19.12.2012 Views

Almost-all results on the p^\lambda problem

Almost-all results on the p^\lambda problem

Almost-all results on the p^\lambda problem

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

<str<strong>on</strong>g>Almost</str<strong>on</strong>g>-<str<strong>on</strong>g>all</str<strong>on</strong>g> <str<strong>on</strong>g>results</str<strong>on</strong>g> <strong>on</strong> <strong>the</strong> p λ <strong>problem</strong> II ∗<br />

Stephan Baier<br />

01.11.04<br />

Abstract:<br />

Suppose that 1/2 ≤ λ < 1. Balog and Harman proved that for any real θ <strong>the</strong>re exist<br />

infinitely many primes p satisfying �<br />

pλ − θ �<br />

< p−(1−λ)/2+ε (with an asymptotic result).<br />

In <strong>the</strong> present paper we establish that for almost <str<strong>on</strong>g>all</str<strong>on</strong>g> θ in <strong>the</strong> interval 0 ≤ θ < 1 <strong>the</strong>re<br />

exist infinitely many primes p such that �<br />

pλ − θ �<br />

< p− min{(2−λ)/6,(14−9λ)/32}+ε . Thus we<br />

obtain a better result for almost-<str<strong>on</strong>g>all</str<strong>on</strong>g> θ than for a single θ if λ > 1/2.<br />

1 Introducti<strong>on</strong><br />

In [1], [2], [3], [4], [5], [6], [8], [9], Balog, Harman and <strong>the</strong> author of <strong>the</strong> present paper<br />

studied <strong>the</strong> distributi<strong>on</strong> of fracti<strong>on</strong>al parts of p λ , p running over <strong>the</strong> prime numbers and<br />

λ being a given real number lying in <strong>the</strong> interval 0 < λ < 1. In <strong>the</strong> sequel, we refer to<br />

this questi<strong>on</strong> as <strong>the</strong> p λ <strong>problem</strong>. Balog’s and Harman’s result for <strong>the</strong> range 1/2 ≤ λ < 1<br />

(see [4], [6, Theorem 2] and [8]) can be formulated as follows.<br />

Theorem 1: Suppose that B > 0, λ ∈ [1/2, 1) and a real θ are given. Then for any<br />

fixed τ in <strong>the</strong> range 0 ≤ τ < (1 − λ)/2 we have<br />

as N → ∞.<br />

�<br />

n ≤ N,<br />

� n λ − θ � < n −τ<br />

Λ(n) =<br />

N 1−τ<br />

1 − τ<br />

+ O<br />

� N 1−τ<br />

(log N) B<br />

Here, as in <strong>the</strong> following, Λ(n) denotes <strong>the</strong> v<strong>on</strong> Mangoldt functi<strong>on</strong>.<br />

It is an interesting questi<strong>on</strong> whe<strong>the</strong>r sharper <str<strong>on</strong>g>results</str<strong>on</strong>g> than Theorem 1 hold true for<br />

almost <str<strong>on</strong>g>all</str<strong>on</strong>g> θ. In [2] <strong>the</strong> author already established almost-<str<strong>on</strong>g>all</str<strong>on</strong>g> <str<strong>on</strong>g>results</str<strong>on</strong>g> <strong>on</strong> <strong>the</strong> p λ <strong>problem</strong><br />

for 0 < λ < 1/2. He obtained <strong>the</strong> following result by using density <strong>the</strong>orems for <strong>the</strong><br />

n<strong>on</strong>-trivial zeta zeros.<br />

∗ Ma<strong>the</strong>matics Subject Classificati<strong>on</strong> (2000): 11N05, 11L20<br />

1<br />

�<br />

(1)


Theorem 2: Let B > 0 and λ ∈ (0, 1/2) be given. Define<br />

⎧<br />

⎪⎨ 5/12 − λ/6 if λ ≤ 1/8,<br />

H(λ) := � �<br />

⎪⎩ 1 + λ/2 �2<br />

/3 − λ if 1/8 < λ ≤ 1/2.<br />

Then <strong>the</strong> asymptotic formula (1) holds true for almost <str<strong>on</strong>g>all</str<strong>on</strong>g> real θ with respect to <strong>the</strong><br />

