir2^J"' = r;
thus
I
if
X be a root of
if
z,
M\ = 0, and
therefore, as
=x+iy,h6
a,
row of
also a root of
E is
we have
or
{\\-r)EzaZ=0.
Therefore as Ez^z, which
is
p
equal to 2 (^^+y^),
is
not zero,
it
= )';
iu
171=1
other words,
all
M\
=0,
A - X =0,
|
unity
A-X =
|
modulus ^r.
are
all
of modulus
:r''
of
are of
any
+ /la;-i+5.T-2 +
is
unity, that
Kronecker, Crelle,
mi.
...,
if all
+iV=0,
X are rational
integers,
and the
coefficient
is
Bd.
i.
e^**^.
(1895), p. 103.
NECESSARY CONDITIONS.
632
Let the roots be
so that
^= -(cosa4-cos;3 +
then
be a
finite
unity.
lies
between - n and
number, say
Thus,
k,
and
n,
B=coa{a+^)+cos{a+y) + ...,
...),
B Hes
between
+^n (n - 1),
etc.
...;
of equations of the above form, whereof all the roots are roots of
x^he the
if x^, ...,
[378
/i,
it
F,{x)=0,
F,{x)=0,...,
Hence, supposing
there
Xj x^^,
F,^,(x)==0,
x^x^^,
(x)=0, F^{x) =
to
....
consisting, suppose, of
I*
V
'
r,'
a equations, we
^fl
XVs,'
Ex.
(^''
i''')-th
if
A = ('',
,)
ii
V
.^ X. .
I'
=x'
root of unity.
M= a + ra, Mt'=P+t^,
[l
and deduce*,
'
infer that
^
and, hence, that x^ is a
U.
.* .
. .
r,
J(a',3') = (o
MJ\i rj'
and
that
_
AHA=rff'.
Hence, when
t'
= t,
if 2
which expresses a self-transformation of the quadratic form ffz\ which has real coeflBcients.
Cf. Hermite, Compt. Rendus, XL. (1855), p. 785
Laguerre, Jowm. de Vic. pol., t. xxv.,
;
379.
p. 285.
Conversely, let
(a
^\
= a'a,
^0' = ^'^,
* Cf. Chap.
afi'
XVm.
325,
Ex.
381]
where r
is
AX =
I
633
AX =
real values
Wi,
of
a matrix t can
we prove now
a symmetrical matrix,
that
for
n^ the imaginary part of the quadratic form tw is
...,
(i)
is
(ii)
+ to') T = /3 + T/S'
matrix A gives rise to
(a
is satisfied.
a complex multiplication.
380.
We utilise the following lemma, of which we give the proof at once.
matrix of n rows and columns, such that the determinant |A+X|, wherein X
order
first
when X
and
vanishes,
if x,
If A be a
ia
a single
y be rows of n quantities
hx=0,
then the quantity xy, =x^yy +
a;
by
^^'^nyn, is
|i, its
j=\,
hy=0,
not zero.
...,
similarly, denoting y by i)i, determine n{n\) further quantities rn^j such that the
determinant 1; does not vanish. Then consider the determinant of the matrix ij (A+A) |
the (r, )-th element of this matrix is
1
(i=l,
...,
n; j=l,
n),
...,
while
thus the
(1,
381.
>
ax+^=\x,
and similarly two rows
z, li,
dx+^'xf='Kaf,
(i),
ii x^
(iz-{-^'!!
+ ^x^'=TXo,
a^)
= 0,
(i),
(ii)
iiif,
\\=r,
T
a'Xfi+^Xo'=-Xo',
by the matrices
^Xq ^Xq^XXq,
*
and hence,
{A-'K){x,
az+^=liz,
from equations
or, say,
a, P,
a, ^,
we have
that
634
[381
which belong to the supplementary matrix rA''- just as the equations (i) belong to the
matrix A
for our purpose however they are more conveniently stated by saying that
;
= x^, i =
= 0;
(A-X)(<,
hence as
x, x' satisfy
the equations
= 0,
(A-X)(,z;,^')
follows from the lemma just proved, putting
words the quantity
it
is
not zero.
\)l{xJ - odz)
we
(i), (ii)
satisfied
(X/i
= 2jd,
- r)
is
not zero
in other
infer
- {ax + /3'a/)
{az
+ /3/)
this is easily
^, a,
a,
- xfz) =
{xz'
that tx+t'j^
found to be the
x^ - x'z = 0.
Also we have
oZq
we
thus
+ /3^o' = fio h
and hence as
is
we put
x=a:^-\-i^2t ^'o=='^i~^'^2>
is
is
if
is
we have
not zero,
3/<Sq
If
^V = Po %'
"'% +
37 Zn
\j.
^''=-^'i' + ^2')
termine only the ratios of the constituents of the rows x, 3/, we may suppose the sign of
tlie imaginary part of X in equations (i), and the resulting values of the constituents
of x and
x', to
be so taken that
XX^ X Xq ^ ^^
this
we
shall suppose to be
done
and we
(jo
1)
With
Ci, p)
Cl.
11
5l,P
x and
x'
be denoted by
>
z, z',
to the root
fi,
be denoted by
C2>ll > S2, PI
and so on
etc.,
for all
the^
roots X, ^
2.
11
) S1!.P>
zZq'-z'z= -2i,
are all expressed by the statement that the diagonal elements of the matrix
matrix
2i.
When
r is not equal to *
is
xzq
~xZq,
(r,
s<p + l),
the
(1,
382]
is
zero
we may
635
therefore write
^lo'-|'lo=-2^.
quantities,
t,
X'=^t,
Z=|<,
so that
-^o~6o'o>
-^0
so
^o!
then
it
be possible to determine
also
A'q'
382.
t,
If
it is
|' is
zero, so that
X' = 0, and
it
would
therefore
impossible.
+ Tx' =
as
0,
+ Tz' = 0,
...,
X, etc., or,
?+rT=o,
then
fl'-fl=rTf-rTr=r(T-T)r;
but the equations of the form xz'
^'
|
x'z
are equivalent to
determined so that
X' =
^t, for
can be
we have
arbitrary value
(T-f)Z'^=0,
and therefore
T
or the matrix t
is
and hence,
it
nant
^'1
is
+ fV = 0, we
have
= p + ia,
since f|'
l
where
= T,
symmetrical.
a row of any
ja
f 'f, = 2i, we
= ro-|',
or
quantities
tot
have
= aX:X',
and X' =
it follows, if
t) I',
a-Xo'X'
is
positive; in particular
rti' is
when
i, ...,
iip
+ ^x' = \x,
a'x
quantities, that
positive.
ax
of
+ ^'x = \x',
636
putting x
[382
TX, we infer
(/3
ar) X
= - \tx',
(/3'
a'r)
= Xx\
and therefore
T
O' - aV) x
+{^- ut) X = 0,
or
[i8
+ T/3'-(a + Ta')T]a;' = 0,
[/3
+ T/3'-(a + Ta')T]|' = 0,
and hence
from which, as
||'
is
;8
We
we obtain
not zero,
+ T/3' - (a +
to!) t
0.
It
may
may form
a theta
2m = 2|',
2ai'=-2f;
ciw' = 0,
Ex.
is
easy to
the matrix
a>
Wq =
2i
= cu,
fij,
...,
w' = cw,
383.
ft)
see.
v!
where
(i)5q
V^MpVi
\M-\ = 0.
|
/3'
are such
integers that
are imaginary
if a'
= 0,
&^)=CoD.
where a^ = , and then the equation for t is satisfied by all values of t this case is that of
a multiplication by the rational number a, and we may omit it here when a is not zero
we have
;
2aV=
-(a-/3')V(<i+/3')^-4r,
itself is sufficient to
equation
383]
637
If
- /3, we obtain a
\a h a)
for a value of r given
by
A-2a
'
where
a'
is
.\/4r-A2
"^^
2a'
'
corresponding value of
a + rd.
is
Hence by the
r, /I
a^
Q\\{ki'Jir-K')w;
when multiplied by a
The expression
(o
be taken positively
to
=^
J,
is
expressible as an integral
When
Mvaa + a'd,
i/u'
u = Mw,
to
the
A-2a + tV4?--An
case of transformation.
is
results of Chap.
if
2ia'|
n'
k, k',
m + k=m^ m'^,
= 7w+M'/3+-(o/;+j3i'),
n,
n\ instead of m,
J/^p(J/;|2<,2o,')
m\
= pO.|2,
as
summation
2')
that
+ 22'[sf>(, + ??^^^|2,
letters,
r- 1
aJc+^V = 0,
sets k,
'''],
2.')
- jJ (??^^^)]
dk+^k' = Q
(mod.
r)
are satisfied*.
This formula
tion
2<i)
*
i.e.
is
is
a complex multiplica-
= 2u = l,
When
2<i>'=2u'=T,
ff=h-a,
-^={a^-ha-irr)jd,
(Appendix
ii.,
k^'),
in
which
= (A - 2a iV4r - /t'')/2a',
k, k^'
have no
common
factor,
when o, a', /3, /3' have no common factor, the remaining solutions are of
where X<r; in that case taking integers x, x' such that k^' - k^'x = l, it is
418)
convenient to take 2vk^ + 2u'k^' and 2vx + 2v'x' as the periods of the functions
jp
on the right
side.
or
+1
or
The appUcation
=2
)-
m=na-n
resulting
the
of
ii.,
iii.)-
r= 1,
[383
CASE.
638
,a^ lia + r
;
m' = na
,
by
+n
,
,i
(/i
a)
^
ha+r
a^
rr
\f
we
MT,
Thus when ^=0 we have '3=0, and if a, a' be any integers such that (a^ + \)la
we have T=(i-a)/a', the upper or lower sign being taken according as a is
is integral,
are integral
when
hence as
\-{li-a)T
immediately find
negative.
positive or
equation
where
= 60 2 2' r
CTo
''^
When
A=
we have
we have
integral,
^2
mn
= ^)
*"d
if
T ('i-2ai'Js)/a'
TPTTi
+ (l-a)/a']*
.
