Negative values of cosine sums (Q4811200)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2092434
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Negative values of cosine sums |
scientific article; zbMATH DE number 2092434 |
Statements
Negative values of cosine sums (English)
0 references
18 August 2004
0 references
0 references
0.73230827
0 references
0 references
0.72794044
0 references
0.72467667
0 references
0.7012869
0 references
0.69972026
0 references
0 references
0 references
In this paper the author studies the so-called ``Chowla's cosine problem'' [see J. Reine Angew. Math. 217, 128--132 (1965; Zbl 0127.02104)]. From the introduction: Let \(A\) be a finite set of positive integers, \(|A|=n\), and write \(f(x)= \sum_{a\in A}\cos ax\). Since \(f(0)>0\) and \(\int_0^{2\pi} f(x)\,dx=0\), we have \(\min f(x)<0\). It is a difficult question to estimate this minimum uniformly for every set of size \(n\). \textit{J. Bourgain} [Acta Arith. 45, 381--389 (1986; Zbl 0615.42001)] proved \[ \min f(x)< -c_1e^{-c_2(\log n)^{c_3}} \] with unspecified positive constants \(c_1,c_2,c_3\). In another paper [Publ. Math. Orsay 84-01, Exp. No.~2, 7 p. (1984; Zbl 0553.10029)] he showed that one can take \(c_3=1/2\) under the assumption that \(A\subset [1,n2^{\sqrt{\log n}}]\). Here the author's aim is to prove this without restriction. Theorem 1. With the above notations we have \[ \min f(x)< -c_4e^{-c_5\sqrt{\log n}} \] with a positive absolute constant \(c_4\) and \(c_5= \sqrt{(\log 2)/8}\). First he gives a reformulation with exponentials, writing \(e(x)= e^{2\pi ix}\). Theorem 2. Let \(A\) be a finite set of integers, symmetric about the origin and not containing~0. Put \(f(x)= \sum_{a\in A}e(ax)\). With \(|A|=n\) we have \[ \min f(x)< -c_4e^{-c_5\sqrt{\log n}} \] with a positive absolute constant \(c_4\) and \(c_5= \sqrt{(\log 2)/8}\). (The number \(n\) and the function \(f\) in Theorem 2 are twice the corresponding quantities of Theorem 1.) In the first stage of the proof (Section 2) he finds a regular subset of \(A\) under the assumption that \(K\) is small. In the second stage (Section 3) it is shown that the existence of a regular subset yields a bound on this minimum. The synthesis and a comparison to Bourgain's paper is given in the final section.
0 references