20.01.2015 Views

savastru-varbanets-0..

savastru-varbanets-0..

savastru-varbanets-0..

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

’iauliai Math. Semin.,<br />

3 (11), 2008, 207219<br />

ON THE REPRESENTATION OF INTEGERS<br />

AS THE SUM OF kTH POWERS<br />

IN ARITHMETIC PROGRESSIONS<br />

Olga SAVASTRU, Pavel VARBANETS<br />

Odessa National University, Dvoryanskaya st.2, Odessa, 65026, Ukraine,<br />

e-mail: sav olga@bk.ru, varb@sana.od.ua<br />

Abstract. We investigate the behaviour of integers represented as the<br />

sum of k-th powers in arithmetic progressions. Using the Waring zeta-function<br />

introduced by A. Vinogradov, we prove the asymptotic formula for the<br />

summatory function of the number of representations of m as the sum of<br />

k-th powers in arithmetic progressions.<br />

Key words and phrases: asymptotic estimate, meromorphic continuation,<br />

Waring zeta-function.<br />

2000 Mathematics Subject Classication: 11N37.<br />

1. Introduction<br />

Dedicated to the 60th birthday<br />

of Professor Antanas Laurin£ikas<br />

Let α 0 , δ 0 ∈ (0, 1], α, δ ∈ (0, 1] n , and let n, k 2 be integers, k n<br />

We denote<br />

< n k.<br />

Θ k (x; α 0 , δ 0 ) =<br />

∞∑<br />

e −πx(m+δ 0) k e 2πimα 0<br />

.<br />

m=0<br />

For Re s > 1, consider the absolutely convergent series<br />

ζ n,k (s; α, δ) =<br />

∞∑<br />

( n∑ ) −s<br />

e 2πim α j +δ j )<br />

j=1(m k , m α = m 1 α 1 +. . .+m n α n ,<br />

m j =0<br />

1jn


208 O. Savastru, P. Varbanets<br />

which, for k = 2, n 2, coincides with the classical Epstein zeta-function,<br />

and, for k = n 3, α = δ = (0, . . . , 0), becomes the Waring zeta-function<br />

introduced for the rst time by A. Vinogradov [7]. Denote<br />

Then we obtain<br />

V n,k (λ; α, δ) =<br />

ζ n,k (s; α, δ) = ∑ λ>0<br />

∞∑<br />

m j =0<br />

n∑<br />

(m j +δ j ) k =λ<br />

j=1<br />

e 2πi(m 1α 1 +...+m nα n) .<br />

V n,k (λ; α, δ)<br />

λ s , Re s > 1.<br />

This function also will be called the Waring zeta-function.<br />

For α = δ = (0, . . . , 0), the function V n,k (λ; α, δ) denes the number of<br />

the representations of λ, λ ∈ N, as a sum of n kth powers of non-negative<br />

integers. A. Vinogradov obtained [7] the asymptotic formula<br />

∑<br />

mN<br />

V n,n (m; 0, 0) = P n<br />

(N 1 n<br />

)<br />

+ O<br />

(<br />

N 3 4 log 2 N<br />

where P n (u) is a polynomial of degree n with coecients depending on u.<br />

C. Hooley (see, [2][4]) considered the problem on the representation of<br />

positive integers as the sums of two or four h-powers. The Hooley method<br />

is very specic, and it is dicult to apply it in general case. Using a result<br />

of I. Vinogradov [8] on the famous Waring problem, S. Salerno studied [6]<br />

∑<br />

the distribution of square-free numbers n represented as r x k i in arithmetic<br />

i=1<br />

progression n ≡ 0( mod q) under condition r > (4+ε)k 2 log k, k > k 0 is large<br />

enough.<br />

In this paper, we study the function ζ n (s; α, δ), in particular, its behaviour<br />

