Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Statement

A sequence A={a1<a2<⋯ }A=\{a_1<a_2<\cdots\} of positive integers has Property P if ai∤aj+aka_i\nmid a_j+a_k for i<j<ki<j<k (p. 1, Erdős and Sárközy's definition). Theorem (p. 1). The set S⊂NS\subset\mathbb N of displays (1)--(2) below has Property P, and its counting function satisfies

S(x)≫xlog⁡x (log⁡log⁡x)2 (log⁡log⁡log⁡x)2.S(x)\gg\frac{\sqrt x}{\sqrt{\log x}\,(\log\log x)^2\,(\log\log\log x)^2}.

Here S(x)S(x) counts the elements of SS below xx, as the paper's A(x)=∑ai<x1A(x)=\sum_{a_i<x}1 (p. 1) does, and, by the paper's notation (Section 2, p. 2), f≫gf\gg g means that there is a constant c>0c>0 with f(x)≥c g(x)f(x)\ge c\,g(x) for all sufficiently large xx; the bound therefore holds for every large xx, not only along a sequence.

The construction (displays (1)--(2), p. 2): S=⋃i≥1SiS=\bigcup_{i\ge1}S_i with Si={n∈N:n=qi4ν2}S_i=\{n\in\mathbb N:n=q_i^4\nu^2\}, where ν\nu runs over the products of exactly ii distinct primes p≡3 (mod 4)p\equiv3\ (\mathrm{mod}\ 4) and qiq_i is the ii-th prime in the class 33 modulo 44; the factor qi4q_i^4 is an indicator of the set SiS_i that keeps the union in Property P.

Source. C. Elsholtz and S. Planitzer, On Erdős and Sárközy's sequences with Property P, Monatsh. Math. 182 (2017), no. 3, 565--575, DOI 10.1007/s00605-016-0995-9 (published online 18 October 2016; Crossref record read). The copy read for this page is arXiv:1609.07935v1 (26 September 2016, 8 pp.), whose pagination is used here; the journal text was not compared. The Theorem on p. 1 and the construction on p. 2, read in the text layer and on the page images.

Read depth. Claims checked: the definition, the Theorem, displays (1)--(2) and the statements of Lemmas 1 and 2 were read clause by clause in the text layer and on the page images. Section 3 (Property P of the union, Lemmas 1 and 2), Section 4 (products of kk distinct primes, Lemma A and Corollary 1) and Section 5 (the counting function, with the proof of the Theorem) were read for their structure and not checked step by step.

Proof pointer

Section 3: Lemma 1 (p. 2) shows that if a prime p≡3 (mod 4)p\equiv3\ (\mathrm{mod}\ 4) divides n1n_1 but not gcd⁡(n2,n3)\gcd(n_2,n_3), then n12∤n22+n32n_1^2\nmid n_2^2+n_3^2, because −1-1 is a quadratic non-residue modulo pp; Lemma 2 (p. 3) deduces that any union of the sets SiS_i has Property P, through the indicator qiq_i when the three elements do not all lie in one SiS_i and through their equal number of prime factors when they do. Section 4 bounds from below the number of squarefree integers up to xx with exactly kk prime factors, all ≡3 (mod 4)\equiv3\ (\mathrm{mod}\ 4), for kk near 12log⁡log⁡x\tfrac12\log\log x (Corollary 1, p. 3, from Lemma A). Section 5 (pp. 5--7) gives Si(x)≫x/(log⁡x (log⁡log⁡x)5/2(log⁡log⁡log⁡x)2)S_i(x)\gg\sqrt x/(\sqrt{\log x}\,(\log\log x)^{5/2}(\log\log\log x)^2) for every ii with ⌊12log⁡log⁡x⌋+2≤i≤⌊12log⁡log⁡x⌋+⌊12log⁡log⁡x⌋\lfloor\tfrac12\log\log\sqrt x\rfloor+2\le i\le\lfloor\tfrac12\log\log\sqrt x\rfloor+\lfloor\sqrt{\tfrac12\log\log\sqrt x}\rfloor, and sums these bounds (p. 7). Not reconstructed here.

Dependencies

Lemma A (p. 3), the special case of X. Meng's Lemma 9 (cited as the arXiv preprint 1607.01882) for squarefree integers with kk prime factors, all ≡3 (mod 4)\equiv3\ (\mathrm{mod}\ 4), uniformly for 2≤k≤Alog⁡log⁡x2\le k\le A\log\log x; the computed bounds 0.0482<M(3,4)<0.04830.0482<M(3,4)<0.0483 of Languasco and Zaccagnini and bounds on the Gamma function and its derivatives obtained with Mathematica (p. 4); Mertens' and Stirling's formulas; the asymptotic 2ilog⁡i2i\log i for the ii-th prime ≡3 (mod 4)\equiv3\ (\mathrm{mod}\ 4) (p. 5); and the fact that −1-1 is a quadratic non-residue modulo a prime p≡3 (mod 4)p\equiv3\ (\mathrm{mod}\ 4).

Bears on

  • Problem 12: Property P with i<j<ki<j<k is the problem's condition (no element divides the sum of two distinct larger elements), so SS is a set of the kind the problem asks about, with counting function ≫N1/2/((log⁡N)1/2(log⁡log⁡N)2(log⁡log⁡log⁡N)2)\gg N^{1/2}/((\log N)^{1/2}(\log\log N)^2(\log\log\log N)^2) for all large NN, the bound the site's commentary credits to Elsholtz and Planitzer. The bound is below N1/2N^{1/2}, so it gives neither the first question's positive liminf against N1/2N^{1/2} nor a set with at least N1−cN^{1-c} elements up to NN for every c>0c>0, and the theorem says nothing about reciprocal sums; it answers none of the problem's three questions. The larger counting functions N/(log⁡N)O(log⁡log⁡log⁡N)N/(\log N)^{O(\log\log\log N)} of the 2026 constructions are pending claims recorded on the problem page.