Linear problems in combinatorial number theory

From MaRDI portal





In many problems of combinatorial number theory one has to estimate the density of a sequence of positive integers, where the only restriction on the sequence is that some linear relations do not hold for its subsets. One such problem is the Sidon problem. A sequence \(a_i\) (finite or infinite) is called a \(B_2\)-sequence if every integer can be written in at most one way in the form \(a_i+a_j\). Let \(F_2(n)\) denote the largest number \(k\) for which there exists a \(B_2\)-sequence \(a_1<a_2<\dots <a_k\leq n\). It is known that \(F_2(n)\sim \sqrt n\). Another problem of similar nature is the following: How many numbers \(r _k(n)\) can be chosen from the first \(n\) integers not containing an arithmetic progression of \(k\) terms. It is known that \(r _k(n)>n^{1-\varepsilon/\sqrt{\log n}}\). One can ask the question for arbitrary \(n\) integers (not necessarily the first \(n\)). The results in this direction are much weaker. In the case of the Sidon problem the corresponding lower bound is only \(n^{1/3}\), for the problem of arithmetic progression, no lower bound is known. However, one can expect that the case of the first \(n\) integers is ``the worst case. This is the statement the authors have formulated as a theorem (too long to be quoted here) and proved in this paper under very general circumstances.




Cited in
(25)








This page was built for publication: Linear problems in combinatorial number theory

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4059030)