in the region 1 2 Re s 1, and applying the Waring zeta-function, we<br />

construct the asymptotic formula for the summatory function of the number<br />

of representations of natural numbers as a sum of n summands of kth powers<br />

of non-negative integers.<br />

In the proof of the main theorem, we shall use the following statement.<br />

Let<br />

ζ(s; 0, δ) =<br />

∞∑<br />

(n + δ) −s , Re s > 1,<br />

n=0<br />

denote the Hurwitz zeta-function.<br />

)<br />

,


On the representation of integers... 209<br />

Lemma 1. Let s = σ + it, −2 σ 2, and let τ be a complex number with<br />

arg τ = ( π<br />

2 − |t|−1) sgn (Im s). Then there exists a constant t 0 > 1 such that<br />

the following approximate equation<br />

ζ(s; 0, δ) = 1 2<br />

where<br />

∑<br />

n∈Z<br />

|n+α|x log x<br />

+π − 1−2s<br />

2<br />

+π − 1−2s<br />

2<br />

F (s; |n + δ|τ 1 2<br />

√ π) + an F (1 − s; |n + δ|τ 1 2<br />

√ π)<br />

|n + δ| s<br />

Γ ( 1<br />

2<br />

(1 − s)) ∑<br />

2Γ( 1 2 s)<br />

n∈Z<br />

|n|y log y<br />

Γ ( 1<br />

2 (1 − s) + 1 ∑<br />

2<br />

2Γ( 1 2 s + 1 2 )<br />

+O ( x −M) + O ( y −M) ,<br />

a n =<br />

)<br />

n∈Z<br />

|n|y log y<br />

{<br />

sgn(n) if n ≠ 0,<br />

1 if n = 0,<br />

F (1 − s; |n|τ − 1 2<br />

√ π)e<br />

2πinδ<br />

|n| 1−s<br />

b n F (−s; |n|τ − 1 2<br />

√ π)<br />

|n| 1−s<br />

b n = −isgn(n)e 2πinδ ,<br />

√<br />

x = |t| 1 2 ( 2|τ|) −1 ,<br />

√<br />

y = |t| 1 2 |τ|( 2) −1 ,<br />

and M > 0 is an arbitrary constant, holds. Moreover, uniformly in all parameters<br />

t, τ, x, y, we have<br />

(<br />

F (ω, z) = l + O exp<br />

(<br />

− |z|2<br />

|t|<br />

)( ) |z| Re ω (<br />

1 +<br />

1<br />

|t| 1 ∣<br />

2<br />

2 |t| 1 |z| 2<br />

) −1 )<br />

2 − |t| 1 ∣ ,<br />

2<br />

where l = 1 if |n + δ| x and |n| y, and l = 0 otherwise; for the integral<br />

representation of F (ω, z), see [5].<br />

The lemma follows easily from the Lavrik work [5].<br />

Lemma 2. Let p be prime, n, k ∈ N, (p, k) = 1. Then the following estimates<br />

hold.<br />

∣<br />

p−1<br />

∑<br />

e 2πi m 1 x 1 +...+mnxn<br />

p<br />

x 1 ,...,x n =0<br />

n∑<br />

x k i ≡a( mod p) i=1<br />

<br />

∣<br />

{(k − 1) n p n−1<br />

2 if n = 3, 4,<br />

(k − 1) n p n 2 if n > 4,


210 O. Savastru, P. Varbanets<br />

Proof. The case n = 3 was considered by Hooley [3]. The case n = 4<br />

can be investigated similarly, and the general case follows from the Deligne<br />

work [1].<br />

Corollary. Let (k, q) = 1. Then, for (m 1 , . . . , m n , q) = 1, we have<br />

∑<br />

e 2πi m 1 x 1 +...+mnxn<br />

q<br />

x 1 ,...,x n ( mod q)<br />

n∑<br />

x k i ≡a( mod q) i=1<br />

≪ q κ τ(q),<br />

where κ = n−1<br />

2<br />

if n = 3, 4, κ = n 2<br />

if n 5, and the implied constant depends<br />

only on k and n.<br />

2. Analytic continuation of ζ n,k (s; α, δ)<br />

For λ > 0, the following equality<br />

λ −s =<br />

πs<br />

Γ(s)<br />

∫ ∞<br />

is true. Thus, we obtain that, for Re s > 1,<br />

ζ n,k (s; α, δ) =<br />

πs<br />

Γ(s)<br />

0<br />

∫ ∞<br />

0<br />

x s−1 e −πxλ dx, Re s > 0,<br />

x s−1 ( ∑<br />

λ>0<br />

V n,k (λ; α, δ)e −πxλ )<br />

dx. (1)<br />

Let σ l (α) denote an arbitrary collection of l elements of the set {α 1 , . . . ,<br />

