You are on page 1of 51

Quadratic Integers: Some Properties and History

By Roger Bilisoly, PhD


Department of Mathematical Sciences
Central Connecticut State University (CCSU)
New Britain, Connecticut
Presented at CCSU
October 1, 2010
|
What is a quadratic integer?
Quadratic integers are solutions of monic quadratics over . Let a, b .
Solve[x^2 - 2 a x + (a^2 + b^2) = 0, x]
Solve[x^2 - 2 a x + (a^2 - 2 b^2) = 0, x]
{{x a b], {x a b]]
x a 2 b, x a 2 b
Let d be square-free (that is, it has no square divisors.)
Then quadratic integers are the following quadratic extensions of :
{a + b d : a, b } = [ d ] if d 2 or 3 (mod 4)
{a + b 1 + d 2 : a, b } = [ 1 + d 2] if d 1 (mod 4)
Above Mathematica code shows [ 1 ] and [ 2 ].
See http://en.wikipedia.org/wiki/Quadratic_integer (accessed 9/2010.)
|
2 CCS U Quadrat ic Int e ge r S LIDES . nb
Quadratic integers are geometric lattices in
2
For example, [ 1 ] is a square lattice in .
max = 7;
line = Range[-max, max];
lattice = Table[I + n + line, {n, -max, max}] // Flatten;
ListPlot[Map[{Re[], Im[]} &, lattice], AspectRatio - 1]
6 4 2 2 4 6
6
4
2
2
4
6
|
CCS U Quadrat ic Int e ge r S LIDES . nb 3
Definition of a commutative ring with identity
For details, see a book on modern algebra such as
definition 1.1 of Chapter 3 of Hungerford (1980).
Example: (, +, *) is a commutative ring with identity because ! a, b, c
a + b (closure of addition)
(a + b) + c = a + (b + c) (addition is associative)
a + b = b + a (addition commutes)
! 0 s.t. a + 0 = 0 + a = a (unique two-sided additive identity)
!a ! -a s.t. a + (-a) = (-a) + a = 0 (existence of unique additive inverse)
a*b (closure of multiplication)
(a*b)*c = a*(b*c) (multiplication is associative)
a*b = b*a (multiplication commutes)
! 1 s.t. a*1 = 1*a = a (unique two-sided multiplicative identity)
a*(b + c) = a*b + a*c and (a + b)*c = a*c + b*c (distributive law)
Thomas Hungerford, Algebra, Springer, 1980.
|
4 CCS U Quadrat ic Int e ge r S LIDES . nb
Ideals of
Every ideal of consists of multiples of some integer, n.
Example. Multiples of 7 are an ideal in :
(7) = {..., -21, -14, -7, 0, 7, 14, 21, 28, ...} = {7 n: n } ,
In general, an ideal of a ring R satisfies the following:
I is closed under addition and multiplication.
(I, +, *) is a ring itself.
Any number in R times a number in I is still in I.
An ideal that consists of all multiples of one element in a ring is called a principal ideal.
An integral domain with only principal ideals is called a Principle Ideal Domain (PID).
|
CCS U Quadrat ic Int e ge r S LIDES . nb 5
Ring homomorphisms and ideals
Theorem 2.8 of Hungerford (1980) states:
If f: R S is a homomorphism of rings, then the kernel of f is an ideal in R.
Conversely, if I is an ideal in R, then the map : R R/I given by r r + I
is an epimorphism of rings with kernel I. R/I is called a quotient ring.
Example: For n , /(n) has kernel
n
, which is equivalent to arithmetic (mod n).
Elements of
n
can be written as a + (n), which shows why adding multiples of n has
no effect in
n
.
Thomas Hungerford, Algebra, Springer, 1980.
|
6 CCS U Quadrat ic Int e ge r S LIDES . nb
Famous example: Gaussian integers, [I]
[I] = { a + b I: a, b }, where I = 1 .
Mathematica shows us below how to add, multiply, and
divide these (typically division produces numbers in [I].)
(a + b I) + (c + d I) // ComplexExpand
(a + b I) + (c + d I) // ComplexExpand
(a + b I) / (c + d I) // ComplexExpand
a c (b d)
a c b d (b c a d)
a c
c
2
d
2

b d
c
2
d
2

b c
c
2
d
2

a d
c
2
d
2
|
CCS U Quadrat ic Int e ge r S LIDES . nb 7
[I] has a norm
For = a + b I [I], the following is a norm: N( a + b I) = a
2
+ b
2

Note that N() =
2
, where | . | is the absolute value in .
Theorem N() = N()*N()
Proof Let = a + b I, = c + d I, and just compute:
(a c - b d)^2 + (b c + a d)^2 // Factor
a
2
b
2
c
2
d
2

Hence, if | , then N() | N().