Lebesgue measure if τ is any fixed real number lying in <strong>the</strong> range 0 ≤ τ < H(λ).<br />

In <strong>the</strong> present paper we use a completely different approach to prove <strong>the</strong> following<br />

almost-<str<strong>on</strong>g>all</str<strong>on</strong>g> result for <strong>the</strong> range 1/2 ≤ λ < 1.<br />

Theorem 3: Let B > 0 and λ ∈ [1/2, 1) be given. Define<br />

H(λ) := min{(2 − λ)/6, (14 − 9λ)/32}.<br />

Then <strong>the</strong> asymptotic formula (1) holds true for almost <str<strong>on</strong>g>all</str<strong>on</strong>g> real θ with respect to <strong>the</strong><br />

Lebesgue measure if τ is any fixed real number lying in <strong>the</strong> range 0 ≤ τ < H(λ).<br />

We note that <strong>the</strong> functi<strong>on</strong> H(λ) defined in Theorems 2 and 3 is c<strong>on</strong>tinuous <strong>on</strong> <strong>the</strong><br />

interval (0, 1) and that for λ0 = 1/2 we have H(λ0) = 1/4 = (1 − λ0)/2. For λ0 = 1/2<br />

Theorem 3 <strong>the</strong>refore does’nt provide any improvement bey<strong>on</strong>d <strong>the</strong> result of Theorem 1.<br />

However, for any λ ∈ (1/2, 1) <strong>the</strong> τ-range in Theorem 3 is wider than that in Theorem<br />

1.<br />

In [3] we improved <strong>the</strong> result of Theorem 3 for single θ’s in <strong>the</strong> range 59/85 =<br />

0.694... < λ < 1. There we proved that in this range <strong>the</strong> exp<strong>on</strong>ent −(1 − λ)/2 + ε in<br />

Theorem 1 may be replaced by −G(λ) + ε, where<br />

G(λ) := min{max{(35 − 22λ)/129, 1/7}, 5/18 − λ/6}.<br />

Even this result is exceeded by that of Theorem 3 for almost-<str<strong>on</strong>g>all</str<strong>on</strong>g> θ since in <strong>the</strong> range<br />

59/85 < λ < 1 we have H(λ) > G(λ).<br />

In a similar manner like [2, Theorem 2] was derived from [2, Propositi<strong>on</strong>], we sh<str<strong>on</strong>g>all</str<strong>on</strong>g><br />

later derive <strong>the</strong> asserti<strong>on</strong> of Theorem 3 from<br />

Theorem 4: Let <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s of Theorem 3 be kept, and suppose that C > 0.<br />

Then we have<br />

as N → ∞.<br />

�1<br />

0<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

n ≤ N,<br />

� n λ − θ � < n −τ<br />

Λ(n) −<br />

N 1−τ<br />

1 − τ<br />

�<br />

�2<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

dθ ≪ N 2(1−τ )<br />

(log N) C<br />

Acknowledgements. This research has been supported by a Marie Curie Fellowship<br />

of <strong>the</strong> European Community programme “Improving <strong>the</strong> Human Research Potential<br />

and <strong>the</strong> Socio-Ec<strong>on</strong>omic Knowledge Base” under c<strong>on</strong>tract number HPMF-CT-2002-<br />

02157.<br />

2<br />

(2)


2 Reducti<strong>on</strong> to exp<strong>on</strong>ential sums<br />

We sh<str<strong>on</strong>g>all</str<strong>on</strong>g> employ a similar method like in [3], where we linked <strong>the</strong> p λ <strong>problem</strong> with<br />

Piatetski-Shapiro primes, that is, primes of <strong>the</strong> form [n c ] with fixed c > 1 and n ∈ �.<br />