[ra
u,
is
in this case
Ex.
i.
+1
is
^u by
jP(m;
cf^,
g^), it is
number
of the
parts
is
of the form
2'^
form
+ l,
4ifc
4, 0);
when n
when
re
is
a prime
is
number
n equal parts
Lxxiv. (1872),
any prime
a prime number
when
is
etc.).
of the form 6^
is
+ 1,
(Fagnano,
p. 362, etc.)
a^lr^
= fp{sla;
It is
0, 4);
These
Bx.
ii.
In case
= 2,
taking a = 4,
a'
= 9, = 0, we
/i
have
= 4 + in'2)/9,
(
and
-2f{i-J2.w)^ip{'w) + ^(w+^-iPir!2).
By expanding
find easily that, if
this equation in
(T/2)=e, then
of the transformation
-^^=^ +
we obtain
81^
r -=JL==W2 Jx
r .Jax^-ibx +
Jt \/8|'-15| +
^"^
verified.
and
It is manifest that
a.
l'
to this
384]
Ex.
Prove that
iii.
if to
= ^ (A + iVS - hF),
the substitution
3to-3
m^^ = x +
639
m* + 4 ^-1
dx
Ji V'(*+4)|3-15f-(TO<-ll)
This includes
for the five cases
Ex.
all
/j
= 0, A= + 1, h=
\a' /3'j
= 2.
Complex multiplication
arises
2.
iv.
"''J:
see
= Vl
1j
o) [o s)
Vo ij (l
is
the
functions 6(^i{w; iV2), &i,i{i''; t'V^)- The actual form of this polynomial is calculable
by the formulae of Chap. XXI. ( 366, 372), by applying in order the linear substitutions
(),(
and
sn[(l + tV2) TF] = (l+iV2)sn H'[l-sn2 W/an'y]l[l-k'^an^ W.sn^y],
where
t.
I.
(1881), p.
379
=jre^, and
elliptic functions
i.,
p.
may
be consulted
Sohnke,
Crelle,
xvi.
491
t.
iK =tK.
the following
(1888); Weber,
.4 c<a
Math.
xi. (1887),
Math. Annal.
(18823), Quart.
Halphen, Liouville,
iv., v.
Journal, xxil. (1887), Proc. Lond. Math. Soc. xix. (1888), xxi. (1890)
Etc.
384.
of
it,
We
come now
surface.
Riemann
For leferences
iv., p.
647.
The theory is considered from the point of view of the theory of functions by Hurwitz, Math.
Annal. xxvm. (1887), p. 561; Math. Annal. xxxii. (1888), p. 290; Math. Annal. xli. (1893),
p.
403.
p.
67.
For
(1,
ii.
1)
and
Klein, Ueber
640
To any point z
[384
all
the points,
called a coincidence
it, is
z.
x^-^,
Such a
A point of
Zn-2 correspond-
z^
of the curve.
and denote by Ci, ..., c,. the positions of ^i, ..., Zr when x is at the
and by aj, ..., a, the positions of x^, ..., Xs when z is at
the particular place c denoting linearly independent Riemann normal integrals of the first kind by i^i, ..., Vp, consider the sum
x-^,
..., Xs',
particular place a,
v7-
as a function of
since
it is
'
+<
,&> <v
we
necessarily finite
clearly
have equations of
the form
Xt
Mi^lVi
where
X,
+Mi^j,Vp
Zif
if
If
now X
+Vi'
'
(t
= tl,...,p),
\
is self-deter-
Chap. VIII.
Vi
Zr, Cr
On
minative
til
=Vi
sign
= can be
replaced by
the sign of
equality
(cf.
153, 158).
describe the kth period loop of the second kind, from the right
to the left side of the kth period loop of the first kind, the places Zj, ..., Zr
will describe corresponding curves
since,
Vi''''
its original
'
position
some
order, the
Mi,ic
oLi^k
+ Tf,! a'i,i +
in
+3fi,pTj,,t
similarly
+n,p a'p.k
/3i,i-(-Ti,,,S',,A:+
+Tj,p^'p,i;
FOR AN
385]
we have
(r,
M=a +
wherein
a, a', yS,
t)^
is
Mr = ^ + T^,
ra',
wherein
641
CORRESPONDENCE.
s)
x,
the integral
d[n2^,<^+
^Zr,Cr],
z, c
independent of
becomes
x,
and
Tlz[^c,
of integration
when any one of the places Zi, ..., Zr coincides with z, or, in
when z is among the places corresponding to x, and this happens
infinite
other words,
when X
The subject
The sum
round the
..., Xg,
when the
a
integral
is
taken, in regard to a
is,
^'"(^^' +
This quantity
is
Riemann
if
x described the
Consider
surface.
direction,
C">
second kind
left side
left side
taken round
the sake of
of the
is
for
paths described by
first,
Zi,
...,Zr
ox, t
first
would, in the
negative circuits of
j,
first
kind (X
1,
is
to
- 2iriTm, k 2
similarly the contribution as
the
first
kind
iv'^';
x describes the
is
2-iriEm,
a'x,
k^
0>.,kV^
a,
dissected surface connecting a^
that simultaneous positions on these paths are simaltaneous posi-
respectively to x,,
tions of I,,
B.
...
a;,.
...,
x,, bo
Cf.
Chap. IX.
165.
41
642
where E^,
m = k, and
unless
a,
v,n
x, as
\ is
Taking therefore
1-
we
1, ...,p,
z,
/^^
i^^, x
Nm,
k=
=Z^K,mi>K
?;
where
so that
is,
E^, m
-i^
= /8'x, m - S
^
i
all
the period
obtain
z
T,, i
t, ^
[385
^,
fc
z,
nT-
=2,iVTO,xfx
t;;^
a'x, t
M=a +
Mt = ^ + ryS'
to',
give
-^ + Ta = (/8'-To')T,
we have
also
Nt = - I3 +
These equations express the sum d^'
in a
^,
of the matrix
The theory
386.
+ ... + Vm'
terms of integrals v/
in
manner analogous
"'
Ta.
'
^^ terms of integrals
taken for
Vi'
...+
fj
_,
of correspondence of points of a
Riemann
'
v^.
M or the matrix
JV^,
surface
now
by elimination,
namely,
TUT + aT-T^'-^=Q,
is
matrices
a, y8,
a, ff,
or,
t, in
is
We
take the
the equation
it is
t.
is satisfied if
= -yE,
/9
= 0,
o'
= 0,
0'
= -yE,
where 7 is any single integer, and E is the matrix unity of p rows and
columns conversely, if the equations are to hold independently of the value
of T, we must have the n' equations
;
1...P
2
i.i
o'i,_,Tm,iTx,j
= 0,
1
2om,iTi,x=
=1
t,_,/8'j,x,
j"=l
t,
these give
y3m,x
= 0,
(m,
=1
pX
386]
With
these
equations of
we have,
values
643
as
385,
Vi
I'm
+v
=0,
+yVi
Vi
{t,m=l
+yvm =0.
of the
left side
p).
of these
firat
equivalences be
91
where
^r,
...,
z; a, c)
<f>{x,
wherein
=e
when x
are integers.
z^.cr
x.a
ov p
'
>
and
at a,
is
+9'pn,p,
''"'
Zr are
...,
Zi,
positions
..., g'p
gp, g^,
+ 91^1,1 +
z,
c,,
c, their
...,
In virtue of the
place c
it
is
a rational function of
Chap. VIII.,
(cf.
form
+Ky;')^
e2iri[K,vt'+
Kp
where Ki,
...,
infinities
are integers;
this quotient, as
it
z,
when
is
.2
a function of
at c
but poles
rational in a.
with
that
z,
We
is,
is
it
which can
construct
a function
fixed places a,
c,
result.
<}>
(x,
z; a,
c),
oi
exist
it
has no
hence ^(x, z;
z,
is
a, c),
can have no
it is
similarly
z^, ..., Zr
coincides
(r, s)
correspondence
surface, it is possible to
and
the
...,
Xa,
which correspond
to z,
and vanishes
to
order y (if y be
positive), at the
and
of
c
is infinite to
z,
Ex.
On a
i.
An
is at c
z^,
...,
which, as a function
Zr,
bj taking
is
negative.
a
which corresponds to
fact obtain
(1, 4)
x,
y=2,
412
644
X meets the curve
again,
and
z.
[386
which correspond to
The value y = 2
is
z,
obtained
=0
+2v
as representative of the fact that a straight line meets the curve twice at
AJ'A,
when
a, c
z,
AM.
x vanishes
and to
z,
and
is infinite
to the
first
first
which corresponds to
Cj,
a,
and
infinite to the
second order at
Ex. ii. More generally for any plane curve of order n, and deficiency p, if
we make correspond the r=n-2 points Zj, ..., z_2, in which the tangent at
curve again, and to a point z the s=27i + 2j9 4 points of contact x^,
drawn
to the curve
points
;rj,
...,
from
same
Xg will be the
If
...,
to a point
u,
meets the
x, of tangents
instance,
for all
a.
k of
s)
the
corre-
^^--U,.J-i^,'
regarded as a function of x, for fixed positions of z, a, c (of which a and c are not to be
multiple points), has for zeros the places x^, ..., x z\ for poles the places a,, ..., a,, c\
and regarded as a function of z, has for zeros the places z^, ..., Zr, x^, and for poles the
places
aK
Ex. iii. If fi-om a point x a tangent be drawn to a plane curve, and the corresponding
points be the points other than the point of contact, in which the tangent meets the curve
again, we have
where z* is the point of contact of one of the tangents drawn from x, there being as many
such equations as tangents to the curve from x since the 2n + 2p-4 points 2' lie on the
;
first
polar of x,
it
follows
therefore
2^""'
so that
y=2+2p-8.