Hence, if is a unit, then N() = 1 = {1, I}.
So the multiplication in [I] is linked to that of .
See Theorem 1.2, Corollary 1.3, Theorem 2.4 of Keith Conrad's "The Gaussian Integers," which is available at http://www.math.u-
conn.edu/~kconrad/blurbs/ugradnumthy/Zinotes.pdf.
|
8 CCS U Quadrat ic Int e ge r S LIDES . nb
Characterizing all primes in [I]
The discussion in Section 9 of Conrad's "The Gaussian Integers"
is crowned by the following result:
Theorem 9.9. "Every prime in Z[i] is a unit multiple of the following primes:
i) 1 + I
ii) or , where N() = p is a prime in Z
+
which is 1 mod 4.
iii) p, where p is a prime in Z
+
with p 3 mod 4."
FactorInteger[1 + I, GaussianIntegers - True]
FactorInteger[1 000 000 009, GaussianIntegers - True]
FactorInteger[1 000 000 007, GaussianIntegers - True]
{{1 , 1]]
{{, 1], {3747 31 400 , 1], {31 400 3747 , 1]]
{{1 000 000 007, 1]]
See Keith Conrad, "The Gaussian Integers," which is available at http://www.math.uconn.edu/~kconrad/blurbs/ugrad-
numthy/Zinotes.pdf.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 9
Brillhart Algorithm for solving p = a
2
+ b
2
The last slides claims 1000000009 = (3747 + 31400 i )(3747 - 31400 i ).
But how does one determine 3747 and 31400? Fermat gave this
problem as a challenge, and Hermite published an algorithm for it.
Brillhart's algorithm does this (Williams (1995), page 410):
(i) Find a solution of z
2
-1 (mod p) where 0 < z < p/2. This can be done
by using quadratic reciprocity: For p = 1000000009 and q = 11, (p/q) = (q/p) = -1
so that q
p12
(q/p) = -1 (mod p).
(ii) Apply the Euclidean algorithm to z and p (in that order)
and determine the first remainder r
k
satisfying r
k
< p .
(iii) p = r
k
2
+ r
k+1
2
.
Now apply this to 1000000009 in [I]. Below shows a z that satisfies step (i).
p = 1 000 000 009;
z = PowerMod[11, (p - 1) / 4, p]
Mod[z^2, p]
569 522 298
1 000 000 008
Kenneth Williams, "Some Refinements of an Algorithm of Brillhart," Canadian Mathematical Society Conference Proceedings,
Volume 15, 1995.
|
10 CCS U Quadrat ic Int e ge r S LIDES . nb
Brillhart Algorithm in Action
Now we implement step (ii).
divide[a_, b_] := {b, a - b Floor[a / b]}
brillhart[p_, z_] := Module[{temp = divide[p, z]},
While[temp[[1]] > Sqrt[p],
temp = divide[temp[[1]], temp[[2]] ] ];
Return[temp[[1]] + temp[[2]] I] ]
brillhart[1 000 000 009, 569 522 298]
31 400 3747
|
CCS U Quadrat ic Int e ge r S LIDES . nb 11
The Euclidean Algorithm and GCDs in [I] using N()
The Euclidean algorithm in truncates the fractional part of the division.
For [I], we round to the nearest Gaussian integer. As Conrad points out,
the latter convention can also be applied to the Euclidean algorithm in .
euclid[_, _] := Module]] = / [ |, ),
= Round[Re[]] + Round[Im[]] I; = - ; Return[{, }]
{euclid[28 + 17 r, 34 - 19 r], euclid[34 - 19 r, 9 - 17 r], euclid[9 - 17 r, -1 + 6 r]}
{{, 9 17 ], {2 , 1 6 ], {3 , 0]]
Hence, 28 + 17 i = i (34- 19 i )+ 9 - 17 i , and
34 - 19 i = (2 + i )(9- 17 i )+ (-1 + 6 i ), and
9 - 17 i = (-3 - i )(-1+ 6 i )+ 0.
So (-1 + 6i )= GCD, as is any unit times this, such as -i (-1 + 6i )= 6 + i .
ExtendedGCD[28 + 17 I, 34 - 19 I]
{6 , {1 2 , 2]]
Note that = + where N() s (1/2)N(). Numbers are from Example 3.2 of Keith Conrad's "The Gaussian Integers," which
is available at http://www.math.uconn.edu/~kconrad/blurbs/ugradnumthy/Zinotes.pdf.
|
12 CCS U Quadrat ic Int e ge r S LIDES . nb
Unique factorization in [I]
A Euclidean domain = Principal Ideal Domain = Unique Factorization Domain.
This follows from Proposition 2.37, Corollary 2.38, and Theorem 2.52 of Weintraub (2008).
Steven Weintraub, Factorization: Unique and Otherwise, A. K. Peters, 2008.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 13
Ideals in [I]
In[1]:= max = 4;
circle = Graphics[Circle[{0, 0}, 1]];
grid = Flatten[Table[{i, j}, {j, -max, max}, {i, -max, max}], 1];
Manipulate[trangrid = Map[({{a, -b}, {b, a}}.) &, grid];
ListPlot[trangrid, AspectRatio - 1] // Show, {a, -5, 5, 1}, {b, -5, 5, 1}]
Out[4]=
a
b
20 10 10 20
20
10
10
20
|
14 CCS U Quadrat ic Int e ge r S LIDES . nb
Pythagorean triples and [I]
It is well known that all Pythagorean triples are generated by
a
2
b
2
, 2 a b, a
2
+ b
2
for a, b . However, the real
and imaginary part of (a + b I
2
equals the first two entries of
the above vector, so there's a bijection between squared
Gaussian integers all Pythagorean triples.
(a + b I)^2 // Expand
a
2
2 a b b
2
(1 + 2 I)^100
64 431 646 909 858 924 948 087 806 774 847 687 61 132 413 077 625 071 791 758 381 011 357 784
CCS U Quadrat ic Int e ge r S LIDES . nb 15
max = 33;
grid = Flatten[Table[{i, j}, {j, -max, max}, {i, -max, max}], 1];
pythagorean = Map[{[[1]]^2 - [[2]]^2, 2 [[1]] [[2]]} &, grid] // Union;
ListPlot[pythagorean, PlotRange - {{-max^2, max^2}, {-max^2, max^2}}, AspectRatio - 1]
1000 500 500 1000
1000
500
500
1000
|
16 CCS U Quadrat ic Int e ge r S LIDES . nb
Second example of quadratic integers: [ 2 ]
For a, b , N(a + b 2 ) = a
2
- 2 b
2
is a multiplicative norm.
f: [ 2 ] [ 2 ] defined by a + b 2 a - b 2 is an automorphism. f() is denoted
-
.
[ 2 ] with norm, N(), is an Euclidean domain by Theorem 2.8 of Weintraub (2008)
So one can compute GCDs in [ 2 ].
Being an Euclidean domain = [ 2 ] is a PID and UFD.
Since [ 2 ] is a PID, ideals are typically rotated, rectangular lattices.
However, there are differences between [ 2 ] and [ 1 ].
Steven Weintraub, Factorization: Unique and Otherwise, A. K. Peters, 2008.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 17
What are the units of [ 2 ]?