A standard procedure for attacking <strong>problem</strong>s related to Piatetski-Shapiro primes is as<br />

follows (see [11] for instance): Firstly, <strong>on</strong>e reduces <strong>the</strong> sum of Λ(n) in questi<strong>on</strong> (in our<br />

paper, this is <strong>the</strong> sum <strong>on</strong> <strong>the</strong> left-hand side of (1)) to exp<strong>on</strong>ential sums of <strong>the</strong> form<br />

�<br />

n∼N<br />

Λ(n)e �<br />

hn λ�<br />

.<br />

Sec<strong>on</strong>dly, <strong>on</strong>e breaks up <strong>the</strong>se exp<strong>on</strong>ential sums into certain bilinear exp<strong>on</strong>ential sums<br />

(so-c<str<strong>on</strong>g>all</str<strong>on</strong>g>ed type I and type II sums) by using Vaughan-type identities. Fin<str<strong>on</strong>g>all</str<strong>on</strong>g>y, <strong>on</strong>e estimates<br />

<strong>the</strong>se bilinear exp<strong>on</strong>ential sums in a n<strong>on</strong>-trivial way. Here we also follow this<br />

approach. However, since <strong>the</strong> first two steps were completely carried out in several papers<br />

(see [11] for example), we omit some details in <strong>the</strong>se steps.<br />

Throughout <strong>the</strong> following, we suppose that λ, τ and t are given real numbers satisfying<br />

1/2 ≤ λ < 1 and 0 ≤ τ < t < 1, and we put<br />

H := N t .<br />

The parameter t will later be choosen suitably subject to λ and τ. Fur<strong>the</strong>rmore, we<br />

suppose that η is a (sm<str<strong>on</strong>g>all</str<strong>on</strong>g>) positive real number. By <strong>the</strong> notati<strong>on</strong> k ∼ K we mean k to<br />

lie in some interval K1 ≤ k ≤ K2 with K/2 ≤ K1 ≤ K2 ≤ 2K.<br />

Using <strong>the</strong> Fourier series expansi<strong>on</strong> of {x}, we obtain<br />

Lemma 1: Suppose that (ah) is a sequence of complex numbers. Assume that we<br />

uniformly have, as N → ∞,<br />

�<br />

ah<br />

�<br />

1≤h≤H N


�<br />

1 − e � ��<br />

−τ<br />

hx ≪ hx −τ−1 ,<br />

∂<br />

∂x<br />

and taking into c<strong>on</strong>siderati<strong>on</strong> that <strong>the</strong> O-c<strong>on</strong>stant in (1) depends <strong>on</strong>ly <strong>on</strong> <strong>the</strong> O-c<strong>on</strong>stant<br />

in (3).<br />

Lemma 2: We have (2) if<br />

for a suitable η > 0.<br />

sup<br />

N


We sh<str<strong>on</strong>g>all</str<strong>on</strong>g> prove <strong>the</strong> following lemma by adapting a method of Heath-Brown (see [11,<br />