As a
W'-''' + (2n+2p-8)if'-'' = 0,
{n Z){2n+2p-4) zeros
For instance
+ 2p-8)th
order.
387]
function of
vanishes to the
z,
sixth order at
We can
of X.
first
its infinities
up
build
645
of
Abel's theorem in the determination of the value of y ; for a fixed position of x, let T(z) =
be the equation, in the variable 2, for the ten tangents to the quartic drawn from z let
;
the quotient
T{z)IP^{z)
when z
10-2(2) = 6; let
vanishes
the function of
is
at the places
z^,
...,
Zi^,
what
T{z), P{z)
2
T{.z)
r(2)
P^{z)i PJ(z)
has the same behaviour as has the function <f){x, z a, c) as a function of 2. From this
function, by multiplication by a factor involving x but independent of 2, we can form a
a, c).
In fact,
this will be the function <f>{x, 2
symmetrical expression in x and 2
denoting the equation of the quartic curve by A/=0, and expressing the fact that the Une
joining the point x of the curve to the point f not on the curve should touch the curve,
viz., by equating to zero the discriminant in X of {Ax + \A()* Ax*, we obtain an equation
;
of the form
4/ [(', ^] =
AxA.^=0
is
the
Replacing f by
x.
first
A^il
polar of
2,
a point on
;c,
A/ A.
hence
,p(x,z, a,
c)
|.g
Ex.
tj){x, z
iv.
;
If a
a, c), is
387.
1)
A problem
of positions of
correspond to
(1,
of order
simple
a;,
(^_^2^^2)2
_ iQA^A.^
is
way
denoted by
to determine the
in
it.
x,
7+ 1-
number
to determine this
when
a;,
is
a,c)^ lim
[<^ {x,
which
putting
(a;
number
a, c)/{x
z)^
- zf],
z=x
and bearing
in
mind that
if
from the properties of the function <f>{x, z; a, c), on the hypothesis that none
of the branch places of the surface are at infinity, the following result the
rational function of x denoted by <j>{x; a, c) vanishes to the first order at
every place x of the surface at which x coincides with one of the places
;
(546
Zy, ..., Zr
which correspond to
it,
first
and
and at each of the places
coincidences by
a,
....
G we
a,
c,
each of the n
2-/ at
is infinite to order
[387
Cj, ..., c,
have
G+2my = (2n+2p-2)y+2y+r+s,
so that*
C = r + s + 2py.
The same result is obtained when there are branch places at infinity.
The argument has assumed 7 to be positive a similar argument, when 7 is
;
same
The number,
i.
given (Ex.
ii.
i,
result.
n and
deficiency
is
386) by
number
the
is
In determining h
it
for instance in the case of a curve with cusps all these cusps
formula
is
will contain
the
which correspond to
Xi, ..., x^
a term 2k
z.
have been
and
k cusps, the
weU-known one
iK=3(m-n),
where
m is the
Ex.
ii.
ra
(n 1)
- 28 3<c.
<t>{x; a, c)
{A,"-^A./{x-zf.A,-^A,.A^''-^A.},,^,
where A^''=0=A,''=Aar=A,''.
Ex.
iii.
The number
when
iii.
(Of.
Ex.
ii.
386.)
386, if
we
p may
P and
be
Q, will
/" is
of double tangents,
we have
2T + A = 2(ra-3)(2w +
where
<r=n+p 3.
For instance
for
number
of those
improper double tangents which are constituted by the tangents to the curve from the
cusps and the lines joining the cusps in pairs. The number of tangents, t, from a cusp is
given (cf. 9, Chap. I., Ex.) by
t+K-\=2{n-2)+2p-2,
or
= 2n-b-K + 2p=n^-n-3-28-3K.
There will not arise any such contribution corresponding to a double point, since the two
*
first
given by Cayley
Ex.
iv.
below.
^^'^]
we have
and therefore
647
(cf.
2,
Chap.
I.)
hence
r=2<r(cr +
p and
o-,
l)-4p-<-J(K2-K);
we find the ordinary formula
t,
equivalent to
r = 8+|(m-7i)(m + -9),
where
iv.
The points of contact of the double tangents of a quartic curve Jj;* = lie
upon a curve whose equation is obtainable by determining the Umit, when z=x, oi the
Ex.
expression
[9
treatises
the
be given here ; Clebsch-Lindemann-Benoist, Zefons sur la G^om^trie (Paris, 18791883), t. i. p. 261, t. ii. p. 146, t. in.
Chasles, Compt. Rendus,
p. 76
t. Lvni. (1864) ; Chasles, Compt. Jiendus, t. lxii.
(1866), p. 584 ; Cayley, Compt. Rendus,
t. LXII. (1866), p. 586, and London Math. Soc. Proc.
t. i. (18656), and Phil.
Trans.
CLViii. (1868) (or Coll. Works, v. 542 ; vi. 9
vi. 263)
Brill, Math. Annal. t. vi. (1873),
and t. vn. (1874). See also Lindemann, Crelle, txxxiv. (1878); Bobek, Sitzher. d. Wiener
Akad., xcill. (ii. Abth.), (1886), p. 899 ; Brill, Math. Annal. xxxi. (1887), xxxvi.
following references
may
(1890)
Castehiuovo, Rend. Ace. d. Lincei, 1889; Zeuthen, Math. Annal. XL. (1892), and the
references there given.
Ex.
V.
If
we use
187)
R (X, a)=lim^
- 1)>
rows of
integers,
and
^H^-^\^)^n^-^\n)
9{x, z;
c=o
= _ l)y
a, c)
[2e' (iQ)
Jit
l<l>{x,z; a,
we have
e(v'.'+ia) e'^'^^-'^lR
{X,
which
is
The
correspond to
position
a)fy
a,
and when x
of the place c
is
it
(i),
when x
e^<<.".
+nt;X-^^9'v'-
")
^
p. 427.
is
at
It is infinite
times,
z,, ..., z,
Ex. vL
a
Zr, Cr
648
where
^j, ..., z,
ing to
a.
Ci, ..., c,
[387
vii.
If there be
to
say
say
latter,
we have
Xi, will
therefore an (/s,
first
correspondence,
rs')
{z, ').
In
"'
ir^'
Al'''i.l
"'
+V'''+v'i.'r''i.^
''
+y'i/-
0,
+ yV'i''''i = 0,
(i=
1, ...,s'),
we have
Hence* we can make the inference. If upon the same Riemann surface there he two
an {r, s) correspondence of places x, z, and an (/, ') correspondence of places
the
number
then
3/, ^,
of common corresponding pairs of these two correspondences, for which
both X, x' coincide, and also z and /, is
correspondences,
r's
We
388.
have so
We
surface.
give
on any Riemann
exist
+ ri 'iyy'p.
more particularly
(1,
1) correspondences.
,p)
v-
t^-'""
'=Mi,,v,- "
?;r'"
M=a
= i\r.y- +
T<x',
Mr=
/8 -I-tj8';
hence
svi
=Mi_,
D,
m=l
= I
+Mi^p S
Vp
ro=l
Mi,,(N,,,v\-''
+ N,,p/;-')
k=l
iz,c +
t,!?;!
J.
z,
+Li,pVp
e
,
389]
CONDITIONS FOR A
where if
, is
the
therefore equal to
MN =M(
(i,
CORRESPONDENCE.
1)
(1,
= MN.
649
This matrix
is
Now
.9.
(a
),
if
it
follows
MN-s = H-s + rB =
that T2B
= 0;
..., rip
must have
B = 0;
hence also
a^ = ^a,
a'y8'
H = s and .4=0;
= y3'a',
not zero
is
We
are identical
ff.).
the matrices
389.
is
or
for
r^ri'
hence we
t.
I)
(1,
\ = must
|iV \| = 0, must be
ilf
A - \ = 0,
|
correspondence*
may
exist,
A=
where
They must
o/j-
there-
(1, 1)
when
A=
^1 c
we introduce other
integrals of the
Vi
and therefore
Supposing
is satisfied, if
in
,p)
x,a,
=Mi,iVi
V-
first
#
+ Mi,pVj,
.
Xt<^
,
=Ci, iV,
+Ci,pVp
=Ci,it',
+Ci.p%
...
Vp", where
also
i
* The
= l is considered in an elementary way
(1, 1) correspondence for the case j)
The reader may prefer to consult that Article before reading the general investigation.
in 394.
PERIODICITY OF A
650
(1,
1)
CORRESPONDENCE.
[389
\M-X\=0,
equations
+Ci^pMp,r
Ci,iil/,,r+
= fHCi,r,
(r
...,p);
1,
be
will
Hence
such
as X;
is
a root of unity,
correspondence
(1, 1)
corresponding to x be
is
z^,
it
Z2,
on,
then after a
coincides with
finite
number
that every
if
is,
z^,
of
be
x,
In order to prove
x.
that
the place
be
p>l,
suppose that
this,
of x,
z^,
and so
z,,
z^,...
Zi, z^,
all
,dF/*'"
= \,dFr''*-';
these give
and therefore
-dvi
Ci,i\dVi
]+
-dvj,
+Ci^p[dVp
0;
where
result
-"r
=K + \rr,i+
Thus
is
by
is
is
z,,
which
p >1,
of the
Co
Zi,
Ct
the
+V''r),
as zeros
and
Zo, c^
now
as poles;
= Co,
= l,...,p),
which
either
x;.-n:'r,,-2"(V''"''+
which
and
(r
we wish
+\rr,p,
but,
on a hyperelliptic
is
surface,
the other, and of the two zeros either determines the other;
it
C;fc
Co,
which
proves
is
is
the
not
fixed,
result
enunciated.