That is, when is N(a + b 2 ) = a
2
- 2 b
2
= 1?
Note that if N(a + b 2 ) = 1, N(a + b 2
n
) = 1
n
= 1.
a
2
- 2 b
2
= 1 is called Pell's equation (mistakenly by Euler),
and all its solutions are generated by 1 + 2
n
. We call
1 + 2 a fundamental unit.
Table[(1 + Sqrt[2])^i // Expand, {i, 1, 20}]
1 2 , 3 2 2 , 7 5 2 , 17 12 2 , 41 29 2 , 99 70 2 , 239 169 2 , 577 408 2 ,
1393 985 2 , 3363 2378 2 , 8119 5741 2 , 19 601 13 860 2 , 47 321 33 461 2 ,
114 243 80 782 2 , 275 807 195 025 2 , 665 857 470 832 2 , 1 607 521 1 136 689 2 ,
3 880 899 2 744 210 2 , 9 369 319 6 625 109 2 , 22 619 537 15 994 428 2
Carefully speaking, each solution above corresponds to 4 solutions.
For example,N(1 + 2 ) = -1 implies that N(1 2 ) = -1.
Article 97 mentions Pell in Leonard Euler (Translated by Rev. John Hewlett), Elements of Algebra, Springer, 1984.
|
18 CCS U Quadrat ic Int e ge r S LIDES . nb
Theorem about primes in [ 2 ]
From Proposition 13.1.3, and the discussion on pages 189-91 of
Ireland and Rosen (1990), the following is proved.
Theorem. Let d = 2,
F
= 4d = 8, , and p = prime in .
If (
F
/p) = 1, then p factors into (a + b d )(a - b d ) (that is, p splits.)
If (
F
/p) = -1, then p is prime in [ d ].
If (
F
/p) = 0, then p is a square (that is, p ramifies.)
Note that p = 2 = 2
2
, so 2 is a square in [ 2 ].
Kenneth Ireland and Michael Rosen, A Classical Introduction to Modern Number Theory, 2nd Edition, Springer, 1990.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 19
Examples of primes in [ 2 ]
Below shows which primes in are still primes in [ 2 ]
and which ones split. Note that the Reduce[] statement below solves
the following generalized Pell's equation: x
2
- 2 y
2
= p.
Do[
If[JacobiSymbol[2, p] = 1,
ans = Reduce[x^2 - 2 y^2 = p, {x, y}, Integers] //. {C[1] - 0} // Last;
f1 = Part[ans, 1, 2] + Part[ans, 2, 2] Sqrt[2];
f2 = Part[ans, 1, 2] - Part[ans, 2, 2] Sqrt[2];
Print[{f1, f2, f1 + f2 // Expand}],
Print[{p}] ],
{p, Table[Prime[i], {i, 1, 25}]}]
20 CCS U Quadrat ic Int e ge r S LIDES . nb
{2]
{3]
{5]
3 2 , 3 2 , 7
{11]
{13]
5 2 2 , 5 2 2 , 17
{19]
5 2 , 5 2 , 23
{29]
7 3 2 , 7 3 2 , 31
{37]
7 2 2 , 7 2 2 , 41
{43]
7 2 , 7 2 , 47
{53]
{59]
{61]
{67]
11 5 2 , 11 5 2 , 71
9 2 2 , 9 2 2 , 73
9 2 , 9 2 , 79
{83]
11 4 2 , 11 4 2 , 89
13 6 2 , 13 6 2 , 97
|
CCS U Quadrat ic Int e ge r S LIDES . nb 21
Prehistory of quadratic integers: Sum of two squares
According to page 14 of Edwards (1977), Diophantus' Arithmetic states
that 65 = 1
2
+ 8
2
= 4
2
+ 7
2
as well as that this "is due to the fact that
65 is the product of 13 and 5, each of which numbers is the sum of two
squares." This suggests that Diophantus was aware of the identity below.
According to page 217 of Stillwell (1997), this identity was explicitly
stated by Abu Ja'far al-Khazin ca. 950, and Fibonacci proved it in his
1225 text, The Book of Squares (Liber Quadratorum).
Note that this identity is also the proof of N() = N()*N() in [I],
which was done in an earlier slide.
(a c - b d)^2 + (b c + a d)^2 = (a^2 + b^2) (c^2 + d^2) // Expand
True
Harold Edwards, Fermat's Last Theorem: A Genetic Introduction to Algebraic Number Theory, Springer, 1977.
John Stillwell, Numbers and Geometry, Springer UTM/RIM Series, 1997.
|
22 CCS U Quadrat ic Int e ge r S LIDES . nb
Fermat and representing primes with quadratic forms
According to page 8 of Cox (1989), the first person to write about
the mathematics of what are now called quadratic integers is Fermat.
He mentions the link between primes and the sum of two squares
to Mersenne in 1640, and writes to Pascal about p = x
2
+ 2 y
2
and
p = x
2
+ 3 y
2
in 1654. The below claims were noted in a letter to
Digby in 1658 (not with this modern notation, however.)
p 1 (mod 4) = p = x
2
+ y
2
.
p 1 (mod 3) = p = x
2
+ 3y
2
.
p 1 or 3 (mod 8) = p = x
2
+ 2y
2
.
It took Euler 40 years to prove the above conjectures!
David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication, Wiley, 1989.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 23
Identity for x
2
+ n y
2
There are analogues of the identity that Diophantus knew, noted on
pages 17-18 of Edwards (1977). These were known by Fermat,
who probably used them in his arguments of infinite descent.
The first factorization is given in Article 169 of Euler (1984).
The second factorization is given in Article 392 of Euler (1984).
See article 175 in Euler (1984) for the general case identity given below.
Factor[(a^2 + b^2) (c^2 + d^2), Extension - I]
Factor[(a^2 + 2 b^2) (c^2 + 2 d^2), Extension - Sqrt[-2]]
(a b) (a b) (c d) (c d)
a 2 b a 2 b c 2 d c 2 d
(a^2 + n b^2) (c^2 + n d^2) = (a c - n b d)^2 + n (b c + a d)^2 // Expand
True
Harold Edwards, Fermat's Last Theorem: A Genetic Introduction to Algebraic Number Theory, Springer, 1977.
Leonard Euler (Translated by Rev. John Hewlett), Elements of Algebra, Springer, 1984. This is based on an 1840 edition
published by Longman, Orme, and Co.
|
24 CCS U Quadrat ic Int e ge r S LIDES . nb
Fermat, 1658, and Euler, 1744, on x
2
+ 5 y
2
Fermat conjectured the following:
p, q 3, 7 (mod 20) = p q = x
2
+ 5 y
2
From page 19 of Cox (1989):
For odd primes, p, not equal to 5, Euler proved:
p = x
2
+ 5 y
2
= p 1, 9 (mod 20)
2 p = x
2
+ 5 y
2
= p 3, 7 (mod 20)
David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication, Wiley, 1989.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 25
Lagrange's 1773-5 analysis of x
2
+ 5 y
2
Page 33 of Cox (1989) states that Lagrange was able to prove the following:
For odd primes p, and for some x, y ,
p 1, 9 (mod 20) = p = x
2
+ 5 y
2