page 257]).<br />

Lemma 4: Let Qh be any positive integer. Then<br />

|Lh| 2 �<br />

≪ QhX �<br />

hX λ Y λ Q −1<br />

�1/2 �<br />

h Q −1<br />

h Y 2 + Y �<br />

Then<br />

Proof: For q ∈ � we define <strong>the</strong> interval Iq by<br />

Iq := �<br />

Y λ (q − 1)/Qh, Y λ �<br />

q/Qh .<br />

|Lh| ≪ �<br />

q≤2Qh<br />

�<br />

m∼X<br />

| �<br />

n ∼ Y,<br />

n λ ∈ Iq,<br />

mn ∼ N<br />

�<br />

2−λ<br />

+ QhX h −1 Y 2−λ + Y X λ��<br />

(log N).<br />

bne �<br />

hm λ n λ�<br />

|.<br />

From that, using <strong>the</strong> Cauchy-Schwarz inequality, we obtain<br />

|Lh| 2 ≪ QhX � �<br />

| �<br />

bne �<br />

hm λ n λ�<br />

| 2<br />

q≤2Qh<br />

≪ QhX �<br />

q≤2Qh<br />

≪ QhX �<br />

m∼X<br />

�<br />

n ∼ Y,<br />

n λ ∈ Iq<br />

n, r ∼ Y,<br />

|t| ≤ 2Y λ /Qh<br />

n ∼ Y,<br />

nλ ∈ Iq,<br />

mn ∼ N<br />

�<br />

r ∼ Y,<br />

r λ ∈ Iq<br />

| �<br />

m ∼ X,<br />

mn ∼ N,<br />

mr ∼ N<br />

| �<br />

m ∼ X,<br />

mn ∼ N,<br />

mr ∼ N<br />

e �<br />

h(n λ − r λ )m λ�<br />

|<br />

e �<br />

htm λ�<br />

|, (5)<br />

where we put t := n λ − r λ . Applying [11, Lemma 1] to <strong>the</strong> inner sum in <strong>the</strong> last line of<br />

(5), we get<br />

�<br />

m ∼ X,<br />

mn ∼ N,<br />

mr ∼ N<br />

e �<br />

htm λ�<br />

�<br />

≪ min X, (ht) −1 X 1−λ + �<br />

htX λ�1/2 �<br />

≤ �<br />

htX λ� 1/2<br />

+ X min �<br />

1, (ht) −1 X −λ�<br />

. (6)<br />

Combining (5) and (6), and using partial summati<strong>on</strong>, we obtain<br />

where<br />

|Lh| 2 ≪ QhX �<br />

hX λ Y λ Q −1<br />

�1/2 h S �<br />

2Y λ Q −1<br />

�<br />

h +<br />

QhX 2 S �<br />

h −1 X −λ�<br />

+ h −1 QhX 2−λ (log N) max<br />

∆≥h−1X −λ S(∆)∆−1 , (7)<br />

5


Obviously, we have<br />

S(∆) := �<br />

n, r ∼ Y,<br />

|t| ≤ ∆<br />

Combining (7) and (8), we obtain <strong>the</strong> desired estimate. ✷<br />

To optimize <strong>the</strong> estimate in Lemma 4, we choose<br />

1.<br />

S(∆) ≪ ∆Y 2−λ + Y. (8)<br />

Qh := 1 + �<br />

h 1/3 X (λ−2)/3 Y (λ+2)/3�<br />

.<br />

Then, by a short calculati<strong>on</strong> using J ≤ H = N t and XY = N, we obtain<br />

when<br />

Lemma 5: We have<br />

�<br />

h≤H<br />

This implies<br />

|Lh| 2 ≪ �<br />

N 1+λ/2+3t/2 + N 2−λ + N 2+t Y −1 + N 4/3+λ/3+4t/3 Y 1/3 +<br />

N 2/3+2λ/3+5t/3 Y 2/3 + N 4/3−2λ/3+t/3 Y 4/3�<br />

(log N).<br />

Lemma 6: For every sufficiently sm<str<strong>on</strong>g>all</str<strong>on</strong>g> fixed η > 0 we have<br />

provided <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s<br />

are satisfied.<br />

4 Treatment of Kh<br />

�<br />

|Lh|<br />

h≤H<br />

2 ≪ N 2−2η<br />

N t+100η ≤ Y ≤ N 2−λ−4t−100η ,<br />

t < (2 − λ)/3, t > 2(1 − λ)/5 (9)<br />

Heath-Brown established <strong>the</strong> following two estimates for Kh (see [11, pages 261-262]).<br />

Lemma 7: (i) We have<br />

6


�<br />

2<br />

Kh ≪ (log N) N 1−λ h −1 + Xh 1/2 N λ/2�<br />

.<br />

(ii) Let (p, q) be any exp<strong>on</strong>ent pair for which 0 < p ≤ 1/2 ≤ q ≤ 1. Let P be any<br />

positive integer. Then<br />

|Kh| 2 ≪ (log N) 8 (X + NP −1 ) �<br />

N + N (p+1/2)λ Y q−2p−1/2 H p+1/2 P p+3/2 +<br />

N 1−λ/2 Y 1/2 h −1/2 P 1/2 + Y P �<br />

.<br />

Taking <strong>the</strong> exp<strong>on</strong>ent pair (p, q) = (2/7, 4/7), and choosing<br />

