Wiki
Wiki

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

Updated

Konyagin 2004 problems set square free numbers

../


S. V. Konyagin, Problems on the set of squarefree numbers, Izv. Math. 68 (2004), no. 3, 493--520, DOI 10.1070/IM2004v068n03ABEH000486 (printed on the first page); English translation, by V. M. Millionshchikov, of the Russian original in Izv. Ross. Akad. Nauk Ser. Mat. 68 (2004), no. 3, 63--90, received 15 August 2003. The problem pages cite the Russian original as [Ko04] under the title "Problems of the set of square-free numbers".

The copy read for this card is the publisher's PDF of the English translation: twenty-eight letter-size pages, printed pp. 493--520 (physical p. nn is printed p. 492+n492+n), with a clean text layer; the statements below were read in the text layer and checked on the page images of pp. 494--495. The Russian original was not compared. Provenance: the copy came from a survey download set of September 2026; the download URL was not recorded. 301,610 bytes. The publisher's PDF prints "© 2004 RAS(DoM) and LMS" in the header of its first page (printed p. 493), every other right reserved.

Read status: claims checked for Theorems 1--4, whose statements were read clause by clause (pp. 494--495); the proofs (sections 2--7, pp. 496--519) were not read.

Contents

For a set SS of positive integers, ESN(S)\mathrm{ES}_N(S) is the maximal cardinality of A⊆{1,…,N}A\subseteq\{1,\dots,N\} with a+a′∈Sa+a'\in S for all a,a′∈Aa,a'\in A, the case a=a′a=a' included; S0S_0 is the set of squarefree numbers and ESN=ESN(S0)\mathrm{ES}_N=\mathrm{ES}_N(S_0) (p. 493). BRN(S)\mathrm{BR}_N(S) is the supremum of max⁡x∣M(x)∣/∫01∣M(x)∣ dx\max_x|M(x)|/\int_0^1|M(x)|\,dx over nonzero trigonometric polynomials MM with frequencies in S∩[1,N]S\cap[1,N], and BRN=BRN(S0)\mathrm{BR}_N=\mathrm{BR}_N(S_0) (pp. 493--494). APN\mathrm{AP}_N is the maximal length of an arithmetic progression in S0∩{1,…,N}S_0\cap\{1,\dots,N\} (p. 495). CiC_i and cic_i are effective positive constants.

