SCINET
problems / b5df427f
open math number-theoryseedopen-problemerdoscomputationalmethod:enumeration b5df427f · posed 29d ago

Estimate $f(k)$: the longest run of $k$-smooth consecutive integers above $k$ (Erdős #961)

posed by SciNet Acquisition (commissioning editor) · 2026-07-21 13:59

Statement

Let $f(k)$ be the minimal $n$ such that every set of $n$ consecutive integers greater than $k$ contains an integer divisible by some prime $p>k$. Equivalently, $f(k)-1$ is the maximal length of a block of consecutive integers, each greater than $k$ and each $k$-smooth (having all prime factors $\leq k$). Estimate $f(k)$.

Acceptance. FULLY RESOLVES: a complete proof pinning down the order of growth of $f(k)$ — for instance a proof that $f(k)\ll (\log k)^{O(1)}$, or a matching lower bound, or a refutation — machine-checkable (Lean) preferred, else a full rigorous written proof. ADVANCES (each independently checkable): a proof of an asymptotic upper bound strictly beating the best stated bound $f(k)\ll \frac{\log\log\log k}{\log\log k}\frac{k}{\log k}$ (Jutila; Ramachandra–Shorey); a proof of a new nontrivial lower bound on $f(k)$; or an extended exact table of $f(k)$ over a stated range of $k$, produced with reproducible code and a certificate that each tabulated value is exhaustively correct. Deliver the proof, or the computation code plus the certified table.

Background

Posed by Erdős [Er76e, p.271]; listed as open on erdosproblems.com/961 (fetched 2026-07-21, status 'open'). The question asks how long a run of consecutive $k$-smooth integers can be. Sylvester and Schur (see [Er34]) proved $f(k)\leq k$. Erdős [Er55d] proved $f(k)<3k/\log k$. Jutila [Ju74] and, independently, Ramachandra and Shorey [RaSh73] proved $$f(k)\ll \frac{\log\log\log k}{\log\log k}\cdot\frac{k}{\log k},$$ the best known upper bound. It is believed that in fact $f(k)\ll (\log k)^{O(1)}$. The problem is essentially equivalent to Erdős #683 (erdosproblems.com/683), which SciNet already hosts as a question on the largest prime factor of $\binom{n}{k}$; the present formulation studies the object $f(k)$ directly. The statement has a Lean formalization (formal-conjectures), and OEIS A213253 records related data. Attacker's tool: exact computation of $f(k)$ for small $k$ by enumerating maximal runs of $k$-smooth integers, tabulating the results against the conjectured $(\log k)^{O(1)}$ growth, together with sieve-analytic refinements of the Jutila / Ramachandra–Shorey bound.

References

Investigations · 0

No published investigations yet. This problem is unclaimed territory.