There
is
known
|i/'-X|
is
no need
= 0,
ii.
p.
486) that
It
= 0; when
the roots
390]
X,, Xj,
dv''-
...
satisfies
'''=Mdv'"\
we can
etc.,
it
infer
d/*'
''''
651
= di/''"
"',
proof of the periodicity of the (1, 1) correspondence, following different lines, and
not assuming that the roots of the equation \M-X\=0 are different, is given by Hurwitz,
Maih, Annal. xxxii. (1888), p. 295, for the cases when p>l. It will be seen below that
p=\
shew that
k be the index of the periodicity, and let <o = e*"/*; let S, T be any
two rational functions on the surface, and let the values of S at the
successive places x, z^, z^, ..., z^^, x which arise by the correspondence be
denoted by 8, Sj, ..., S^-i, S, and similarly for T; then the values of the
let
functions
= S+a)-'/Sf, +
T, +
t = T+
+ w-'*-"-S^,
r,_,
+
Sr=Sr +
hence
it
can be inferred
relation connecting
s*
+ G)-'*^'' (Sr+t-i =
(o-'Sr+i
(cf.
and t
Chap.
(o^'s,
4) that there
I.,
S and T
Conversely
and
exists
a rational
generality that
the original
Such a surface
is
is
a multiple of
k.
The
(a)
(/3)
o)S,
='t.
may
be mentioned *
The index
When k>2p
1).
2 the
Riemann
the
Riemann
(7)
When
A;
> 4p
4,
= <*( -
less
1)*'.
than
k.
652
[391
We
can deduce from 389 that in the case of a (1, 1) correspondence the number of coincidences is not greater than 2p + 2. In the case of
a hyperelliptic surface, when the correspondence is that in which conjugate
391.
places
of the
when
For,
which
is
of order 2p
2,
z^,
which corresponds
1 at
the
number
of
is
and to order
a.t
+2
number
pairs,
2p
is
4.
let g'
to Zq
the surface
is infinite
number
of coincidences
is
now
thus
In the case of a
2.
hyperelliptic surface
2/'
we may
similarly consider
We
3/'
= y,
x'
x,
= x, for
be greater than
It
1)2^-1.
the function x
= (a'.
4,
of order
which x
number
(r, s)
unless
the
identically
of coincidences cannot
x is
y'
= y,
x'
= x.
equations
^"'''
+tf *=J/i,i<'''+
+i^i.,"^'",
/d^K\%^
to the value obtained for the integral taken
tion, that
(1,
1)
ai,
+j3'pp is the
sum
correspondence, in which
C^2p+2.
Forany(r,
formula gives
of the form
round the
+ ap-|-/3',i +
^-^^).
the 2p roots of
|A-\|=0
|A-X|=0,
it
follows for a
A = (~l'
), the
same
as already found.
( 386,
J,
any such
+n^\U
C=r+s+2py,
We have remarked
a matrix
all
Jfr=/3+r|3',
is
C=r+s-(ai,-iSince
ir=a + ra',
Ex.
iv.) for
the case of a
(1, 1)
+y.
For
number
i-r+K'^^O,
(i=l,...,p);
NUMBER OF
393]
(1, 1)
CORRESPONDENCES
IS
653
LIMITED.
conversely, for 7 = 1, equations of this form are known to hold for any hyperelliptic surface,
associated with the correspondence of the conjugate places of the surface.
From the
considerations here given, it follows for p>\ that for a (1, 1) correspondence the number
>2p+2.
392.
In conclusion it is to be remarked that on any Riemann surface
which p >1, there cannot be an infinite number of (1, 1) correspondences.
For consider the places of the Riemann surface that can be the poles of
for
<(^+l) which
34
36,
such a
Chap.
(1, 1)
III.).
correspondence
is
( 28, 31,
^r-places.
As
of the first kind, which does not affect the zeros of the determinant A, of 31, it follows that the place corresponding to a ^r-place
integrals
must
^r-place
of coincidences in a
be
>2p + 2.
(1, 1)
Therefore,
number
for
surface
in
no case
not hjrperelliptic, a
is
1)
(1,
a permutation
rise to
correspondence,
when the
is
finite,
But the
we have shewn
result
is
number
of distinct ^f-places
is
2p
+ 2,
every
(1, 1)
since
4,
the
As
number of such permutations is finite, the number of (1, 1) correspondences must equally be finite.
It is proved by Hurwitz* that the number of (1, 1) correspondences,
when ^ > 1, cannot be greater than 84 (p l). In case p = 3, a surface is
known to exist having this number of (1, 1) correspondences "f.
this particular one
must give
rise to
the
393.
ingenuity.
let (z, s)
functions
It can
(x,
s)
y)
then
z,
ii.
s are
each rational
1. 1.,
1890, p. 701.
654
[393
functions
rational
z,
s.
x^\ ...,
= Q,
which x
for
a;^"'
and
a}^^
irS,"'
we
let
z-^ ,
which x
for
...,z'n
is infinite; as
a;
is
a rational
VIII. 154)
{m)
(0)
(<x>)
(0)
v*i'*i
=0,
-{-i,*>^n
-I-
{i=\, ...,p);
^r
z, c
=Mi,tVi
Vi
X,
Tir
+Mi^pVp
a
,
we have
<''''
+ v^-"- = 0.
{i=l,...,p);
there exists therefore (Chap. VIII., 158) a rational function having the
2r' as poles
2r as zeros, and the places z,",
regarding this as
^1
places
. .
.
is
z,
. ,
tf>
may
reader
Bd.
(Berlin, 1890), p.
II.
Riemann's Theorie
138
for
^>1
the
u.
s.
285
;
of
(1,
1)
is Umited the
Geiamm. Math. Abhand.,
correspondences
x/(f> {z, s)
p.
386
Noether, Math.
;
Klein, Ueber
Noether, Math. Annal., xxi. (1883),
(Berlin, 1895), p. 241
Hurwitz, Math. Annal.,
number
p.
a function of
II.
394.
In regard to the (1, 1) correspondence for the case p = l, some remarks may be
made. The caae/) = needs no consideration here any (1, 1) correspondence is expressible
by an equation of the form
;
Atf+Bt+Cf + D=0;
thus there exists a triply infinite number of
In case
p = l,
corresponds to
shewn,
if v^- "
correspondences.
(1, 1)
of,
if
there be a
and
a, a'
the form
wherein
ji
is
we deduce equations
fi
where
a,
= a+Ta',
From
this equation,
by supposing x
^r=^+T/3',
By supposing
a, ^, j? are integers.
x.
of the form
x'
(i),
we deduce
where
y,
y,
8, 8'
are integers.
= f'(y+T7'),
The
r=ft(84-r8'),
(ii),
a', /3,
^ is
394]
if
now yy-ys=0,
IS
we may
and y are
zero,
we deduce
is
inconsistent with
1 =/i (7
77'), or else
in that case the deficiency of the surface is not 1
therefore exclude that case if yS'-y'S be not zero, we have
a rational fraction
but
-ry' = fi(y8'-y'd),
either y'
however interesting to
is
(ii)
655
CASE.
it is
known that
''=y-8W8='"^"'' ''*=y^^d=^+^^''
we have
= o>
..y_./;> =/^'
:7=^>
and therefore
= (aff-a'P)(,yS'-y'8);
thus
a^ a'^=yS' y'8= + l
S'
(ii)
or
= fa,
let
y = ^(,
is,
to the equations
(i)
we deduce
if a'
be
/3/(a
/S"), the
fi,
=a,
is
an
equations corresponding to ;=
to
if next,
But
a'=j3=0,
(i)
/i=
= ^,a
and
they differ
the equations
=e
= +1
and
^= + 1
or /i=
the
are
V*''^'
0/
from
aj3'-a'/3
existing correspondences
y being +1*.
whenp>\,
(i)
second of equations
+1 and
1!x;a'
do belong
=0;
/3
H = a=ff,
coefficient
then we deduce
these
in turn
a-n
now
value
;
/3+T/3'=fir,
T2a'+r(a-/3')-/3=0,
/i is
-/3t
(i).
so that
5=
lead to
a+Ta'=ii.,
that
common
denote their
y'=-a't,
+ V'<^ =
the
the (1,
2ra'= -(a-^')v'(a+^?-4f,
SO that, as r cannot be
real,
we must have
(a+j3')2-4f<0,
* For instance, on a plane cubic curve, the former equation is that in which to point of
argument u we make correspond the point of argument u + constant ; the line joining these two
points envelopes a curve of the sixth class, which in case the difference of arguments be a
half-period becomes the Cayleyan, doubled ; while the latter equation is that in which we
make correspond
656
and
this
shews
= l.
[394
precisely
the same form as those already considered for the special transformation of theta functions
(
383)
(1, 1)
and the
may
(1, 1)
there exists
-.r,
j/=iy
2iir
\ij oi
e^
x,
y'y
is
2 in the
Ex.
Consider the
(1, 2)
shew that a
(1, 2)
correspondence
.
is
given
(cf.
Ex.
ii.
iV2
383) by
.r2-22:-|,
jo
=1
in a similar way.
395]
CHAPTER
XXII.
The
p = 2,
the case
for
= x+
in fact, putting ^
{x
'JkX)
kKJx,
first
dx
is
we immediately
verify that
df
^x{x-\){x-k){x-\){x-kK)
396.
kind
k\),
'J(^+2^/kX)(^-1-k\)(^-k-\)
for
any value of
there exists an
C = XiWi +
wherein
\,Up,
u^, ...,Up
first
kind, which
is
/.
