Wiki
Wiki

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

Updated


Claim. There is a comeager set G⊂(2,∞)G\subset(2,\infty) such that for every x∈Gx\in G the set Sx={⌊xn⌋:n≥1}S_x=\{\lfloor x^n\rfloor : n\ge 1\} is Sidon and meets every infinite arithmetic progression; the complements of these uncountably many sets contain no infinite arithmetic progression, so the answer is no.

For x≥2x\ge 2 the terms of SxS_x at least double, so SxS_x is Sidon. For d≥1d\ge 1, 0≤r<d0\le r<d and N≥1N\ge 1, let Ud,r,NU_{d,r,N} be the set of xx with m<xn<m+1m<x^n<m+1 for some n≥Nn\ge N and some m≡r(modd)m\equiv r\pmod d, so that ⌊xn⌋=m\lfloor x^n\rfloor=m; it is open, as a union of the intervals (m1/n,(m+1)1/n)(m^{1/n},(m+1)^{1/n}), and dense, since bn−anb^n-a^n grows without bound for fixed b>ab>a. The Baire category theorem makes the intersection of the Ud,r,NU_{d,r,N} over all dd, rr and NN comeager, and for xx in it SxS_x meets every residue class modulo every dd infinitely often, hence every infinite progression. The comment states the argument on (1,∞)(1,\infty) with SxS_x Sidon for x≥2x\ge 2; the library's construction page writes the density, tail and distinctness steps out on (2,∞)(2,\infty).

Postings. Sayan Dutta's comment of 2025-09-02 in the site's discussion thread; the site's page records Dutta's observation that these sets give examples for uncountably many xx and thanks Dutta. The page's record of the remark appears in a web archive capture of 2026-01-28 and not in one of 2025-12-06.

Acceptance. Reviewed: the site's curator, Thomas Bloom, records the construction on the problem page as a further example and credits it to Dutta, a documented acceptance outside this project and independent of the claimant. No refereed publication exists, and no formalization of this argument is posted.

Depends on. No wiki page; the claim rests on the argument stated above.