P := �<br />

N (2−(1+2p)(λ+t))/(2p+3) Y (4p+1−2q)/(2p+3)�<br />

= �<br />

N (14−11(λ+t))/25 Y 7/25�<br />

,<br />

from Lemma 7, we obtain<br />

Lemma 8: (i) We have<br />

(ii) We have<br />

�<br />

h≤H<br />

�<br />

h≤H<br />

|Kh| 2 �<br />

4<br />

≪ (log N) N 2(1−λ) + X 2 N λ+2t�<br />

.<br />

|Kh| 2 �<br />

8<br />

≪ (log N) N (36+11λ+36t)/25 Y −7/25 + N (43−7λ+18t)/25 Y 9/25 + N 1+t Y +<br />

provided <strong>the</strong> c<strong>on</strong>diti<strong>on</strong><br />

is satisfied.<br />

when<br />

N 2+t Y −1 + N (57−18λ+7t)/25 Y −9/25�<br />

,<br />

From <strong>the</strong> preceding lemma, we derive<br />

t < 14/11 − λ (10)<br />

Lemma 9: (i) For every sufficiently sm<str<strong>on</strong>g>all</str<strong>on</strong>g> fixed η > 0 we have<br />

�<br />

|Kh|<br />

h≤H<br />

2 ≪ N 2−2η<br />

N (λ+2t)/2+100η ≤ Y.<br />

(ii) For every sufficiently sm<str<strong>on</strong>g>all</str<strong>on</strong>g> fixed η > 0 we have (11) when<br />

N max{(11λ+36t)/7−2,t}+100η ≤ Y ≤ N min{1−t,7(1+λ)/9−2t}−100η ,<br />

7<br />

(11)


provided <strong>the</strong> c<strong>on</strong>diti<strong>on</strong> (10) is satisfied.<br />

We are now in a positi<strong>on</strong> to prove<br />

Lemma 10: For every sufficiently sm<str<strong>on</strong>g>all</str<strong>on</strong>g> fixed η > 0 we have (11) when<br />

N t+100η ≤ Y,<br />

provided <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s in (9) and (10) as well as <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s<br />

are satisfied.<br />

t < (2 − λ)/5, t < (14 + 5λ)/54, t < (14 − 9λ)/32 (12)<br />

Proof: We first note that <strong>the</strong> sum �<br />

same manner as �<br />

h≤H<br />

h≤H<br />

|Kh| 2 can alternatively be estimated in <strong>the</strong><br />

|Lh| 2 was estimated in Lemma 6. Thus, we also have (11) when<br />

N t+100η ≤ Y ≤ N 2−λ−4t−100η ,<br />

provided <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s in (9) are satisfied. By this observati<strong>on</strong> and Lemma 9, for proving<br />

Lemma 10 it suffices to show that<br />

and<br />

(λ + 2t)/2 < min{1 − t, 7(1 + λ)/9 − 2t} (13)<br />

max{(11λ + 36t)/7 − 2, t} < 2 − λ − 4t. (14)<br />

Indeed, (13) follows from <strong>the</strong> first and <strong>the</strong> sec<strong>on</strong>d c<strong>on</strong>diti<strong>on</strong> in (12), and (14) follows<br />

from <strong>the</strong> first and <strong>the</strong> third c<strong>on</strong>diti<strong>on</strong> in (12). This completes <strong>the</strong> proof. ✷<br />

5 Proof of Theorem 4<br />

We suppose that 1/2 ≤ λ < 1 and<br />

We choose t in such a manner that<br />

and we take<br />

0 ≤ τ < min{(2 − λ)/6, (14 − 9λ)/32}.<br />

max{τ, 2(1 − λ)/5} < t < min{(2 − λ)/6, (14 − 9λ)/32},<br />

u := N t+100η ,<br />

v := N 2−λ−4t−100η ,<br />

z := N t+100η .<br />

8


Then <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s in (9), (10), (12) to λ and t are satisfied. Moreover, since t <<br />