α n } (the elements α i dier not by values but by indices). Let ∑ (α) be the<br />

l<br />

set of all collections of σ l (α). It is clear that ∑ (α) contains ( n<br />

) l<br />

elements.<br />

l<br />

So, from the denitions of V n,k (λ; α, δ) and Θ k (x; α 0 , δ 0 ) we obtain that<br />

∑<br />

V n,k (λ; α, δ)e −πxλ =<br />

λ>0<br />

=<br />

n∑ ∑<br />

∏<br />

l=1 σ l (α) α i ∈σ l (α)<br />

n∏<br />

j=1<br />

Θ k (x; α i , δ i )<br />

(Θ k (x; α j , δ j ) + 1) − 1, (2)<br />

where ∑ means that the sum runs over all m of the form m = (m 1 +<br />

m>0<br />

δ 1 ) k + . . . + (m n + δ n ) k , and ∑<br />

∑<br />

denotes . Here we take into<br />

σ l (α)<br />

σ l (α)∈ ∑ (α)<br />

l


On the representation of integers... 211<br />

account that the collections (α i1 , . . . , α il ) and (δ i1 , . . . , δ il ) are found in the<br />

one-to-one correspondence.<br />

Since the terms of the series for the Θ k decrease exponentially, we have<br />

that the behaviour of the integrand in (1) on [1, ∞) guarantee an analytic<br />

continuation of this part of integral to the whole complex plane. Therefore,<br />

it remains to consider the integral<br />

where<br />

F (s) =<br />

=<br />

=<br />

∫ 1<br />

0<br />

I(σ l (α)) =<br />

x s−1 ( ∑<br />

λ>0<br />

n∑ ∑<br />

∫ 1<br />

l=1 σ l (α) 0<br />

V n,k (λ; α, δ)e −πxλ )<br />

dx<br />

x s−1<br />

∏<br />

α i ∈σ l (α)<br />

Θ k (x; α i , δ i )dx<br />

n∑ ∑<br />

I(σ l (α)), (3)<br />

l=1 σ l (α)<br />

∫ 1<br />

0<br />

x s−1<br />

∏<br />

α i ∈σ l (α)<br />

Θ k (x; α i , δ i )dx. (4)<br />

In order to investigate the behaviour of the function Θ n,k (x; α i , δ i ) on the<br />

right of the point x = 0, we use the integral representation for this function.<br />

Consider the Mellin pair<br />

Γ(z) =<br />

∫ ∞<br />

0<br />

e −y y z−1 dy, e −y = 1<br />

2πi<br />

∫<br />

u+i∞<br />

u−i∞<br />

y −z Γ(z)dz, Re z = u > <strong>0.</strong><br />

Taking y = πx(m + δ i ) k , u = 1 n<br />

+ ε, ε > 0, we arrive at the relation<br />

ε+ 1 n +i∞ ∫<br />

Θ k (x; α 0 , δ 0 ) = 1 (πx) −z Γ(z)ζ(kz; α 0 , δ 0 )dz, (5)<br />

2πi<br />

ε+ 1 n −i∞<br />

where ζ(w; α 0 , δ 0 ), 0 < δ 1, is the Lerch zeta-function dened, for Re w > 1,<br />

by the Dirichlet series<br />

ζ(w; α 0 , δ 0 ) =<br />

∞∑<br />

m=0<br />

e 2πiα 0m<br />

(m + δ 0 ) w .


