Introduction to Fourier analysis methods in arithmetic combinatorics
These lecture notes correspond to a session of additive combinatorics taught by Pablo Candela at the AGRA II school (August 2015, Universidad de San Antonio Abad del Cusco, Perú). They are aimed at graduate students and young researchers. The goal of these notes is to provide an introduction to Fourier-analytic methods in additive combinatorics. More precisely:\N\NIn \S 1, there is a brief introduction to Fourier analysis on finite abelian groups, including concepts such as the inversion formula, the Plancherel and Parseval identities, and the Cauchy-Schwarz inequality.\N\NIn \S 2, we are presented with a method for estimating the number of 3-term arithmetic progressions in a subset \(A\) of a finite abelian group \(G\) by analyzing the functional\N\[\N\Lambda_3(1_A,1_A,1_A) := \mathbb{E}_{x,y\in G} 1_A(x)1_A(x+y)1_A(x+2y).\N\]\NIt is shown that, using Fourier analysis, this functional \(\Lambda_3\) \emph{diagonalizes} in the sense that\N\[\N\Lambda_3(1_A,1_A,1_A) = \sum_{\chi\in \widehat{G}} \widehat{1_A}(\chi)^2 \widehat{1_A}(\chi^{-2}).\N\]\NIf we assume that \(A\) contains no (non-trivial) 3-term arithmetic progressions, it follows that the zero-mean function \(1_A - \mathbb{E}_{x\in G}1_A(x)\) must have large correlation with some non-trivial Fourier character.\N\NThis large correlation leads to a \emph{density increment argument}. This argument is explained in two settings: when \(G = \mathbb{F}_p^n\) for fixed prime \(p\) and large \(n\), and when \(G = \mathbb{Z}_N\) for large prime \(N\). In the former case, it is shown that there exists a large affine subspace \(V \leq \mathbb{F}_p^n\) such that for some \(\epsilon>0\) not depending on \(n\) we have\N\[\N\frac{|A \cap V|}{|V|} \ge (1+\varepsilon)\frac{|A|}{|G|}.\N\]\NIn the latter case, the paper introduces the concept of a Bohr set and shows that there exists a large arithmetic progression \(P \subset \mathbb{Z}_N\) such that for some \(\varepsilon'>0\) not depending on \(N\) we have\N\[\N\frac{|A \cap P|}{|P|} \ge (1+\varepsilon')\frac{|A|}{|G|}.\N\]\N\NIn \S 3, several applications of the aforementioned tools are presented. Notably, Meshulam's theorem is proved: any subset \(A \subset \mathbb{F}_p^n\) lacking 3-term arithmetic progressions must have density \(\tfrac{|A|}{3^n}=O(1/n)\). Analogously, Roth's theorem is established: any subset \(A \subset \mathbb{N}\) with no non-trivial 3-term arithmetic progressions must satisfy \N\[ \N\frac{|A \cap \{1,\ldots,N\}|}{N} = O\big(\frac{1}{\log\log N}\big). \N\] \NThe notes also discuss the best known lower and upper bounds (as of 2015) for the maximal size of subsets avoiding 3-term arithmetic progressions. Finally, a lemma of Bogolyubov is proved, which states that for any finite abelian group \(G\), if \(A \subset G\) has large density, then the set\N\[\N2A - 2A := \{a + b - c - d : a, b, c, d \in A\}\N\]\Ncontains a large Bohr set.\N\NFor the entire collection see [Zbl 1553.11001].
This page was built for publication: Introduction to Fourier analysis methods in arithmetic combinatorics
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6968905)