p 3, 7 (mod 20) = p = 2 x
2
+ 2 x y + 3 y
2
.

Moreover, the forms x
2
+ 5 y
2
and 2 x
2
+ 2 x y + 3 y
2
are closely related.
p, q 3, 7 (mod 20) = p q = x
2
+ 5 y
2
because of the first identity below, which is
(2.30), on page 37, of Cox (1989). And from an earlier slide (see 2nd identity below),
we know that p, q 1, 9 (mod 20) = p q = x
2
+ 5 y
2
.
Finally, it can be shown that p 1, 9 (mod 20) and q 3, 7 (mod 20) =
p q = 2 x
2
+ 2 x y + 3 y
2
by the third identity below, which is on pages 18-19
of the translator's introduction of Dedekind (1996).
David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication, Wiley, 1989.
Richard Dedekind, Theory of Algebraic Integers, Cambridge University Press, 1996. Translated by John Stillwell from 1877
edition.
|
26 CCS U Quadrat ic Int e ge r S LIDES . nb
Lagrange's analysis summarized as C
2
The above can be summarized by the following multiplication table of forms,
where form1 is x
2
+ 5 y
2
and form2 is 2 x
2
+ 2 x y + 3 y
2
.

+ form1 form2
form1 form1 form2
form2 form2 form1
(2 a^2 + 2 a b + 3 b^2) (2 x^2 + 2 x y + 3 y^2) == (2 a x + a y + b x + 3 b y)^2 + 5 (a y - b x)^2 // Expand
(a^2 + 5 b^2) (x^2 + 5 y^2) = (a x - 5 b y)^2 + 5 (b x + a y)^2 // Expand
(a^2 + 5 b^2) (2 x^2 + 2 x y + 3 y^2) =
2 (a x - b x - 3 b y)^2 + 2 (a x - b x - 3 b y) (a y + 2 b x + b y) + 3 (a y + 2 b x + b y)^2 // Expand
True
True
True
David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication, Wiley, 1989.
Richard Dedekind, Theory of Algebraic Integers, Cambridge University Press, 1996. Translated by John Stillwell from 1877
edition.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 27
Quadratic forms: Lagrange and Gauss
The material on the next few slides is from chapter 1, section 2A of Cox (1989),
which follows Lagrange's "Recherches d' Arithmetique" of 1773-75 and Gauss'
Disquisitiones Arithmeticae of 1801.
f(x,y) = a x
2
+ b x y + c y
2
, where x, y , is called a quadratic form.
Note that a x
2
+ 2 b x y + c y
2
was the notation used in the 19th century.
f is called primitive if gcd(a, b, c) = 1.
m is represented by f if m = f(x, y) has a solution in integers.
m is properly represented by f(x, y) if gcd(x, y) = 1.
Forms f and g are equivalent if f(x, y) = g(p x + q y, r x + s y),
where p s - q r = 1 and p, q, r, s
Gauss makes the following distinction:
Forms f and g are properly equivalent when p s - q r = 1.
Forms f and g are improperly equivalent when p s - q r = -1.
Chapter 1, Section 2A of David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication,
Wiley, 1989.
|
28 CCS U Quadrat ic Int e ge r S LIDES . nb
Linear algebra and forms
Today we use the following language with quadratic forms and equivalence.
f(x, y) = (x y)
a b 2
b 2 c
x
y
= x
T
Ax .
Let
x