212 O. Savastru, P. Varbanets<br />

The function ζ(w; α 0 , δ 0 ) is entire, except for a simple pole at w = 1 with<br />

residue 1 if α 0 ∈ Z.<br />

In equality (5), move the contour of integration to the line Re z = −1+<br />

2k.<br />

1<br />

The integrand has a simple pole at z = 1 k<br />

(if α 0 ∈ Z), and a simple pole at<br />

z = <strong>0.</strong> Thus, we obtain that<br />

Θ k (x; α 0 , δ 0 ) = ε(α 0 ) Γ( 1 k )<br />

(<br />

+ ζ(0; α<br />

(πx) 1 0 , δ 0 ) + I − 1 + 1 )<br />

k<br />

2k ; x, α 0, δ 0 , (6)<br />

where<br />

I(b; x, α 0 , δ 0 ) = 1<br />

2πi<br />

∫<br />

b+i∞<br />

b−i∞<br />

{ 0 if α<br />

ε(α 0 ) =<br />

0 ∉ Z,<br />

1 if α 0 ∈ Z.<br />

Since, for x → +0, ∫∞<br />

I(b; x, α 0 , δ 0 ) ≪ x −b<br />

≪<br />

1<br />

∫∞<br />

x −b<br />

≪ x −b ,<br />

we obtain from (6)(7) that<br />

Θ k (x; α 0 , δ 0 ) = ε(α 0 ) Γ( 1 k )<br />

1<br />

(πx) −z Γ(z)ζ(kz; α 0 , δ 0 )dz, (7)<br />

|Γ(b + iv)||ζ(kb + ikv; α 0 , δ 0 )|dv<br />

|Γ(b + iv)|v 1 2 −b |ζ(1 − kb − ikv; −δ 0 , α 0 )|dv<br />

(πx) 1 k<br />

Now from (3), (4) and (8) we have that<br />

where<br />

F (s) =<br />

=<br />

n∑ ∑<br />

∫ 1<br />

l=1 σ l (α) 0<br />

× ∏<br />

α i ∈σ l (α)<br />

n∑<br />

l=0<br />

b k (l)<br />

s − l k<br />

b k (0) =<br />

x s−1<br />

(<br />

ε(α i ) Γ( 1 k )<br />

(πx) 1 k<br />

(∣ ∣∣∣<br />

+ O Re s + 1<br />

2k ∣<br />

n∑ ∑<br />

l=1 σ l (α) α i ∈σ l (α)<br />

+ ζ(0; α 0 , δ 0 ) + O(x 1− 1<br />

2k ). (8)<br />

+ ζ(0; α i , δ i ) + O ( x 1− 1<br />

2k<br />

) ) dx<br />

∏<br />

−1)<br />

, (9)<br />

ζ(0; α i , δ i ),


On the representation of integers... 213<br />

b k (n) =<br />

r =<br />

( ( ) ) 1 r<br />

Γ (π) − 1 k ,<br />

k<br />

n∑<br />

i=1<br />

α i =1<br />

1, (10)<br />

and b k (l), 0 < l < n, are polynomials in Γ( 1 k<br />

) with coecients which depend<br />

on α 1 , . . . , α n and δ 1 , . . . , δ n .<br />

Equality (9) gives meromorphic continuation of the Waring zeta-function<br />

to the half-plane Re s −1 +<br />

2k. 1 In view of (9), we see that ζ n,k (s; α, δ) has<br />

poles at the points s = k, l l = 0, 1, . . . , n.<br />

Now we will estimate the function ζ n,k (s; 0, δ) for 1 2<br />

< Re s 1 + ε. In<br />

(1), the integrand is holomorphic for Re s > <strong>0.</strong> Therefore, if |t| 1, we can<br />

turn by the angle ϕ = ( π<br />

2 − |t|−1) sgn(t) the ray of integration x ∈ [0, +∞).<br />

Hence, we have<br />

πs<br />

ζ n,k (s; α, δ) =<br />

Γ(s)<br />

∫ ∞<br />

x=0<br />

z s−1 ( ∑<br />

λ>0<br />

V n,k (λ; α, δ)e −πzλ )<br />

dz, (11)<br />

z = xe iϕ . This choice of the ray of integration shows that the Waring zetafunction<br />