(2 − λ)/6, <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s in Lemma 3 to u, v, z are satsified if η is sufficiently sm<str<strong>on</strong>g>all</str<strong>on</strong>g> and<br />

N is sufficiently large.<br />

Now, putting Lemmas 2, 3, 6 and 10 toge<strong>the</strong>r, we obtain <strong>the</strong> result of Theorem 4. ✷<br />

6 Proof of Theorem 3<br />

Fin<str<strong>on</strong>g>all</str<strong>on</strong>g>y, we derive Theorem 3 from Theorem 4 by using <strong>the</strong> following lemma in metric<br />

number <strong>the</strong>ory which is essenti<str<strong>on</strong>g>all</str<strong>on</strong>g>y due to W.M.Schmidt.<br />

Lemma 11: (Lemma 1.5 in [10]) Let X be a measure space with measure µ such<br />

that 0 < µ(X) < ∞. Let fk(x) (k = 1, 2, ...) be a sequence of n<strong>on</strong>-negative µ-measurable<br />

functi<strong>on</strong>s, and let fk, ϕk be sequences of real numbers such that<br />

Write<br />

0 ≤ fk ≤ ϕk (k = 1, 2, ...). (15)<br />

Φ(M) := �<br />

and suppose that Φ(M) → ∞ as M → ∞. Suppose that for arbitrary integers m, n<br />

(1 ≤ m < n) we have<br />

�<br />

X<br />

k≤M<br />

ϕk,<br />

�<br />

�<br />

�<br />

� �<br />

�2<br />

�<br />

�<br />

�<br />

� (fk(x) − fk) �<br />

� dµ ≤ D0 ϕk<br />

�m≤k


k<br />

(log rk) B+1 ≪ rk − rk−1 ≪<br />

rk<br />

(log rk) B+1<br />

for <str<strong>on</strong>g>all</str<strong>on</strong>g> k ≥ 1, where B is <strong>the</strong> c<strong>on</strong>stant in Theorem 3. We note that<br />

k ≪ (log rk) B+2<br />

by this c<strong>on</strong>structi<strong>on</strong>. We put X := [0, 1], define µ to be <strong>the</strong> Lebesgue measure, and<br />

define<br />

fk(θ) :=<br />

fk := r1−τ k<br />

�<br />

rk−1 < n ≤ rk,<br />

� n λ − θ � < n −τ<br />

− r1−τ k−1<br />

1 − τ<br />

,<br />

Λ(n),<br />

r<br />

ϕk := D1 ·<br />

2(1−τ)<br />

k<br />

,<br />

(log rk) 3B+6<br />

where D1 is choosen in such a manner that (15) holds for <str<strong>on</strong>g>all</str<strong>on</strong>g> positive integers k. We<br />

note that<br />

fk ≪<br />

r1−τ<br />

k<br />

(log rk) B+1<br />

by (18) and Taylor’s formula.<br />

Then <strong>the</strong> c<strong>on</strong>diti<strong>on</strong> “Φ(M) → ∞ as M → ∞” in Lemma 11 is satisfied since we<br />

have rk → ∞ as k → ∞. By Theorem 4, also <strong>the</strong> inequality (16) is satisfied for a<br />

suitable c<strong>on</strong>stant D0 (to see this, we apply <strong>the</strong> Cauchy-Schwarz inequality). Hence,<br />

<strong>the</strong> asymptotic estimate (17) indeed holds for almost <str<strong>on</strong>g>all</str<strong>on</strong>g> x ∈ [0, 1] with respect to <strong>the</strong><br />