R (I)
functions on
f,
Riemann
the original
surface;
= 6/n-a/n',
wherein
a,-, bi,
a/,
hi
\Ti,i
{i=i,...,p),
+Xj,Ti,p=-bin+ain,',
are integers
we may suppose
fi' to
be chosen so that
the 2p integers
Oi, ..., ttp, Oi', ...,
have no
common
Oi6i'-l-
factor
aA' +
Up
and so that
- ap'hp = r,
42
TRANSCENDENTAL CONDITIONS
658
where
Eliminating
a positive integer; we assume that r is not zero.
..., Xp, and putting m =1'/^, we have the p equations
is
r*
[396
the quantities \,
bi+ 6i'Ti,i+
if therefore
...
+hp'Ti,p
= a>(ai + ai'Ti^i+
A=
...
p/j
of
( 420,
Appendix
= a'a,
18^'
then
II.),
= ^'^,
and the
he determined
in order,
..., Op'
in order,
(i=l, ..-.p);
+apTi^p),
o/3'-o';S
r,
= r,
it
for
that
t'ii
= a),
t'i2
= 0,
Tip
+ Ta')T' = /8 + Ty3',
to see this
compare the
it is sufficient to
elements of the
first
(^1)
variables,
Wj, ...,Wp.
It can
397.
it is
possible by a
= 0,
ax
ax = 1,
bx'
x,
b'x = 0,
such that the 2p elements of rx b, rx' b' have unity as their greatest
common factor, a denoting the row Oj, ...,ap, etc., and suppose ( 420) a
matrix of integers, of 2p rows and columns,
\7
bj
\a,rx b,
\x
...
...J
77=77.
wherein the
column
column
This theorem
evski, Acta
first
Math.
The proof
is
7S'-7S = 1,
BB'=^8,
and
iv. (1884), p.
395.
II.,
d'.
rx'
b',
Kowal-
(1881), p. 519.
419.
It
may
be necessary, before-
398]
column
afifect
o (ra;'
and x
ax
-dx=\,
{rx-h)oo -{rx
for
(7
leads to
t's,
Ki +
= 0,
. .
.
(T7')i.i] t".,i
t"p,
+ t7')t" =
only the
if
-h)x = ^,
a'h = r.
equations
(i
matrices 8
+ t8',
(7
Then the
+ tS',
[7.-,2
659
satisfied
these
first
p),
columns
equations are
[a
t",,i
= to/r,
= 1/r
t'^,,
we
for
398.
(a
+ ra') = b+ rb'.
p=2
<t",
/^
n, l/r\
U/r, t"J
hence
it
'
immediately follows that beside the integral of the first kind already
is expressible as an elliptic integral, there is another
considered, which
form
U=(b'-rT\,a')u,
where u denotes the row of 2
integrals
iCj,
r=[rt'-a'- rr\
U2; consider
{rx'
now the
integral
- b')] u,
where t' is a row of two elements, these being the constituents of the first
column of the matrix 8' the periods of V at the first set of period loops are
given by the row of quantities
;
rH
*
t.
XI.
(18823),
de France,
t.
xil.
p. 25,
and
(1883
Tv. (1884), p.
400
4), p.
124
viii.
(1886), p. 289.
422
660
(i
(i
(7
- a')]i - rr'V
Fat
- b%,
[t {ri
('^
1,
[398
[r (rx'
= 1,
2)
gives
= l,
2),
where
is
[ffl
matrix S;
therefore, as T'\^,
the periods of
l/r,
first
column of the
may be
399.
It has
{i
= 1,
2),
it
V is reducible
last
periods t there exist transformations of the theta functions into theta func-
A=
^=
^,
,),
exists, associated
same period
t'
then
</>
with a matrix
is
it is
transformed
easy to prove
whereby it becomes changed into a theta function with the same period
namely the transformation is that associated with the matrix
W/S'JUVA-a
\f' 9')
to prove this
it is
a)
{\
(a + ra') t'
/3
+ t/3'
{f+rf')T=g +
* Wiltheiss,
Tg'.
p. 127.
S^
t,
401]
Hence
a-
it
661
400.
Thus for the case p = 2, supposing r = 2, the original function ^ is
transformed into a theta function with unaltered period t, by means
of the
transformation of order 4 associated with the matrix,
C^)(::)(-?;i). -^-y.
where
m denotes the
matrix {~
j
the matrix
is
it
easy to see that this transformation of order 4 is equivalent to a multiplication, with multiplier 2, together with a linear transformation associated
is
A^(iV).
We
when, in case p = 2, there exists a transformawhereby the periods t are changed into periods t' for
which t'i_ 2 = 0, then there exists a linear transformation whereby the periods
T are changed into the same periods t, or what we have called in the last
chapter a complex multiplication.
401.
The transcendental
first
kind which
is
Thus
for instance
when p =
2,
A=f
an
,
elliptic integral
^,j,
it
is
which an
by means of a
sufficient to consider
may
for
when
t',_2
second order,
it
Chap. XX.
^(
By means
up
jj
into
of the formulae*
t'
364.
p. 77.
[401
ALGEBRAIC CONDITIONS.
662
may
algebraic equation
y''
= X {x
1) (x
k) (x \) {x k\),
by Jacobi
is
2,
_p
2,
_ A^ + B
'^'G^-A'
and
which
is
x
the quantities a,
h,
= ^'i^k{z + \)Hz-\),
being respectively
(V\-^-l)V(\//a-l)^ (V
1,
Similarly for
+ \/\)V(Vk - V\).
^ = 3, when
an
first
0.
^
* Cf.
taken at
Ex.
4
= -xl{\+Fx+Gy),
V. p. 341.
= 0,
1, , A,
/i,
By means
wherein,
r,
= -yl{\+Fx + Gy).
of the Bubetitution
ii c^,
Mem.
p. divers
REFERENCES.
403]
663
But the problem of determining the algebraic equations for which an associated
an elliptic integral may be considered algebraically, by
beginning with an elliptic integral and transforming it into an Abehan integral. The
402.
may consult Richelot, Crdle, xvi. (1837); Malet, Crelle, lxxvi. (1873), p. 97;
Brioschi, Compt. Rendu*, lxxxv. (1877), p. 708; Goiirsat, Bvlletin de la Soc. Math, de
reader
Fraiwe,
t.
Math. Soc.
vol.
xxin. (1892),
c.
(1885), p. 622
p. 173.
403.
The paper of Konigsberger already referred to (Crelle, Lxvii.) deals with the case
a transformation of the second order, for/) = 2. For the case of a transformation of the
third order, when p = 2, consult, beside the papers of Goursat (loc. cit. 402), also
Hermite, Ann. de la Soc. Scient. de BrtuceUes, 1876, and Burkhardt, Math. Annal. xxxvi.
For the case p = 2, and a transformation of the fourth order, see Bolza,
(1890), p. 410.
Ueher die Reduction hyperelliptigcher Integrale u. s. w., Getting. Dissertation (Berlin,
of
Schade, 1885), or Sitzungsher. der Naturforsch. Ges. zu Freiburg (1885). The paper of
Kowalevski {Acta Math, iv.) deals with the case of a transformation of the second order for
See further the references given in this chapter, and Poincard, Compt. Rendus,
p = 3.
t.
(1883), p. 983.
(ii.
Abth.)
[404
APPENDIX
I.
On Algebraic Curves
in Space.
404.
Given an algebraic curve (C) in space, let a point be found, not on the curve,
such that the number of chords of the curve that pass through
is finite ; let the curve
be projected from
on to any arbitrary plane, into the plane curve (/), and referred to
homogeneous coordinates f, i;, r in that plane, whose triangle of reference has such a
position that the curve does not pass through the angular point ij, and has no multiple
points on the line t=0; let the curve (C) be referred to homogeneous coordinates f, ij, f, t
of which the vertex f of the tetrahedron of reference is at 0.
Putting x=^/t, y=n/r,
^
= f/r,
it is
and
its
sufficient to
think of
x, y, z as
infinity.
equation
is
of the
is
such that y
is
finite
being at
value of x,
to be the deficiency of (/) ; to any point (x, y) of (/) corresponds in general only one
point {x, y, z) of (C), and, on the curve (C), z is not infinite for any finite values of x, y.
Now
let
f {y)='if{y, x)fdy;
let <^
(/) at which
vanishes to the first order at least
of
{^-a)<f)/f'
(j/)
^'~y^\\
2 (c-yi)
e-!/i
where
c is
L/'
(y)J'=!'.-
89,
it
+ Mm),
yt for
c,
after division
by
a,
we
where
x and
yjf
<f>/f' (y).
'
'
p. 614).
404]
appears therefore
It
a curve in space
that
may
665
be regarded as arising as an
Riemann
surface
and, within a finite neighbourhood of any point of the curve in space, the coordinates
of the points of the curve
may
we have
Riemann
surface;
t,
to
More
And
in space precisely as for a plane curve, or can be deduced from the latter case
identity of the deficiency for the curve in space and the plane curve
may
the
be regarded as
Also we can deduce the theorem that, of the intersections with a curve in
all can be arbitrarily assigned, the number of those whose
positions are determined by the others being, for a surface of sufficiently high order, equal
a corollary.
Ex.
If
through
^1
+!
/>
y^={x,
l)2p + 2.
is
same system, a
representable by
2Ml=2.
{x, \)^ + ^\aa.n integral polynomial in x of order 2p + 2, and Mj, Mj are respectively
and quadric polynomials in x and y.
For the development of the theory consult, especially, Noether, Ahh. der Akad. zu
Halphen, Journ. cole Polyt., Cah. Lll. (1882),
Berlin vom Jahre 1882, pp. 1 to 120
where
linear
pp.
1200;
[405
APPENDIX
On
A
405.
II.
Matrices*.
SET of n quantities
(Xj, ..., ic)
is
often denoted
By
sum
the
by a
which
(or difference) of
The
406.
single letter x,
set of
x^]/^
+ x^^
n equations denoted by
^i=i,ili +
where n
may be
+ t,,^p,
(i
= l,
n)
matrix
we
call a^, ,
the
Oj, ,.