=
p q
r s
x
y
= Px, then f and g are equivalent if f(x) = g(Px).
Proper equivalence means det(P) = 1, and improper equivalence means det(P) = -1.
The discriminant of f is D = b
2
4 a c = 4 det
a b 2
b 2 c
Also note that a change of coordinates on x
T
Ax can be written as:
(Px
T
AP x = x
T
P
T
AP x, and that det(P
T
AP) = detP
2
det(A) = det(A)
since det(P) = 1. Thus if f and g are equivalent, they have the same discriminant.
Chapter 1, Section 2A of David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication,
Wiley, 1989.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 29
Reduced forms
Define a primitive, positive definite form f to be reduced if we have either
(1) |b| < a < c or
(2) 0 s |b| = a s c or
(3) 0 s |b| s a = c.
Theorem 2.8. Every primitive, positive definite form is properly equivalent
to a unique reduced form.
Recall that f primitive means gcd(a, b, c) = 1, and proper equivalence means
the transformation matrix P satisfies det(P) = 1.
Chapter 1, Section 2A of David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication,
Wiley, 1989.
|
30 CCS U Quadrat ic Int e ge r S LIDES . nb
Gauss introduces class numbers
Gauss makes the following definitions in his Disquisitiones Arithmeticae
(see page 29 of Cox (1989).)
Two forms are in the same class if they are properly equivalent.
Let h(D) = the number of reduced forms with discriminant D.
Gauss computed about 100 values of h(D) in Article 303 of his
Disquisitiones Arithmeticae.
Chapter 1, Section 2A of David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication,
Wiley, 1989.
Carl Friedrich Gauss, Disquisitiones Arithmeticae, Yale University Press, 1965. Translated by Arthur A. Clarke.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 31
Table of h(D) and reduced forms
D h (D) Reduced Forms
4 1 x
2
y
2
8 1 x
2
2 y
2
12 1 x
2
3 y
2
20 2 x
2
5 y
2
, 2 x
2
2 x y 3 y
2
28 1 x
2
7 y
2
56 4 x
2
14 y
2
, 2 x
2
7 y
2
, 3 x
2
2 x y 5 y
2
Above is equation 2.14, page 29 of Cox (1989). Gauss computed a more
extensive version of the first two rows here in Article 303 of Gauss (1965)
though in a different format.
Chapter 1, Section 2A of David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication,
Wiley, 1989.
Carl Friedrich Gauss, Disquisitiones Arithmeticae, Yale University Press, 1965. Translated by Arthur A. Clarke.
|
32 CCS U Quadrat ic Int e ge r S LIDES . nb
Gauss' conjecture for D < 0
Below is a plot of square-free D vs. class number.
Gauss conjectured that as D -o, min h(D) o.
This was proved by Heilbronn (1934) according to
http://mathworld.wolfram.com/GausssClassNumberConjecture.html.
class = {}; max = 5000;
Do[If[SquareFreeQ[d], AppendTo[class, {-d, NumberFieldClassNumber [Sqrt[d]]}], Null],
{d, -1, -max, -1}]
ListPlot[class, AxesLabel - {"Square-free D", "Class number"}]
1000 2000 3000 4000 5000
Squarefree D
20
40
60
80
100
120
Class number
"Gauss's Class Number Conjecture" at Wolfram MathWorld. http://mathworld.wolfram.com/GausssClassNumberConjecture-
.html
accessed 9/2010.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 33
Gauss' conjecture for class number = 1, D > 0
Gauss conjectured that there are an infinite number of rings with
class number D for D > 0. This is still an open problem.
class = {}; max = 5000;
Do[If[SquareFreeQ[d],
AppendTo[class, {d, NumberFieldClassNumber [Sqrt[d]]}], Null], {d, 1, max}]
ListPlot[class, AxesLabel - {"Square-free D", "Class number"}]
1000 2000 3000 4000 5000
Squarefree D
2
4
6
8
Class number
|
34 CCS U Quadrat ic Int e ge r S LIDES . nb
Class number = 1 and unique factorization
From Alaca and Williams (2004):
Theorem 12.1.1. Let K be an algebraic number field. Then
h(K) = 1 = O
K
is a PID = O
K
is a UFD.
From Weintraub (2008):
Theorem 2.8. Let R = O( D ). Then R is a Euclidean domain for
D = -11, -7, -3, -2, -1, 2, 3, 5, 6, 7, 11, 13, 17, 21, and 29.
Proof: Weintraub proves this with elementary methods. .
Note that D = 33, 37, 41, 57, 73 also make R a Euclidean domain, and
these are all the values of D that make R a Euclidean domain with
respect to its norm. However, note the following theorem:
Trivia: Theorem (Clark (1994)). The ring of integers of ( 69 ) is
Euclidean but not norm-Euclidean.
From Wikipedia:
Stark-Heegner Theorem. -D = 1, 2, 3, 7, 11, 19, 43, 67, 163 is an
exhaustive list of D that make R = O( D ) a UFD.
Gauss conjectured that this was true in 1801. The proof was finally
finished in the 1950s and 1960s.
Saban Alaca and Kenneth Williams, Introductory Algebraic Number Theory, Cambridge, 2004.
David Clark, "A Quadratic Field which is Euclidean but not Norm-Euclidean," Manuscripta Mathematica, Vol 83, pages