Lebesgue measure. Moreover, by (19) we have<br />

Φ(M) ≪<br />

r2(1−τ)<br />

M<br />

(log rM) 2B+4<br />

as M → ∞. Using (17), (20) and (21), we obtain, as M → ∞,<br />

�<br />

n ≤ rM ,<br />

� n λ − θ � < n −τ<br />

Λ(n) = r1−τ M<br />

1 − τ<br />

+ O �<br />

r 1−τ<br />

M (log rM) −(B+1)�<br />

for almost <str<strong>on</strong>g>all</str<strong>on</strong>g> θ ∈ [0, 1].<br />

Fin<str<strong>on</strong>g>all</str<strong>on</strong>g>y, we pick θ ∈ [0, 1] for which (22) holds, and we seek to show that <strong>the</strong>n even<br />

(1) holds for this θ. To do so, we proceed as follows: For a given N ≥ 2 let M be <strong>the</strong><br />

positive integer for which rM−1 < N ≤ rM. Then, by c<strong>on</strong>structi<strong>on</strong> of <strong>the</strong> sequence (rk),<br />

we have<br />

rM − N ≪<br />

rM<br />

≪<br />

(log rM) B+1<br />

10<br />

N<br />

.<br />

(log N) B+1<br />

(18)<br />

(19)<br />

(20)<br />

(21)<br />

(22)


Using this inequality, (22) and Taylor’s formula, we obtain <strong>the</strong> asymptotic estimate (1).<br />

This completes <strong>the</strong> proof of Theorem 3.<br />

References<br />

[1] S. Baier, On <strong>the</strong> p λ <strong>problem</strong>, Acta Arith. 113 (2004), 77-101.<br />

[2] S. Baier, <str<strong>on</strong>g>Almost</str<strong>on</strong>g>-<str<strong>on</strong>g>all</str<strong>on</strong>g> <str<strong>on</strong>g>results</str<strong>on</strong>g> <strong>on</strong> <strong>the</strong> p λ <strong>problem</strong>, to appear in Acta Math. Hungarica.<br />

[3] S. Baier, An extensi<strong>on</strong> of <strong>the</strong> Piatetski-Shapiro prime number <strong>the</strong>orem, preprint<br />

(2004).<br />

[4] A. Balog, On <strong>the</strong> fracti<strong>on</strong>al part of p θ , Arch. Math. (Basel) 40 (1983), 434-440.<br />

[5] A. Balog, On <strong>the</strong> distributi<strong>on</strong> of p θ mod 1, Acta Math. Hung. 45 (1985), 179-199.<br />

[6] A. Balog, G. Harman, On mean values of Dirichlet polynomials, Arch. Math.<br />

(Basel) 57 (1991), 581-587.<br />

[7] S.W. Graham, G. Kolesnik, Van der Corput ′ s Method of Exp<strong>on</strong>ential Sums, Cambridge<br />

University Press, 1991.<br />

[8] G. Harman, On <strong>the</strong> distributi<strong>on</strong> of √ p modulo <strong>on</strong>e, Ma<strong>the</strong>matika 30 (1983), 104-<br />

116.<br />

[9] G. Harman, Fracti<strong>on</strong>al and integral parts of p λ , Acta Arith. 58 (1991), 141-152.<br />

[10] G. Harman, Metric Number Theory, Oxford Science Publicati<strong>on</strong>s, Clarend<strong>on</strong> Press,<br />

Oxford (1998).<br />

[11] D.R. Heath-Brown, The Pjateckii-Shapiro Prime Number Theorem, J. Number<br />

Theory 16 (1983), 242-266.<br />

[12] R.C. Vaughan, On <strong>the</strong> Estimati<strong>on</strong> of Trig<strong>on</strong>ometrical Sums over Primes, and Related<br />

Questi<strong>on</strong>s, Report No. 9, Institut Mittag-Leffler, 1977.<br />

HARISH-CHANDRA RESEARCH INSTITUTE<br />

CHHATNAG ROAD<br />

JHUSI<br />

ALLAHABAD 211 019<br />

INDIA<br />

E-MAIL: SBAIER@MRI.ERNET.IN<br />

11

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

Saved successfully!

Ooh no, something went wrong!