Status
On this page
Status
Topics
Status
On this page
Status
Topics
Let be the maximal such that there exists a choice of congruence classes for all primes such that every integer in is congruent to at least one of the .
Give good estimates for . In particular, can one prove that or even ?
Source: erdosproblems.com/687
No claim settles this problem.
Open. No source proving , or any upper bound below Iwaniec's , was found in the search whose scope the Current assessment records. The bounds in hand: (the Corollary of [Iw78], p. 226, for the longest run of consecutive integers each divisible by one of arbitrary primes, taken at ; attested in this form by the introduction of [FGKMT18] and, in the inverse form , by [Er79d] p. 79); (display (1.2) of [FGKMT18], J. Amer. Math. Soc. 2018, refereed, cited from the arXiv version), improving Rankin's ; and the site's account, since 31 August 2026, of a further improvement attributed to GPT 5.6 Pro (the name the site gives) prompted by a forum contributor, supported by a proof claim and a maintainer's exposition on the site's Problem 4 page and recorded here as the site's account, not as a refereed result. The conjectured truth is (Maier and Pomerance, as attested by [FGKMT18]); Erdős expected . This is a bounded negative finding, not a certificate of openness. Since that search, the OpenAI mathematics release of 25 September 2026 claims for Jacobsthal's function; its Theorem 1.2 states the covering form directly, for large , a yes to the first displayed question if it stands, without naming or this problem. The manuscript is unreviewed, and the Lean declaration the corpus built and audited concerns alone, so it is a pending partial claim on its claim page (OpenAI, 2026) and the standing stays open. Erdős's offer in [Er80] p. 106 is a prize "for clearing up of this problem", and the site lists a prize.