has the power growth for |t| → ∞. We put<br />

ζ n,k (s; α, δ) =<br />

def<br />

π s<br />

Γ(s)<br />

= ζ (1)<br />

n,k<br />

⎛<br />

⎝<br />

∫ 1<br />

Now, using the well-known estimate<br />

we nd that<br />

ζ (2)<br />

n,k (s; α, δ) = ∑ λ>0<br />

0<br />

I λ (s) = 1<br />

Γ(s)<br />

By a trivial estimate ∑<br />

deduce from (14) that<br />

+<br />

∫ ∞<br />

1<br />

⎞ (<br />

⎠<br />

z s−1 ∑ λ>0<br />

V n,k (λ; α, δ)e −πzλ )<br />

dz<br />

(s; α, δ) + ζ(2)<br />

n,k<br />

(s; α, δ). (12)<br />

∫ ∞<br />

x=λ<br />

z=xe iϕ<br />

V n,k (λ; α, δ)<br />

m s I m (s) ≪ ∑ λ>0<br />

λx<br />

∫∞<br />

ζ (2)<br />

n,k (s; α, δ) ≪ ∑<br />

V n,k (λ; α, δ) ≪ x n k<br />

λ 0<br />

λx<br />

z s−1 e −z dz ≪ e − λ<br />

|t|<br />

, (13)<br />

V n,k (λ; α, δ)<br />

λ σ e − λ<br />

|t|<br />

. (14)<br />

and partial summation, we<br />

(<br />

V n,k (λ; α, δ)<br />

λ σ e − x<br />

|t|<br />

1 + x ) dx<br />

|t| x σ+1


214 O. Savastru, P. Varbanets<br />

≪<br />

min<br />

(<br />

λ n k −σ<br />

)<br />

0<br />

n<br />

k − σ , |t| n k −σ log |t| + |t| n k −σ , (15)<br />

λ 0<br />

λ 0 = δ k 1 + . . . + δk n. Hence, in view of (3)(5), we obtain that, for Re s > 1 2,<br />

where<br />

n∑<br />

ζ (1)<br />

n,k<br />

(s; α, δ) =<br />

∑<br />

l=1 σ l (α)<br />

π s<br />

Γ(s)<br />

∫ 1<br />

x=0<br />

x s−1 e iϕs G(x, σ l (α))dz,<br />

and<br />

G(x, σ l (α)) =<br />

∏<br />

α j ∈σ l (α) 1<br />

1<br />

2k +i∞ ∫<br />

ζ (1)<br />

n,k (s; α, δ) = πs e iϕs<br />

Γ(s)<br />

2k −i∞ e iw j<br />

(πx) −w j<br />

Γ(w j )ζ(kw j ; α j , δ j )dw j ,<br />

×<br />

l∏<br />

j=1<br />

n∑ ∑<br />

∫<br />

∫<br />

· · ·<br />

l=1 σ l (α)<br />

Re w j = 1<br />

2k<br />

1jl<br />

e iϕw j<br />

Γ(w j )<br />

ζ(kw j ; α j , δ j ) dw 1 . . . dw l<br />

s − w 1 − w l<br />

. (16)<br />

The integrals on w j , j = 1, . . . , l, can be calculated by the method of stationary<br />

phase using the Stirling formula for Γ(w) and the approximate functional<br />

equation for ζ(s; α, δ), see Lemma 1. We have<br />

n∑<br />

ζ (1)<br />

n,k<br />

(s; α, δ) ≪<br />

∑<br />

∑<br />

l=1 σ l (α) α∈σ l (α)<br />

∑<br />

V l,k (λ; α, δ)<br />

(17)<br />

(σ − 1 λλ 0<br />

2 ) + ||t| − λ|.<br />

The main contribution in (16) is given by the summand with l = n , and<br />

then σ l (α) = (α 1 , . . . , α l ), λ 0 = δ k 1 + . . . + δk n. Hence, and from (12), (15)<br />

and (17) we obtain that, for Re s > 1 2,<br />

( )<br />

ζ n,k (s; α, δ) ≪ |t| n 1<br />

k −σ min<br />

n<br />

k − σ , log(|t| + 3)<br />

+ ∑ V n,k (m; α, δ) λ<br />

(σ − 1 m 2<br />

) + ||t| − m|e−<br />

|t|<br />

. (18)