The sum
(i,
is called
(or difference) of
two
same number of rows and columns, is the matrix formed by adding (or
component matrices. Two matrices are
equal only when all their elements are equal
a matrix vanishes only when all its
elements are zero. If ^j
, X^ by the equations
^p be expressible by m. quantities Xj
matrices, of the
^r=6r.i^l +
so that
^=hX, where
is
= Ci,l^l +
m,
('=1,2,
+6r.m^m,
or
,p),
+Cj,m-J'm,
).
x=cX, where
Ci..=a,i6,..
The
ii.
The
+a^ph^.,
and
^^j,
following references
(1889), p. 475
[^^^
may
be given
Cayley, Grelle,
l.
(1855)
lxviii. (1868)
1874, p. 347.
408]
Cj, ,
being the
(t,
)th
may be
667
columns
from the
arises
it
x=abX
hence we may
formulate the rule
A matrix a may he multiplied into another matrix b provided the
number of columns of a he the same as the numher of rows of b the (i, a)th element of the
resuUiTig matrix is the result of m,ultiplying, in accordance with the rule given above, the
i-th row of A hy the s-th column of b.
Thus, for multiplication, matrices are not generally
commutative, but, as is easy to see, they are associative.
The matrix whose (i, )th element is c,, <, where c,_j is the {s, i)th element of any
matrix c of n rows and m columns, is called the transposed matrix of c, and may be
equations
result
written
denoted by
be matrices,
if a, h, c
ahc='&>a.
The matrices which most commonly occur are square matrices, having an equal
and columns. With such a matrix is associated a determinant, whose
407.
number
of rows
elements are the elements of the matrix. When the determinant of a matrix, a, of p rows
and columns, does not vanish, the p linear equations expressed by x=a^ enable us to
represent the quantities fj, ..., |p in terms of x^, ...,Xp; the result is written ^ = a-^x, and
o~i is called the inverse matrix of
a,
the
(i,
minor of
itself.
The
a,^
in
inverse of a
product of square matrices is obtained by taking the inverses of the factor matrices in
reverse order ; for example, if a, 6, c be square matrices, of the same number of rows and
columns, for each of which the determinant is not zero, we have
(a6c)-i = c-i6-%-i.
of a matrix is the transposed of its inverse
(a)-i
thus
= (^).
,
|
we
clearly
have a6
|
|
|
(i) If a
Finally, the following results are of frequent application in this volume
denotes
row
of
quantities,
the
symbol
a|
a
and
columns,
and
rows
|
n
of
p
p
matrix
be a
a row of n quantities ; if ij be a row of n quantities, the product of these two rows, or
When n=p this must be distinguished from the matrix
(af)(i,), is denoted by a|ij.
which would be denoted by a $i)thia latter never occurs. We have then
408.
and this
called
a bilinear form
we
p
I
also clearly
(ii) if
is
is
n
=1
5 be
;;
of
columns.
668
[409
409.
If a bilinear
i.
square matrix of
SaR=a.
Ex.
ii.
If
prove that
(a+A)a~i
hence shew that
{a
if
the substitutions
R=S
see
Taber, Amer. Joum., vol. xvi. (1894) and Proc. Lond. Math. Soc,
Ex.
give
iii.
The
Ex.
iv.
'10
,=,
The
0\
010
000
1
is
Of. also
when
of the matrix
x=^0,
y=Q.
/OOlOv
Ov
Ov
o),
I
-I
'
_/
j^^{
,
'n -1
^
o'
-1
1\
M^
00 1ly
\
ooo/'
,-
_/
j^^l
oo-io),
/
-10
o'
which
505.
-10
oioo), >,=(-!
analysis.
II. p.
xxvi. (1895).
and columns,
vol.
vol.
0'
equal to {n^+x^-irx^+x^f, vanishes, for real values of x, x^, x^, x^, only when
is zero.
(Frobeniiis, Crelle, lxxxiv. (1878), p. 62.)
In the course of this volume we are often concerned with matrices of 2p rows
Such a matrix may be represented in the form
and 2p columns.
=(: 3.
wherein
o, 6, c,
matrix given by
"=lo'
d')
if
/i'
be another such
*11]
the
(t,
this is the
sum
of the
jt)
(i,
+1
less
hence we
/a b\
may
than^ + 1
is
+ 6',,pey,r,
/a' b'\
when
+a'i,pap,T + h'i,iCi,r +
<i<i.r+
and
/I'/i)
669
b'c
similarly
when
and r
write
+ d'dj'
the law of formation for the product matrix being the same as
if a, b, c, d, a',
V, c\ d! were
single quantities.
Ex.
Ex.
iv.
~)
respectively
by
can be denoted by
'>(-'
=G?)' ^'=("03' '^=(-?J)'
~o)-
411. We proceed now to prove the proposition* assumed in 333, Chap. XVIII.
Eetaining the definitions of the matrices J^, B, C, D there given, and denoting
Z>-' respectively
by
a^,
b, c, d,
we
find
6=
REDUCTION OF GENERAL
670
be formed
[411
which are specified in the
in respects
following statements
(i)
ttj.
(p + l)th
also a,,
interchanges the
first
(ii)
adds the
(iii)
(iv)
d adds the
new
and k-ih columns (of A), and, at the same time, the
For the sake of uniformity we introduce
{l<k<p+l).
and
(/)
+ l)th
column.
first
first
first
(j}
+ 2)th
and
the (^ + l)th.
results
if
placed at the right side of any matrix A, of 2p rows and columns, so that the matrix
acts
lji=ajs?ai,
results
mentioned
will accrue
(j>+k)iJii
columns
(of A),
<i=ai6ai, interchanges the i-th and {p-\-k)\h columns (of A), giving the new k-th
column an opposite sign to that it had before its change of place,
i!^=aj^ai, interchanges the i-th and (p + k)ih columns, giving the
column a changed
adds the
mjc^ajt^ch'a,,,
/fc-th
column
to the {p + k)th.
= ajtfichai,
new (p + i)th
sign.
i-th.
g^^,=a/i^fl^dha^a,a^ri subtracts the -th column from the r-th, and, at the same
time, adds the (/> + r)th column to the (p+4)th.
g\^,=a^^fl.j)dl^a^fi,^r, adds the -th column to the r-th, and, at the same time,
subtracts the (^+r)th from the (p + a)th colunm.
/r, 8='e^r,
/'r,
/) adds the (p + r)ih and (^ + 3)th columns respectively to the -th and
r-th columns.
= 'S^r, /)
and
-th
r-th columns.
To this list we add the matrix a^, whose efiect has been described, and the matrix 6^,
which changes the sign both of the first and of the (/> + l)th colimins; then it is to be
shewn that a product, P, of positive integral powers of these matrices, can be chosen such
that, if A be any Abelian matrix of integers, given by
=(: where
a/3
a/3'-|3a'
= l.
aP is the matrix unity of which every element is zero except those in the
di^onal, each of which is 1. Hence it will follow that fi = P~^ ; namely that every such
Abelian matrix can be written as a product of positive integral powers of the matrices
the product
Ai, B,
C, D.
Up
we
to be
r.
In the matrices a^, a,, a,, each of k, r, is to be <.p-\-\ and in general each of
ia >1 ; but for the sake of uniformity it is convenient, as already stated, to
introduce a matrix ai = l then each of k, r, may have any positive value less than ^-(-1.
;
k,
r, s
ABELIAN MATRIX.
412]
412.
oj,
p,
the
first
/9j,
if
first
671
we begin with
Then,
AZp.
let
(i.e. if
from
y, is
a',
which, after
zero.
Then use
first
row of the
all
Let a similar process of alternate reduction and transposition be applied to A', until
the (1, 2/) l)th element of the resulting matrix is zero. And so on. Eventually we
arrive, in continuing the operation, at a matrix instead of A, in which there is a zero in
each of the places formerly occupied by ^j, ,
, ^l, p,
Now
apply the processes given by 6^, Ip, gi,p, S'p,!, and eventually Up, if necessary, to
reduce the (1, /')th element to zero. Then the processes 6^, Zp_i, ^i, p-i, gv-\,i, ^p-n as
and so on, till the places,
far as necessary, to reduce the (1, p l)th element to zero
;
matrix were occupied by a,,2, ..., oj.p, are all filled by zeros.
Consider now the second row of the modified matrix. Beginning with the (2, p)th and
Next
(2, 2/))th elements, use the specified processes to replace the latter by a zero.
replace, similarly, the (2, 2p-l)th element by a zero; and so on, finally replacing the
which
(2,
in the original
^ + 2)th
element by a
zero.
first
(2,
The necessary
..., (2,
(ii)
the
(r,
(1,
processes
wiU not
the
(r,
+^)th element
r<p and
and every s<p, in
r<p
which >r. In other words the matrix has a form which may be represented, taking p = 4,
by the matrix
p.
P=(
"11
REDUCTION OF GENERAL
672
[412
the value
and
is
r=l
being introduced
not zero
tmnsformatiou of order
later,
1st,
2nd,
...,
is
+1*
( 324,
a|3
r,
Ex. L)
and
/3a
we deduce
that in the matrix p the elements ^jd /831, j332, ... of the matrix ;3 which are on
the left side of the diagonal are also zero thus, in p, every element of the matrix ^ is zero.
;
Apply now
a^'-|3a'=r,
which
the matrix
fi'
o0 = r.
Then
it is
Oll^'ll =
The
may
left side
If now to the matrix o- we apply the processes given by the matrices ^j, 2 or ^j, 2 and l^,
we may suppose oji numerically less than 022, and 022 positive if then we apply the
processes given by the matrices g^^ 3 or jr',, 3 and l^, and the processes given by the matrices
9% 3 "' S^z, 3 ^"d Z3, we may suppose ajj, 032 numerically less than 033, and may suppose 033
to be positive.