327-330, 1994.
Steven Weintraub, Factorization : Unique and Otherwise, A. K. Peters, 2008.
"StarkHeegner theorem," http://en.wikipedia.org/wiki/Stark%E2%80%93Heegner_theorem.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 35
Non-unique factorization of irreducibles in [ 5 ]
A common example in books or on the web of
non-unique factorization is 2*3 = (1+ 5 ) (1- 5 ).
For example, this appears on Wikipedia under
"Unique Factorization Domain." This is an old example,
one that appears in Dedekind's Theory of Algebraic Integers,
first example of equation (1), page 87. However, finding
examples using Mathematica does not take that much work.
The steps are:
(1) Produce irreducibles
(2) Multiple these irreducibles together
(3) Match up identical products
"Unique Factorization Domain" on Wikipedia at http://en.wikipedia.org/wiki/Unique_factorization_domain. Accessed 9/2010.
Richard Dedekind, Theory of Algebraic Integers, Cambridge University Press, 1996. Translated by John Stillwell from 1877
edition.
|
36 CCS U Quadrat ic Int e ge r S LIDES . nb
Some irreducibles in [ 5 ]
list = {}; max = 100;
Do[ If[1 < a^2 + 5 b^2 < max, AppendTo[list, a + b Sqrt[-5]], Null],
{a, -Floor[Sqrt[max]], Floor[Sqrt[max]]}, {b, -Floor[Sqrt[max / 5]], Floor[Sqrt[max / 5]]}]
reducibles = Table[Expand[i j], {i, list}, {j, list}] // Flatten // Union;
irreducibles = Complement[list, reducibles];
irreducibles = SortBy[irreducibles, Abs[]^2 &]
2, 2, 5 , 5 , 1 5 , 1 5 , 1 5 , 1 5 , 3, 3, 2 5 , 2 5 ,
2 5 , 2 5 , 3 5 , 3 5 , 3 5 , 3 5 , 4 5 , 4 5 ,
4 5 , 4 5 , 1 2 5 , 1 2 5 , 1 2 5 , 1 2 5 , 3 2 5 , 3 2 5 ,
3 2 5 , 3 2 5 , 6 5 , 6 5 , 6 5 , 6 5 , 1 3 5 , 1 3 5 ,
1 3 5 , 1 3 5 , 7, 7, 2 3 5 , 2 3 5 , 2 3 5 , 2 3 5 , 4 3 5 ,
4 3 5 , 4 3 5 , 4 3 5 , 8 5 , 8 5 , 8 5 , 8 5 , 7 2 5 ,
7 2 5 , 7 2 5 , 7 2 5 , 9 5 , 9 5 , 9 5 , 9 5 , 3 4 5 ,
3 4 5 , 3 4 5 , 3 4 5 , 7 3 5 , 7 3 5 , 7 3 5 , 7 3 5
|
CCS U Quadrat ic Int e ge r S LIDES . nb 37
Examples of non-unique factorizations in [ 5 ]
6 (3) (2)
6 (1 5 ) (1 5 )
4 2 5 (1 5 ) (1 5 )
4 2 5 (2 5 ) (2)
3 3 5 (3) (1 5 )
3 3 5 (2 5 ) (1 5 )
9 (3) (3)
9 (2 5 ) (2 5 )
8 2 5 (3 5 ) (1 5 )
8 2 5 (4 5 ) (2)
2 4 5 (3 5 ) (1 5 )
2 4 5 (1 2 5 ) (2)
11 5 (3 5 ) (2 5 )
11 5 (1 2 5 ) (1 5 )
9 3 5 (4 5 ) (1 5 )
9 3 5 (3 5 ) (3)
9 3 5 (1 2 5 ) (1 5 )
|
38 CCS U Quadrat ic Int e ge r S LIDES . nb
Dedekind's solution: Unique factorization of ideals
Dedekind showed that one could define divisibility for ideals of a ring
in his book Dedekind (1996). Here is an example of his
theory from page 173, Section 5.6 of Weintraub (2008).
Convert 6 = 2*3 = (1 + 5 )(1 - 5 ) to principal ideals in [ 5 ]:
(6) = (2)(3) = (1 + 5 )(1 - 5 ) (*)
Let I
1
= 2, 1 + 5 and I
2
= 3, 1 + 5 .
One can show that (2) = I
1
2
, (3) = I
2
I
2
, (1 + 5 ) = I
1
I
2
, and (1 - 5 ) = I
1
I
2
.
So (*) becomes (6) = (I
1
2
I
2
I
2
) = I
1
I
2
I
1
I
2
), which is just an reordering of the ideals.
Hence uniqueness (with respect to ideal factorization) is regained.
Richard Dedekind, Theory of Algebraic Integers, Cambridge University Press, 1996. Translated by John Stillwell from 1877
edition.
Steven Weintraub, Factorization: Unique and Otherwise, A. K. Peters, 2008.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 39
Plots of the ideals (1 + 5 ), I
1
, I
2
max = 10; plotmax = 20;
m0 = Graphics[Circle[{0, 0}, 1]];
grid = Flatten[Table[Table[{i, j Sqrt[-d]}, {j, -max, max}], {i, -max, max}], 1];
a = 1; b = 1;
trangrid1 = Map[({{a, -b Sqrt[5]}, {b Sqrt[5], a}}.) &, grid];
g1 = ListPlot[trangrid1, AspectRatio - 1, PlotMarkers - {m0, 0.02},
PlotRange - {{-plotmax, plotmax}, {-plotmax, plotmax}}]
max = 4; plotmax = 20;
m0 = Graphics[Circle[{0, 0}, 1]];
grid = Flatten[Table[Table[{i, j Sqrt[-d]}, {j, -max, max}], {i, -max, max}], 1];
a1 = 2; b1 = 0; a2 = 1; b2 = 1;
trangrid1 = Map[({{a1, -b1 Sqrt[5]}, {b1 Sqrt[5], a1}}.) &, grid];
trangrid2 = Map[({{a2, -b2 Sqrt[5]}, {b2 Sqrt[5], a2}}.) &, grid];
trangrid = Table[i + j, {i, trangrid1}, {j, trangrid2}] // Union;
g = ListPlot[trangrid, AspectRatio - 1, PlotMarkers - {m0, 0.012},
PlotRange - {{-plotmax, plotmax}, {-plotmax, plotmax}}]; Show[g]
a1 = 3; b1 = 0; a2 = 1; b2 = 1;
trangrid1 = Map[({{a1, -b1 Sqrt[5]}, {b1 Sqrt[5], a1}}.) &, grid];
trangrid2 = Map[({{a2, -b2 Sqrt[5]}, {b2 Sqrt[5], a2}}.) &, grid];
trangrid = Table[i + j, {i, trangrid1}, {j, trangrid2}] // Union;
g = ListPlot[trangrid, AspectRatio - 1, PlotMarkers - {m0, 0.012},
PlotRange - {{-plotmax, plotmax}, {-plotmax, plotmax}}]; Show[g]
20 10 10 20
20
10
10
20
40 CCS U Quadrat ic Int e ge r S LIDES . nb