  • Prior results (pp. 493--494): Erdős and Sárközy (the paper's [6], filed as erdos_1987_divisibility_properties_integers_form) proved (log⁡N)/248<ESN<3N3/4log⁡N(\log N)/248<\mathrm{ES}_N<3N^{3/4}\log N for large NN (1.1); Sárközy [15] (Acta Math. Hungar. 60 (1992), the problem page's [Sa92c]) improved the upper bound to ESN≪N3/4\mathrm{ES}_N\ll N^{3/4}; Elsholtz (oral communication) noted that the method of [6] gives ESN≫log⁡Nlog⁡log⁡N\mathrm{ES}_N\gg\log N\log\log N; Gyarmati [9] (the problem page's [Gy01]) found A,B⊆{1,…,N}A,B\subseteq\{1,\dots,N\} with ∣A∣,∣B∣≫log⁡2N|A|,|B|\gg\log^2N and all sums a+ba+b squarefree; Balog and Ruzsa [1] proved BRN≪N3/4log⁡N\mathrm{BR}_N\ll N^{3/4}\log N (1.2). Proposition 1 (p. 494): ESN(S)≤BR2N(S)\mathrm{ES}_N(S)\le\mathrm{BR}_{2N}(S). The author expects ESN=o(Nε)\mathrm{ES}_N=o(N^\varepsilon) for every ε>0\varepsilon>0, notes BRN≫N2/3\mathrm{BR}_N\gg N^{2/3} by [2], and that ESN=o(BRN)\mathrm{ES}_N=o(\mathrm{BR}_N) is unproved (p. 494).
  • Theorem 1 (p. 494; proofs in section 3, pp. 499--502): there are effective positive constants C1,C2C_1,C_2 with BRN≤N11/15exp⁡(C1log⁡N/log⁡log⁡N)\mathrm{BR}_N\le N^{11/15}\exp(C_1\log N/\sqrt{\log\log N}) (1.3) and ESN≤N11/15exp⁡(C2log⁡N/log⁡log⁡N)\mathrm{ES}_N\le N^{11/15}\exp(C_2\log N/\sqrt{\log\log N}) (1.4) for all N≥3N\ge3.
  • Theorem 2 (p. 495; proof in section 2, pp. 496--499): a large sieve inequality for square moduli: for distinct positive integers n1,…,nZ≤Nn_1,\dots,n_Z\le N and Z(q,h)=∣{j:nj≡h(modq)}∣Z(q,h)=|\{j:n_j\equiv h\pmod q\}|, ∑p≤Xp2∑h=0p2−1(Z(p2,h)−Z/p2)2≤C3NZ+X16/5Z6/5exp⁡(C3log⁡X/log⁡log⁡X)\sum_{p\le X}p^2\sum_{h=0}^{p^2-1}(Z(p^2,h)-Z/p^2)^2\le C_3NZ+X^{16/5}Z^{6/5}\exp(C_3\log X/\sqrt{\log\log X}) for all integers X≥3X\ge3. Through a result of Bombieri and Zannier [3] on elliptic curves it improves the Erdős--Sárközy inequality (1.5) for some NN, ZZ and XX (p. 494); for X≤Z1/4X\le Z^{1/4} it gives nothing beyond (1.5) (p. 495).
  • Theorem 3 (p. 495; proof in section 7, pp. 518--519, after the balanced sifting of section 6): ESN≥c1log⁡2Nlog⁡log⁡N\mathrm{ES}_N\ge c_1\log^2N\log\log N for all N≥3N\ge3 (1.6); announced in [11] (Debrecen, 2000).
  • Theorem 4 (p. 495; proof in section 5, pp. 508--510): c2log⁡2N≤APN≤C4log⁡2Nc_2\log^2N\le\mathrm{AP}_N\le C_4\log^2N for all N≥2N\ge2 (1.7); the lower bound gives ESN≥c2log⁡2N/6\mathrm{ES}_N\ge c_2\log^2N/6 by passing to the odd terms of a coprime progression and taking every other one (p. 495).
  • Section 4 (pp. 502--508) develops Brun's sieve with quadratic moduli for the proofs of Theorems 3 and 4. Not read.

Compiled scope

The introduction (pp. 493--495) was read and Theorems 1--4 are recorded as checked; the proofs were not read and nothing here is independently reviewed.

Bears on. #1109 (the problem's f(N)f(N) is the paper's ESN\mathrm{ES}_N, both including the doubles 2a2a; Theorem 1 (1.4) gives f(N)≤N11/15exp⁡(C2log⁡N/log⁡log⁡N)f(N)\le N^{11/15}\exp(C_2\log N/\sqrt{\log\log N}), improving Sárközy's N3/4N^{3/4}, Theorem 3 gives f(N)≥c1log⁡2Nlog⁡log⁡Nf(N)\ge c_1\log^2N\log\log N, improving Erdős--Sárközy's (log⁡N)/248(\log N)/248, and the expectation ESN=o(Nε)\mathrm{ES}_N=o(N^\varepsilon) on p. 494 is the problem's first question, unproved there), #1103 (for an infinite AA with A+AA+A squarefree, A∩{1,…,N}A\cap\{1,\dots,N\} is one of the sets counted by ESN\mathrm{ES}_N, so (1.4) bounds its counting function by N11/15exp⁡(C2log⁡N/log⁡log⁡N)N^{11/15}\exp(C_2\log N/\sqrt{\log\log N}), an immediate consequence noted here and not stated in the paper; otherwise the paper treats the finite problem).

No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.