You are on page 1of 4

Topics in analytic number theory, Lent 2013.

Lecture 19: Bombieris theorem, initial steps


Bob Hough
March 4, 2013
Reference for this lecture: Bombieri, Sections 5, 6.
Thus far we have seen analytic, arithmetic and multiplicative versions of the
large sieve. Now we add an integral version and combine them in the form that
we need for Bombieris Theorem.
Theorem 19.1P(Large sieve, integral form (Gallagher)). Let T 2. Let
{an }
|an | < . We have
n=1 satisfy

2
2

Z X
Z T X


dy

it
2

an n dt  T
an
.



T n
0
yne/T y y
As in the proof of the analytic large sieve we will use Fejers kernel. We
recall that this is defined by

2
sin x
F (x) =
,
x
with Fourier transform
F () = max(0, 1 ||).
We also recall that we defined FN (x) = F (x/N ) and that this satisfies FN () =
N F (N ).
Note also our convention for the Fourier transform:
Z
f() =
f (x)e2ix dx.

Proof. We first bound the LHS. We have


2

2
Z T X

X

2 Z





an nit dt
F2T (t)
an nit dt.





4
T
n
n
Opening the square and exchanging sum and integral, this becomes
Z
n1
2 X
2 T X
T
n2
an1 an2
F2T (t)ei log n2 dt =
an1 an2 F ( log )
4 n ,n
2 n ,n

n1

T X
T
n2
=
an1 an2 max(0, 1 | log |).
2 n ,n

n1
1

Now the integral on the RHS is given by



2

Z X
Z
X

dy
dy


=
a
an1 an2 ,
n

y
y
0
0
yne/T y
yn1 ,n2 e/T y
which, after exchanging sum and integral, and keeping careful track of the range
of integrations, gives
Z min(n1 ,n2 )
X
dy
an1 an2
,
e/T max(n1 ,n2 ) y
n ,n
1

with the convention that the integral is zero if lower limit is larger than upper
limit. This is equal to


X
min(n1 , n2 )
an1 an2 max 0, /T + log
max(n1 , n2 )
n1 ,n2



X
T
n2
=
an1 an2 max 0, 1 log ,
T n ,n

n1
1

and comparison with the LHS proves the theorem. Note that absolute convergence of the last sum justifies the exchange of sum and integral above.
The hybrid version of the large sieve combines the integral and multiplicative
versions. Recall that the multiplicative version stated the following.
Theorem (Multiplicative large sieve, variant). Let N, Q 1 and M an arbi+N
trary integer. Let (bk )M
k=M +1 be complex numbers which satisfy bk = 0 whenever
k has a prime factor less than or equal to Q. We have
2

M
+N
M +N

X
X
Q X X
2

2
(N + 2Q )
|bk |2 .
log
bk (k)



q
4
qQ

mod q

k=M +1

k=M +1

Theorem 19.2 (Hybrid


large sieve). Let T 2. Let {an }
n=1 be complex
P
numbers satisfying
|an | < and the additional condition that an = 0 if n
has a factor T . Then
2
Z

X
X
T X T X
it
log
(|an |2 (n + T 3 )).
an (n)n dt 


q
T n
n
qT

mod q

[Note: it may be that the RHS = .]


Proof. We use the notation A  B to mean A is within a fixed constant of B.
By the integral large sieve, the left hand side is bounded by

2


Z
X
X
X

dy

T


 T2
log
a
(n)
(1)
n

y .
q
T e/T

/T
qT

mod q

yne

Note that we have set the lower limit of integration to T e/T since an = 0 if
n T . For y > T e/T the interval [y, e/T y] has length y(e/T 1)  y
T  1.
2

Thus exchanging sum and integral, and applying the multiplicative large sieve,
we find that (1) is bounded by
Z
X
dy
y
 T2
(T 2 + )
|an |2
T
y
/T
e
T
/T
yne

Exchanging sum and integral, and noting that n  y, this is

n
T
|an | (T + )
T
n=1
2

ne/T

X
dy

|an |2 (T 3 + n),
y
n=1

as wanted.
We now have the necessary technology to prove Bombieris theorem. We
recall the statement.
Theorem (Bombieri). There exists c > 0 such that, for all T > 2, and all
1/2 < 1,
X X
N (, T ; ) = O(T c(1) ).
qT mod q

If there exists a real primitive character 0 mod q 0 with q 0 < T , such that
(s, 0 ) has zero > 1 log T , then we have the improved estimate

X X

N 0 (, T ; ) = O(T c(1) (1 ) log T ).

qT mod q

Here the
cluded.

on N 0 (, T ; ) indicates that the single exceptional zero is to be ex-

Note that, although the first claim of the theorem would include a hypothetical exceptional zero in the count, there is at most one such zero. Since the
claimed bound is  1, this zero may be ignored, so that it suffices to estimate
N 0 (, T ; ) for both statements.
We also recall the statement of the zero-detection lemma.
Lemma (Bombieri Lemma B). Let mod q primitive, q T , |w| T , and
also 2 |w| in the case that = 1. Let log1 T r < 1012 . Suppose that L(s, )
has non-exceptional zero 0 with |0 (1 + iw)| < r. There exist constants
A, B, C, D > 1 such that, if x > T B then

2

Z xA X

ap (p) dy

D(log x)3 xCr .

p1+iw y
x

xp<y

Here ap = log p if there is no exceptional zero, and ap = log p(1 +


is an exceptional zero for L(s, 1 ).

1 (p)
p11

) if 1

Proof of Bombieris Theorem, initial steps. We may suppose that < 1 logc T
since for larger we are in the zero-free region for the non-exceptional zeros.
Also, note that the sum in the theorem contains O(T 2 ) L-functions, each of
which has O(T log T ) zeros in the interval T T . Therefore, a trivial
3

estimate for the LHS is O(T 3 log T ). It follows that by choosing the constant C
in the theorem sufficiently large, we may assume that > 1c0 for a sufficiently
small constant c0 > 0. In particular, we may safely assume that the Riemann
zeta function has no zeros in the region || < 2, > , so the restriction |w| > 2
in the case of = 1 in Lemma B will not come into play.
Set r = C 0 (1 ) such that C 0 2 and 1/ log T < r < 1012 . Notice, in
particular, that 1 < r/2. Suppose that = + i is a non-exceptional zero
of L(s, ) satisfying and || < T . Then for all |w | < r/2 we have
|(1 + iw) | |1 | + |w | r, and therefore for such w, if x > T B is
sufficiently large, then

2

Z xA X

dy
a
(p)
p
3 Cr


.

y  (log x) x
1+iw
p
x


xp<y

Therefore
+r/2

xA

r/2

2



X ap (p) dy


dw  r(log x)3 xCr ,

1+iw y

xp<y p

and so
Z

X
(,)=0
>,||<T

+r/2

r/2

xA

2



X ap (p) dy


dw  r(log x)3 xCr N 0 (, T ; ).

1+iw y

xp<y p

Since for any given w [T r/2, T + r/2], the number of zeros of L(s, ) with
R +r/2
P
| (1 + iw)| < r is  r log T (Lemma 13.2), the sum
(,)=0 r/2 covers
>,||<T

R T +r/2

with multiplicity  r log T . Therefore, we deduce

T r/2

T +r/2

T r/2

xA

2


X
dy

(log x)3 Cr 0
a
(p)
p


dw

x
N (, T ; ).

1+iw y
log T

xp<y p

We are now in position to sum over characters and apply the hybrid Large Sieve
to estimate the LHS. We will do this next lecture and conclude the proof.

You might also like