20 10 10 20
20
10
10
20
20 10 10 20
20
10
10
20
As Stillwell (1998) points out on pages 244-5, the last two ideals are not rectangular lattices, hence
they are not principal ideals.
John Stillwell, Numbers and Geometry, Springer UTM/RIM Series, 1997.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 41
References
Sukumar Adhikari, An Introduction to Commutative Algebra and Number Theory, CRC Press, Narosa Publishing House, 2001.
Saban Alaca and Kenneth Williams, Introductory Algebraic Number Theory, Cambridge, 2004.
Michael Artin, Chapter 11 of Algebra, Prentice Hall, 1991.
David Clark, "A Quadratic Field which is Euclidean but not Norm-Euclidean," Manuscripta Mathematica, Vol 83, pages
327-330, 1994.
Keith Conrad, "The Gaussian Integers" is available at http://www.math.uconn.edu/~kconrad/blurbs/ugradnumthy/Zinotes.pdf.
David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication, Wiley, 1989.
Richard Dedekind, Theory of Algebraic Integers, Cambridge University Press, 1996. Translated by John Stillwell from 1877
edition.
Harold Edwards, Fermat's Last Theorem: A Genetic Introduction to Algebraic Number Theory, Springer, 1977.
Leonard Euler (Translated by Rev. John Hewlett), Elements of Algebra, Springer, 1984. This is based on an 1840 edition
published by Longman, Orme, and Co.
Carl Friedrich Gauss, Disquisitiones Arithmeticae, Yale University Press, 1965. Translated by Arthur A. Clarke.
William Gilbert, "The Division Algorithm in Complex Bases," Canadian Mathematical Bulletin, Volume 39, 1996.
William Gilbert, "Arithmetic in Complex Bases," Mathematics Magazine, Volume 57, 1984.
Thomas Hungerford, Algebra, Springer, 1980.
Kenneth Ireland and Michael Rosen, A Classical Introduction to Modern Number Theory, 2nd Edition, Springer, 1990.
I. Katai and J. Szabo, "Canonical number systems for complex integers," Acta Sci. Math., Volume 37, 1975.
Donald Knuth, The Art of Computer Programming, Volume 2, Seminumerical Algorithms, 2nd Ed., Addison-Wesley, 1981.
William LeVeque, Fundamentals of Number Theory, Addison-Wesley, 1977.
William Stein, Algebraic Number Theory, A Computational Approach, 2008. Available at http://modular.math.washington.edu/-
books/ant/ant.pdf.
John Stillwell, Numbers and Geometry, Springer UTM/RIM Series, 1997.
Steven Weintraub, Factorization: Unique and Otherwise, A. K. Peters, 2008.
Wikipedia, http://www.wikipedia.org/.
Kenneth Williams, "Some Refinements of an Algorithm of Brillhart," Canadian Mathematical Society Conference Proceedings,
Volume 15, 1995.
Wolfram MathWorld, http://mathworld.wolfram.com/.
|
42 CCS U Quadrat ic Int e ge r S LIDES . nb
Does Bill Gates do number theory on the side?
"Because both the system's privacy and the security of
digital money depend on encryption, a breakthrough in
mathematics or computer science that defeats the
cryptographic system could be a disaster. The obvious
mathematical breakthrough would be the
development of an easy way to factor large prime numbers."
(Gates 1995, The Road Ahead, p. 265).
We have seen above that some primes in do factor in [I] and [ 2 ].
For example, 97 = (4 + 9 i )(4 - 9 i )= 13 6 2 13 6 2 .
William Stein suggests "However, perhaps Gates is an algebraic
number theorist, and he really meant what he said: then we might imagine that
he meant factorization of primes of Z in rings of integers of number fields."
See page 47 of Stein (2008).
William Stein, Algebraic Number Theory, A Computational Approach, 2008. Available at http://modular.math.washington.edu/-
books/ant/ant.pdf.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 43
Characterizations of all ideals in [ 5 ]
Theorem 9.5 of Adhikari (2001).
"If I is a non-zero ideal of the ring [ 5 ] and a non-zero element of I of
minimal absolute value r, then either I is the principal ideal () with lattice
basis (, 5 ) or I is not a principal ideal and has the lattice basis
(, ( + 5 )/2). The second case occurs only when ( + 5 )/2
is an element of I."
Sukumar Adhikari, An Introduction to Commutative Algebra and Number Theory, CRC Press, Narosa Publishing House, 2001.
|
44 CCS U Quadrat ic Int e ge r S LIDES . nb
How to find z for the Brillhart-Hermite Algorithm?
This can be done by quadratic reciprocity: (p/q)(q/p) = (-1)^((p-1)/2 (q-1)/2)
(This is Theorem 1 of Ireland and Rosen (1990).)
For p = 1000000009 and q = 11, (p/q) (q/p) = 1
since (p-1) is divisible by 4, hence (p-1)/2 is even.
Note that (p/q) = (8/11) = (2/11)^3, but (2/11) = -1
by quadratic reciprocity: (2/q) = (-1)^((q^2-1)/8) = (-1)^15 = -1.
Hence (p/q) = (q/p) = -1.