Proceeding thus we may eventually suppose all the elements of any row of
;
when
and
is
that
all
a prime
Further we
a,
left
may suppose
number
may
We
ff.
'<
by means
or
left
of this diagonal,
by means
r.
^pp
the matrix
than
/^p,
we may suppose
a'pp to
o';
by means of the
or f'p^, we may suppose a'p,, to be numerically less than /y^p. Similarly by the
processes given by ip_, or 7i'p_, we may suppose a'p_i,
p_i numerically less than ^'p_i,p_i,
and by the processes /p_i_, or/'p_j, ,, where s<p l, we may suppose a'p_,, , numerically
/p,
than j3p_],p_i. The general result is that in every row of the matrix a we may
suppose the diagonal element, and the elements to the left of the diagonal, to be all
numerically less than the diagonal element of the same row of the matrix /y.
less
If then
413.
a product
may
we take the
of the
^+2
case
matrices Oi,
be represented, taking jb= 3, by
h, c,
d,
AO
0),
AQ = ( 10
1
1
a'i2 a'i3 1
a'23
0*12 ^"13
1
/3'23
wherein
zero,
all
and
all
415]
ABELIAN MATRIX.
matrix
673
the condition
a'/3'=j3'a',
equivalent to
a'
=5',
we have a'=0.
the matrix unity of 2p rows and columns, and A, =a-\ is expreised as a prodtuit of
p+2 matrices A^, B, C, >, as desired. Since the determinant
of each of the matrices A^, B, C, D is +1, the determinant of the Hnear matrix A ia also
+ 1.
414.
In the pari;icular case ^ = 1 the only matrices of the ^ + 2 matrices A^, B, C,
which are not nugatory are the two matrices B and C we denote these here by U and
and put further
M=?7~> = {
D=V-i=l
1,
1,
(-7,
,-;)
1)
(v,-i)
(-l,-'7)
first
{S,v-^)
(.i,v+^)
if
-'?,'?)
{$+1,
and columns, and we form the product Am, then the columns |,
new first column
t)
v)
of 2 rows
of the matrix
is
changed
are
we have
in fact
(a'/yji-l o) =
hence
it is
\-0
a')'
A=(
ff)
which the integers a, |3, o', ff satisfy the relation off -a'P=l, can be expressed as a
product of positive integral powers of the two matrices
for
-c -;) HI -;)
Combining the final result for the decomposition of a linear Abelian matrix with
415.
the results obtained for any Abehan matrix of order r we arrive at the following statement,
whereof the parts other than the one which has been formally proved may be deduced from
that one, or established independently
let
order r
integral
it is possible to find a linear matrix
powers of the (j>+2) matrices At, B, C, D, which wiU enable us to write A=AiQ, where Aj
chosen, of the four forms repreis an Abelian matrix of order r having any one, arbitrarily
then
by
sentable
^.-(fe^)'
and
A<
is
it is
also
^-(^^)- ^KSk)'
/'^P\.
^<^)
of the matrix
instance
r is a prime
number
is
l+r+r^+r*; forp=3,andr
43
674
a prime number,
it is 1
+ r + r= + Sr' + r* + r* + ?*
[415
XL. (1855), p. 253, Wiltheiss, Crdle, xcvi. (1884), pp. 21, 22, and the
of Krause, Die Transformation der Hyperdliptischen Functionen (Leipzig, 1886),
C<m.pt. Rendus,
t.
book
which deal with the case
(1878), p. 139,
jb
=2
= 3,
see
A=Q
xll
O', in
2', t. ix.
213 ; Eisenstein,
(1851), p. 192
Ser.
Crelle, xxvill.
Considering
416.
where
a linear matrix,
is
(cf.
(i)
the
determinant of the matrix B' is the same, save for sign, (ii) if ji be a row of p positive
integers each less than r (including zero), the rows determined by the condition,
-
5 = integral,
'/I
be such that
where
is
is
namely
vanish,
A,
a linear matrix
j'.4
= 0,
0\
/a
therefore
j'
|
|
if
/i
-p2 + o-jD\
/pj',
= +1
which
equal to -B'/x,
is
the determinant of
be written
fpA,pB+qB\
q'B'
J'
a'
is also
A = Q'A'o can
Vq', -pq+<r'pj- \0
\a' ffj
;
Now pq' = l
if
is
first result.
is
also a
integers,
=n
q'B'fi,
Also,
=q'm,
row of integers
since q'B'
is
the
matrix which, for the reduction A=Q'A'o, occupies the same place as that occupied, for the
reduction A = OAo, by the matrix B', the second result is also proved.
Considering any rectangular matrix whose constituents are integers, if all the
(Z + 1) rows and columns formed from this matrix are zero, but not all
determinants of I rows and columns, the matrix is said to be of rank I. The following
417.
determinants of
theorem
is
often of use,
and
is
XXII.
In
ordej- that
coefficients,
system,
of
should be capable
of being satisfied hy integer values of the variables, it is necessary and sufficient that the
rank I of, and the greatest common divisor of all determinants of order I which can be
formed from, the matrix of the coefficients of the variables in these equations, should be
unaltered when to this matrix is added the column formed by the constant terms in these
equations.
For the proof the reader may be referred to H. J. S. Smith, Phil. Trans, cli.
(1861), Art. 11, and to Frobenius, Crelle, lxxxvl (1879), pp. 1712.
418.
Consider a matrix of
+l
^1 bi
ki, t|, ei
419]
MATRICES OF INTEGERS.
675
be the greatest common divisor of the determinants formed from this matrix with
rows and columns; let A' be the greatest common divisor of the determinants
formed from this matrix with n rows and columns then, since every determinant of the
(to + l)th order may be written as a linear
aggregate of determinants of the -th order,
let
n+l
integral,
is
=M,
Then
say.
the
n+l
congruences
U(=aiX+biJf+
have just
A incmigruent
their highest
common
Also, if in the
sets
+i:j^+lit
or
>n+l)
in particular,
divisor,
integer
Frobenius,
and
x so
m(n + l)
when n=l,
determine
2m numbers
aj, bi
The theorem of
419.
m forms
the
loc. dt., p.
if the
(i=l, ...,),
+kiZ+lit+eiU,
common
And an
M)
(mod.
divisor.
Ui=aiX-\-hiy +
divisor;
+ eiU =
357, p.
589
it
( 418)
which
satisfy the
2p congruences
X6/+/iaj'=0, \bi+iiai=0
(mod. M),
(i=l, ...,p),
6iQ - a!ia'=.BiQi
AfQ^',
bi'Q
- ai'a'=Bi'Qi-Ai'Qj',
and the greatest common divisor of the p{2p-l) binary determinants AtBj-AjBi,
integers B^, ..., Bp';
AiB'-Aj'Bi, AiBj-Aj'Bi, which is equal to if, divides the 2p
integers next put Q2= J/iQi, a2' = ^i')
these
divisor
of
common
2p
greatest
the
is
thus
. .
haf-yx=0,ao
aar'-a'a;=l,
haf-h'x=0;
^ar/-b/+hr^i' have no
common
Putting
X=x+H,
676
Z'=^+h|', the
first
397
will consist of
[420
the elements of
(a, a'),
the (p + l)th column will consist of the elements of (b, b'), the second column will
consist of the elements of rX-b, rX' h'; and since these latter have unity as their
common
greatest
(p + 2)th and
all
other columns
420.
theorem is assumed in 396, which has an interest of its own If of an
Ahelian matrix of order r there he given the constituents of the first r columns, and also the
constituents of the {p + V)th, ..., (p+r)th columns {r<p), it is always possible to determine
Frobenius,
columns
{a X
\a' x^
);
the elements of the given columns will satisfy the relation ah' a'h=r ; it is required to
determine in order the second, the fifth, the third and the sixth columns ; the relations
arising from the equations
aa'-a'a = 0,
0/3'-0'3 = O,
a^ -a'fi = r
a3^a'x=Q'\
'-a'x=(i\
hxf^-h'x=o]
a'y=0
by'-b'y^O
xy' x'y=r
ay'
^'^'
(ii),
_^
au'
bt'-b't=0
bu'-b'u=0
xu' x'u'=0
yu' y'u=0
xf-x't=0
yf-y't=Ol
(iii),
tu'
now let
a'u=0\
af-a't=0\
(iv);
-tu = r)
y=ri-(Ab-Ba), y'=ri-(Ab'-Ba'),
satisfy equations
(ii)
constituents have no
next let
common
(t,
')
be a solution of equations
(iii)
in
which the 2p
Yx),
ff.,
221,
225
ff.,
Dersch 647
Dini 239
Dirichlet 246, 600
Bacharach 141
Bachmann 666
Baur 57, 112
Epstein 342
Euler 159
Fagnano 638
Biermann 663
Bobek 647
Forsyth
2, 3,
25, 29,
7, 9,
90, 114,
39,
233, 296,
198, 212,
122, 123,
144,
150,
327,
395,
421,
373,
Frahm 383
Fricke 639, 648
Frobenius 342, 387, 447, 474, 486, 491, 500,
516,
517,
521,
Briot 90
598,
628,
629,
630,
Broch 221
Burkhardt
674, 676
Burmide
Casorati 579
Gordan
296,
340, 342,
374,
387, 459,
295, 392,
168,
241, 244,
255,
544,
545,
556,
474,
183,
423, 448,
578, 623
Goursat 663
Grossman 137
Greenhill 639
Christoffel 666
142,
131,
288,
Frost 389
663
Cantor 239
230,
632,
392,
Darboux 666
Dedekind 57
de wTonquieres 137
168,
183,
423, 448,
Halphen
124,
165,
364,
639, 665
Hamburger 2
Hancock 296, 326
Harkness
2,
10,
14,
15,
25,
79,
678
Hamack
Pringsheim 445
222
Prym
296,
2,
342,
392, 477,
486,
511, 566,
600, 627
Hubert 665
Humbert 222, 255, 340, 486
Hurwitz 41, 392, 639, 648, 651, 653, 654
Riemann
1,
2,
9, 13, 45,
6,
Roch 29
Jacobi
Rosanes 666
Salmcm
Jiirgensen 221
11,
5, 6, 7,
39,
117,
124,
136,
144,
340,
343,
345,
Klein
9,
25,
156,
159,
169,
343,
383,
360,
373,
433,
378,
177, 342,
639,
648, 653,
Schepp 239
Schlafli 666
Schottky
283,
101,
32,
371,
360,
372, 373,
296,
387,
448,
461, 469,
654
Schubert 665
Kohn 387
Konigsberger 337, 342,
448,
459,
477,
529,
Schwarz
14,
654
Scott 473
Sylvester 136
Lagrange 230
Taber 668
Lindemann
Luroth 239
Toeplitz 383
Malet 663
Mathews 165
Minding 221
Valentin 101
Mittag-Leffier 202
MorUy
2, 10, 14,
Weber
56,
8,
460,
486,
270, 272,
373, 387,
392,
533,
559, 600,
615, 620,
538,
430,
Netlo 20, 90
Neumann
Weierstrass
531
134,
99,
101,
177,
195,
205, 231,
239, 242,
246,
301,
311, 317,
32,
93,
137,
142,
326, 339,
443,
474,
486,
292,
572,
577,
579,
594,
145,
149,
156,
165,
295,
390,
392,
573,
658, 666
White 165
654, 665
Picard
659
197,
Zeuthen 647
first
...