On the representation of integers... 215<br />

3. Sum of n kth powers in arithmetic progression<br />

Recall that V n,k (m) denotes the number of representations of a natural<br />

m as a sum of n kth powers of non-negative integers, i.e.,<br />

∑<br />

V n,k (m) =<br />

1. (1)<br />

x 1 ,...,x n0<br />

x k 1 +...+xk n=m<br />

Since we have not an acceptable estimate for V n,k (m), we cannot use the<br />

lemma on partial sums of Dirichlet series. Therefore, we apply the Perron<br />

formula for calculation of the weighted mean for V n,k (m). We have that<br />

∑<br />

m≡a( mod q)<br />

mN<br />

V n,k (m)(1 − m N ) = 1<br />

2πi<br />

where δ = ( l 1<br />

q<br />

, . . . , ln q ), ∑<br />

n∑<br />

li k<br />

i=1<br />

≡ a( mod q).<br />

(a,q)<br />

c+i∞ ∫<br />

c−i∞<br />

1 ∑<br />

q ks<br />

(a,q)<br />

N s<br />

ζ n,k (s; 0, δ) ds, (2)<br />

s(s + 1)<br />

denotes a summing over l 1 , . . . , l n ( mod q), and<br />

We consider only the case k 2 < n k because for n k 2<br />

we cannot dene<br />

the main term in the asymptotic formula.<br />

By (14) and (16) it is clear that we can move the contour of integration<br />

in (2) to the line Re s = σ 0 > 2. 1 This gives<br />

∑<br />

m≡a( mod q)<br />

mN<br />

V n,k (m)<br />

= N n,k (a, q) ∑<br />

+ 1<br />

2πi<br />

∫<br />

σ 0 +i∞<br />

(<br />

1 − m N<br />

)<br />

( ) l<br />

N k<br />

a n,k (l)<br />

q k k<br />

2


216 O. Savastru, P. Varbanets<br />

Let, for Re s > 1,<br />

Obviously,<br />

Z n,k (s; a, q) =<br />

∞∑<br />

m=1<br />

m≡a( mod q)<br />

Z n,k (s; a, q) = 1<br />

q ks ∑<br />

(a,q)<br />

V n,k (m)<br />

m s .<br />

ζ n,k (s; 0, δ).<br />

Then, similarly to the proof of relation (12), we can obtain that, for Re s 1 2,<br />

|Im s| > 1,<br />

where<br />

and<br />

Z n,k (s; a, q) = Z (1)<br />

n,k<br />

(s; a, q) + Z(2)<br />

n,k<br />

(s; a, q), (4)<br />

∫ 1<br />

Z (1)<br />

πs ∑<br />

n,k<br />

(s; a, q) = x s−1 e iϕs q −ks<br />

Γ(s)<br />

(a,q) 0<br />

n∑ ( n<br />

) ( ( ) 1<br />

× Γ (πxe iϕ ) − 1 k<br />

l k<br />

l=1<br />

( )) 1<br />

l<br />

+I<br />

2k ; x, 0, δ i dx, (5)<br />

Z (2)<br />

∞ n,k<br />

(s; a, q) =<br />

∑<br />

m=1<br />

m≡a( mod q)<br />

V n,k (m)<br />

m s I m (s). (6)<br />

The denitions of I(b; x, α 0 , δ 0 ) and I m (s) are given in (7) and (13), respectively.<br />

After some simple transformations of the integrand in (5), we obtain that<br />

Z (1)<br />

πs<br />

n,k<br />

(s; a, q) =<br />

Γ(s) eiϕs q −ks<br />

( π<br />

s<br />

+O<br />

Γ(s)<br />

( ∑<br />

×<br />

(a,q)<br />

n∑ a n,k (l)<br />

s − l N n,k (a, q)<br />

l=1 k<br />

n∑<br />

∫ 1<br />

∣ x s−1 e iϕs q −ks<br />

l=1<br />

0<br />

l∏<br />

( 1<br />

I<br />

2k ; x, 0, l ))<br />

)<br />

i<br />

dx<br />

q ∣<br />

i=1<br />

. (7)