By Proposition 5.1.2 of Ireland and Rosen (1990), q^((p-1)/2) (q/p) = -1 (mod p).
Hence z q^((p-1)/4) 11^250000002 (mod p).
p = 1 000 000 009; q = 11;
JacobiSymbol[q, p]
z = PowerMod[q, (p - 1) / 4, p]
1
569 522 298
Kenneth Ireland and Michael Rosen, A Classical Introduction to Modern Number Theory, 2nd Edition, Springer, 1990.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 45
Modular arithmetic in [I]
Since the Euclidean algorithm works in [I], we are able to do
modular arithmetic in [I]. One example of this is that there's
an analog of Fermat's Little Theorem that holds in the Gaussian integers.
Below combines Keith Conrad's Theorems 7.12 and 7.14 of his
"The Gaussian Integers."
Theorem. Let be a Gaussian prime and note that the number of Gaussian integers
modulo is N() =
-
. If 0 mod , then
N1
1 mod .
Steven Weintraub, Factorization: Unique and Otherwise, A. K. Peters, 2008.
Keith Conrad, "The Gaussian Integers" is available at http://www.math.uconn.edu/~kconrad/blurbs/ugradnumthy/Zinotes.pdf.
|
46 CCS U Quadrat ic Int e ge r S LIDES . nb
Gaussian integer bases
Gilbert (1996) points out that Knuth (1981) notes that one can write
complex numbers in binary notation for the base (-1 + I). Gilbert gives
the example (-2-I)/2 = 110.01
1+I
. Gilbert (1996) also notes that
using the digits {0, 1, 2, ..., N(b)-1}, Katai and Szabo (1975) proved that
the bases (-n I), where n is a positive integer, are the only ones possible.
Finally, Gilbert (1984) shows how to add, subtract, multiply complex numbers
in this notation, while Gilbert (1996) shows how to divide in this notation.
William Gilbert, "The Division Algorithm in Complex Bases," Canadian Mathematical Bulletin, Volume 39, 1996.
William Gilbert, "Arithmetic in Complex Bases," Mathematics Magazine, Volume 57, 1984.
I. Katai and J. Szabo, "Canonical number systems for complex integers," Acta Sci. Math., Volume 37, 1975.
Donald Knuth, The Art of Computer Programming, Volume 2, Seminumerical Algorithms, 2nd Ed., Addison-Wesley, 1981.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 47
Rank plot for D > 0
class1 = {}; max = 1000;
Do[If[SquareFreeQ[d] && NumberFieldClassNumber [Sqrt[d]] = 1,
AppendTo[class1, d], Null], {d, 1, max}]
class1[[1 ;; 50]]
ListPlot[class1, AxesLabel - {"Rank", D}]
{1, 2, 3, 5, 6, 7, 11, 13, 14, 17, 19, 21, 22, 23, 29, 31, 33,
37, 38, 41, 43, 46, 47, 53, 57, 59, 61, 62, 67, 69, 71, 73, 77, 83, 86,
89, 93, 94, 97, 101, 103, 107, 109, 113, 118, 127, 129, 131, 133, 134]
50 100 150 200 250
Rank
200
400
600
800
1000
D
"Gauss's Class Number Conjecture" at Wolfram MathWorld. http://mathworld.wolfram.com/GausssClassNumberConjecture-
.html
accessed 9/2010.
|
48 CCS U Quadrat ic Int e ge r S LIDES . nb
Euler proves Fermat's conjectures for x
2
+ n y
2
Page 12 of Cox (1989) says it took Euler 40 years from the time he first heard
about Fermat's conjectures concerning x
2
+ n y
2
, n = 1, 2, 3, until 1772, when
he was able to prove all of them. His proof used his knowledge of quadratic
reciprocity, a result that he failed to prove in general. In Euler's notes
(see page 16 of Cox (1989)), he conjectures that (below uses modern notation):
(N/p) = 1 = p = (mod 4N) for odd values of that need to be determined.
David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication, Wiley, 1989.
|
CCS U Quadrat ic Int e ge r S LIDES . nb 49
Definite and indefinite forms
Equation (2.4) of Cox (1989) states: 4 a f(x, y) = (2a x + b y
2
D y
2
.
So if D > 0, then f can be both positive and negative and is called indefinite.
If D < 0 and a > 0, then f is always positive and is called positive definite.
If D < 0 and a < 0, then f is always negative and is called negative definite.
From now on, we are only interested in positive definite forms such as x
2
+ n y
2
.
Chapter 1, Section 2A of David Cox, Primes of the Form x
2
+ n y
2
: Fermat, Class Field Theory, and Complex Multiplication,
Wiley, 1989.
|
50 CCS U Quadrat ic Int e ge r S LIDES . nb
[(1+ 3 )/2] vs. [ 3 ]
NEED CONCRETE EXAMPLE HERE.
Let = e
2 i3
= (-1 + 3 )/2. That is, is a cube root of unity. Let a, b below.
= Exp[2 Pi I / 3]
Solve[x^2 - (2 a - b) x + (a^2 - a b + b^2) = 0, x] // Expand

2
3
x a
b
2

1
2
3 b, x a
b
2

1
2
3 b
The real and imaginary parts of the solution above are either
both integer or both half-integers. Hence, the Eisenstein integers are
not [ 3 ] but are []. This is a technicality that Euler missed.
See http://en.wikipedia.org/wiki/Quadratic_integer and http://en.wikipedia.org/wiki/Eisenstein_integer (accessed 9/2010.)
|
CCS U Quadrat ic Int e ge r S LIDES . nb 51

You might also like