r^; periods
of second kind,
For periods,
r'', p. 15.
Qp, or
2, (2), ...,
p. 16,
kind, ^j,
P- 61.
..., <t>n-i<
Elementary integral
Also 0i,
of third kind,
...,
P^",
0p, p. 146.
p. 68.
(Canonical
periods
a;
^(ar,
...,
Hf
Also
of, 196.
z, Cj,
c,,),
p. 182,
and F*",
p. 291, are
This
is
first
function.
f(x,a;
i, c),
E (x,
E (X,
z),
z),
e(u,T;
Q, Q') or
^u, t
^j
or
e L,
or
^^
e(u;
(?,
Q')
_2e2TC(7i+Q')+i^(+Q7'+2(J(n+Q')_ p 248.
5(;
g') or
<2,
'\ = Se'+2*<+')+*<+W+3"Q(+Q'),
5(|
p. 283.
ft(")=^.log&(), p.287.
Ipi. i
()
--
See also
W{x,z;
or
(f,
Ki'
^^,(^^
for in (x,
(|).P-487.
A), p. 509.
(u),
But
P- 367.
(f, /*).
kind), p. 169.
Also
/ii
174.
|<?|,|(3,B|,
# (a, a
first
P- 288.
7), p.
p. 516.
K, K' + ,j), *,
ff.
z),
(x), p.
192.
in-
153
of surface into
Invariants,
653.
itself,
See
and Curves.
Abel's theorem,
functions, 397
of,
for factorial
of,
222.
of, 16,
integral
532
of,
352,
ff.;
functions,
transforma-
see Characteristics.
number
notation
for,
Weirstrass's
number
tables of
303, 305
p = 2,
syzygetic, azygetic,
543, 564;
380,
564
of radical
Gopel groups
ff.
surface, 83,
characteristics,
439,
by Elein, 159
p = 3,
AutomorpMc
Biemann
ff.
see Places.
Associated
Branch places,
ff.,
functions,
639, 660.
264, 487.
Birational transformation of a
face
150
Biemann
sur-
number
3, 7
of invariants in, 9, 144, 148,
of plane curves, 11
nomials, 142
152
by ^-poly-
for hyperelliptic
when p = l,
or
0,
Riemann
surface,
of,
256,
236,
ff.,
Biemann
a,
sur-
136, 213.
3,
ft.,
flexions of a plane
inflexions and
40
bitangents in connection with theory
;
129
coresidual
136
formation
see
of,
of,
deducible,
of half-integer charac-
by means of a fundamental
teristic
ff.
definition
396
of,
426
functions, 439,
automorphic
with
connection
In-
correspon-
Correspondence
see
and
trans-
Birational,
dence
algebraically
references, 14.
121,
theorems,
78
quartic in con-
curves,
Existence
tions, 105,
sets
Elementary
681
special
finite,
contact curves,
381 ; general form of Pliicker's equations for, 124 ; Weierstrass'a canonical equation for, 93, 103
Cayley's
399
poles,
408
integral,
ff.
function by
form of that
simplified
411
expression of factorial
means
of that integral,
theorem
for,
160,
ff.,
141
412.
see
ff.,
Factorial
439,
fi.;
Integral,
231.
tion, 292, 324, 333,
Deficiency of a
Biemann
DeOniag relation
516
see
^ func-
fac-
prime, 172,
radical,
Tbeta,
see
Theta
Transformation
functions,
and
Differential equations of
225,
ff.
denda
inversion
problem,
(p. xz).
Fundamental
ff.,
182.
105,
ff.,
ff.
characteristics,
B.,
55
Biemann
surface, 26,
529,
Gap theorem,
468
see
Addenda
(p. xx).
44
B.
682
p+
of
fanctions
Jaoobian
same
functions, 105, S.
of in-
Klein,
ff.,
and Transforma-
tion.
447
See Linearly.
123,
at
the
place
infinite
of
629.
Biemann
surface,
or
of
rational
27,
see
ff.;
Eesidne.
on a Biemann
torial; Biemann's,
15
all
127,
ff.,
normal elementary
kind, 22
629,
185,
163,
hyperelliptic, 195
surface, 135.
function,
Biemann
1, 2, 3.
on an algebraic curve,
Jacobian func-
tions, 594.
tion, 27;
189,
194;
formulae connect-
functions, 527.
Normal equation
for
Biemann
a,
surface, 83,
Normal
Number
dental.
functions,
of theta
fl. ;
of independent products of
/i
^-poly-
functions, 261
number
144,
p,
148,
nomials,
7;
the
3p-3
153;
the
for
contact
0-
ff.,
270;
by
2;
iP
in
the
algebraic
equation,
see
Constants.
Period loop, see Loops.
functions, 594.
the ^-places,
transformation of the
their expression
of theta functions of
9,
moduli,
588
by theta fanctions,
3^7
surfiMse.
S^
Kind,
liucisir
on a h^peieUiptio
gtxneial
trausiformsUioTi.
ff.
184.
tv5^.
1, j,
branch
places,
15>9,
Slsmann-Boch theorem,
the plaoers w,,
,..,
355:
11,.,
3t>C)
trauafonuatiou,
afVer
deterS63
for a Bieniann surface
mination of,
with assigned period
567
loc^is,
for a
ment
theorem
S(tDB<e.
130
of.
130:
of,
adjoint.
<^polynamials. 141
treat-
al^i^ehraic
13^:
131.
and
functions,
Sequent
br
of,
of places. ISa.
curve, 185.
expression of rational
al^giebraic inte^iTals
sets
Sets of places
tranformatioa
J.*:.
Polysomlal. grade
581,
li>7. -Jjio.
.Mv3.
134
for factorial
44. 183
functions, 405.
their gieo-
luetrical iDterpretatiou,S6S,
linear
Weier-
156:
ff.,
polynomials,
5S^
SSti.
68S
4.-i7.
Special cvineepondences on a
Biemann sur&ce.
64^.
see Carrea.
p^od loop.
PodtlTe dirvction of
539.
184, 147.
fonetious, SS7.
sin^
essential singu-
639.
a,
6S9,
66a
Supplementary transformaticsis of
larity. aCkS.
tbeta
function, 5S3.
Qaazttc
^10.
SyiysvUc
Sll. S90,
TaWea
436.616.
Kiemann
are
surfiee
37
pities.
Roeh
whos*
infinities
.n-.!y
Ssl.
91
fl.
of the
fundamental
tbew
of orJer
47.
s!cond oivler.
intii^^ rational
A\
T%ta foncUans
Biemann 's
Tei^ihse
from
347:
349;
446
number
<rf
B.
SOS; hyperer.iptic
eivrt-sji.or.
oC
period ptofwr-
ff..
ttano
con-
d;; termination
periodicity, 444
:.t^ of.
3,^1,
of,
3!>6,
'2>S.
ff.;
35;:.
?c7
371.
j7i5,
algehraiv-
684
390, 426
elliptic, 332,
Transformation
of
of
390, 426,
ff.
275, 824,
of,
not
283,
integers,
625
when
transformations, 552
551;
of,
p = 2,
Gopel relation
by means
of,
composition
transformations,
special
of,
629,
630, 660;
expression
see Gopel;
of Jacobian functions
for,
coefficients
supplementary
587;
444
654,
functions,
first order,
Biemann
152,
solution of inver-
means
theta
formation, 539;
sion problem by
(or
151,
for hyperelllptio,
594.
expression
of,
number
509,
510;
of linearly
every
2, of
p+2
characteristics
of
p + 1,
ff.,
93, 174;
tions, 629,
ff.,
276, 303.
34,
ff.
fundamental
182,
ft.;
hyperelliptic
theorem
case, 433.
prime
when
zeros
the function
j.
and
o.
ff.,
of
Biemann
theta
252.
integr&l, 444.
oambbidge: printed by
Laurent's theorem
is
189;
p. clay,
at
function,
-UHNEiiUNIVERSITYLIBRARY
AUG
8 1991
MATHEMATICSUBRARy