On the representation of integers... 217<br />

The main contribution in the error term in (7) is given by the summand with<br />

l = n. In order to estimate the integral<br />

Φ(s) =<br />

∫ 1<br />

0<br />

π s e iϕs<br />

Γ(s) xs−1 q −ks ⎛<br />

we use (7). This leads to the estimate<br />

where<br />

∫<br />

Φ(s) ≪<br />

∣<br />

∫<br />

· · ·<br />

Re w j = 1<br />

2k<br />

1jn<br />

q −ks πs e iϕs<br />

Γ(s)<br />

∑ ∑<br />

(ζ) = ζ<br />

(a,q)<br />

⎝ ∑ (a,q)<br />

n∏<br />

i=1<br />

Γ(w 1 ) . . . Γ(w n )<br />

π (w 1+...+w n )<br />

( 1<br />

I<br />

2k ; x, 0, l ) ⎞<br />

i<br />

⎠ dx<br />

q<br />

(<br />

kw 1 ; 0, l ) (<br />

1<br />

. . . ζ kw n ; 0, l )<br />

n<br />

.<br />

q<br />

q<br />

∑<br />

(ζ)<br />

dw 1 . . . dw n<br />

s − w 1 − w n<br />

∣ ∣∣∣∣∣∣∣∣∣<br />

, (8)<br />

Applying the approximate functional equation for the Hurwitz zeta-function<br />

ζ(kw; 0, δ) with |τ| = |t| − 1 2 log −1 (|t| + 3), see Lemma 1, we easily nd that<br />

