Covering shrinking polynomials by quasi progressions
Paul Erdős introduced the quantity \(S=T\sum_{i=1}^T |X_i|\), for a set of \(T\) arithmetic progressions, \(X_1,X_2,...,X_T\), where \(\cup_{i=1}^T X_i\) contains all squares up to \(N\). Erdős conjectured that \(S\) cannot be much smaller than \(N,\) and this conjecture was confirmed by \textit{A. Sarközy} [Ann. Univ. Sci. Budap. Rolando Eötvös, Sect. Math. 25, 267--272 (1982; Zbl 0498.10033)] showing \(S> \frac{1}{700} \frac{N}{\log^2 N}\) for sufficiently large \(N\). The paper under review shows that \(S \geq \frac{N}{4K\log N}\), where \(K\) is the maximum of the number of divisors of the differences of the arithmetic progressions. (For a one-term arithmetic progression, take the number of divisors 1.) A \textit{quasi arithmetic progression} in this paper has difference \(d\) or \(d+1\) between consecutive terms. It is shown that Erdős' conjecture for quasi arithmetic progressions is no longer true, namely allowing different \(d\)'s for different quasi arithmetic progressions, \(S<C n^{3/4}\log_2 N\) holds, with an explicit constant \(C\). Finally, the paper extends Erdős' problem from squares to values of shrinking polynomials.
This page was built for publication: Covering shrinking polynomials by quasi progressions
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7026905)