Φ(s)<br />

≪<br />

∞∑<br />

m j =1<br />

1jn<br />

∫<br />

× ∣<br />

∣ eiϕs<br />

∣∣ ∑ Γ(s)<br />

(a,q)<br />

∫<br />

· · ·<br />

Re w j = 1<br />

2k<br />

1jn<br />

n∏<br />

j=1<br />

e 2πi m 1 l 1 +...+mnln<br />

q<br />

Γ(w j )<br />

π w jm w jk<br />

j<br />

∣<br />

dw 1 . . . dw n<br />

s − w 1 − w n<br />

∣ ∣∣.<br />

Computing the integrals by the method of stationary phase, we obtain<br />

Hence,<br />

Φ(s)<br />

≪<br />

≪<br />

∞∑<br />

m j =1<br />

1jn<br />

∣ ∑ (a,q)<br />

e 2πi m 1 l 1 +...+mnln<br />

q<br />

e − (mk 1 +...+mk n )<br />

|t|<br />

×<br />

(σ − 1 2 ) + ||s| − mk 1 − . . . − mk n| q−kσ<br />

∞∑<br />

m=1<br />

∣<br />

V n,k (m)e − m |t|<br />

(σ − 1 2 ) + ||t| − m| qκ−kσ . (9)<br />

Z (1)<br />

n,k (s; a, q) = N n,k(a, q) πs<br />

Γ(s) eiϕs q −ks<br />

n∑ a n,k (l)<br />

l=1<br />

s − l k<br />

N l q


218 O. Savastru, P. Varbanets<br />

+<br />

Moreover, using the estimates<br />

and<br />

∑<br />

m≡a( mod q)<br />

mX<br />

V n,k (m) = ∑ (a,q)<br />

∑<br />

m≡a( mod q)<br />

mX<br />

∞∑<br />

m=1<br />

V n,k (m)e − m |t|<br />

(σ − 1 2 ) + ||t| − m| qκ−kσ . (10)<br />

∑<br />

m i<br />

∑ (mi +δ i ) k <br />

V n,k (m) e− m |t|<br />

we obtain by partial summation that<br />

Z (2)<br />

n,k<br />

(<br />

) 1<br />

k<br />

X<br />

q k<br />

(<br />

n<br />

X<br />

)<br />

k<br />

1 ≪ N n,k (a, q)<br />

q n + 1<br />

m σ ≪ q n k −σ if |t| q, (11)<br />

(s; a, q)<br />

⎧<br />

⎨q n k −σ if |t| q,<br />

≪ ( n<br />

⎩ |t|<br />

−σ<br />

)<br />

k<br />

q<br />

+ |t| −σ min ( 1<br />

n n<br />

k −σ , log(|t| + 3)N n,k(a, q) ) if |t| < q.<br />

Thus, taking σ = 1 2 + 1<br />

log N<br />

, we obtain by (4), (10) and (12) that<br />

∑<br />

m≡a( mod q)<br />

mN<br />

(<br />

V n,k (m) 1 − m )<br />

N<br />

= N n,k (a, q)<br />

(<br />

+O<br />

n∑<br />

l=1<br />

a n,k (l)<br />

)<br />

q l N l q + O<br />

(N 1 n<br />

2 q k − 1 2 log 2 N<br />

)<br />

N n,k (a, q)N 1 2 q<br />

− 3 2 log 2 N<br />

)<br />

+O<br />

(N 1 2 q<br />

κ− k 2 τ(q) log N<br />

(12)<br />

, (13)<br />

where a n,k (n) =<br />

(<br />

1<br />

π Γ( 1 k ) ) n<br />

, the constants a n,k (l), 1 l n, are computable,<br />

and the O-constants do not depend on N and q. From this, by standard way,<br />

we obtain the following statement.<br />

Theorem. Let V n,k (m) denote the number of the representations of m as the<br />

sum of n kth powers, and let a, q, N be natural numbers. Then the following<br />

asymptotic formula


On the representation of integers... 219<br />

∑<br />

m≡a( mod q)<br />

mN<br />

V n,k (m)<br />

= N n,k (a, q)<br />

+O<br />

n∑ a n,k (l)<br />

N l q + O<br />

(N 3 2n−k<br />

4 q 2k<br />

q l<br />

l=1 )<br />

(N 3 4 q<br />

κ− k 2 τ(q) log N<br />

)<br />

− 1<br />

2 log 2 N<br />

(<br />

)<br />

+ O N n,k (a, q)N 3 4 q<br />

− 3 2 log 2 N<br />

holds. Here N n,k (a, q) = #sol{x k 1 + . . . + xk n ≡ a( mod q), 0 x i q − 1},<br />

{ n−1<br />

κ = 2<br />

if n = 3, 4,<br />

n<br />

2<br />

if n 5,<br />

and τ(q) is the number of divisors of q.<br />

References<br />

[1] P. Deligne, La conjecture de Weil. I, II, IHES Publ. Math. 43 (1974), 273307,<br />

52 (1980), 137252.<br />

[2] C. Hooley, On the representation of a number as the sum of four cubes. I,<br />

Proc. London Math. Soc.(3) 36 (1978), 11714<strong>0.</strong><br />

[3] C. Hooley, On the numbers that are representable as the sum of two cubes. I,<br />

Reine Angew. Math. 314 (1980), 146173.<br />

[4] C. Hooley, On another sieve method and the numbers that are a sum of two<br />

hth powers, Proc. London Math. Soc. (3) 43 (1981), 73109.<br />

[5] A. Lavrik, Approximate functional equation for the Dirichlet L-functions, Tr.<br />

Mosk. Mat. O.-va 18 (1968), 91104 (in Russian).<br />

[6] S. Salerno, On the distribution of x k 1 + . . . + x k s in the arithmetic progression,<br />

Studio Scient. Math. Hung. 20 (1985), 357374.<br />

[7] A. I. Vinogradov, Waring zeta-function, in: Collection of Papers Dedicated<br />

to the 60th Birthday of Prof. V. Sprindzuk, 3340, 1997 (in Russian).<br />

[8] I. M. Vinogradov, The Method of Trigonometric Sums in the Theory of Numbers,<br />

Interscience, London, 1954.<br />

Received<br />

31 January 2008

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!