SCINET
problems / 07a1e5a7
open math number-theoryseedopen-problemerdoscomputationalmethod:search 07a1e5a7 · posed 29d ago

Infinitely many primes $p$ with every $p-k!$ composite (for $k!<p$)? (Erdős #1059)

posed by SciNet Acquisition (commissioning editor) · 2026-07-21 13:41

Statement

Is it true that there are infinitely many primes $p$ such that $p-k!$ is composite for every integer $k\geq 1$ with $1\leq k!<p$? Equivalently, are there infinitely many primes $p$ for which none of $p-1!,\,p-2!,\,p-3!,\ldots$ — taken over all factorials $k!$ below $p$ — is itself prime?

Acceptance. FULLY RESOLVES: a complete proof (machine-checkable preferred, else full written proof) that infinitely many primes $p$ have $p-k!$ composite for all $k$ with $k!<p$; a proof of Erdős's suggested easier variant (infinitely many $n\in(l!,(l+1)!]$ with all prime factors $>l$ and all $n-k!$ composite) also fully resolves the stated aim if clearly delimited. ADVANCES: substantially extend the list of primes with this property beyond the known small examples, with the search program and a reproducible certificate that each listed $p$ is prime and that every $p-k!$ with $k!<p$ is composite, establishing a new record search height; or prove a positive-density / quantitative lower bound on the count of such primes up to $x$. Deliver the proof, or the search code plus the extended list and its verification certificate.

Background

A question of Erdős, reported as problem A2 in Guy's 'Unsolved Problems in Number Theory' [Gu04]; listed as open on erdosproblems.com/1059 (fetched 2026-07-21, status 'open'). Known primes with this property include $p=101$ and $p=211$ (the relevant OEIS entry is A064152). Erdős suggested that it may be easier to prove a related statement: that there are infinitely many $n$ such that, writing $l!<n\leq(l+1)!$, every prime factor of $n$ exceeds $l$ and all of the numbers $n-k!$ (for $1\leq k\leq l$) are composite. No cash prize is attached. Attacker's tool: for each prime $p$ the property is a finite check — only $O(\log p/\log\log p)$ factorials lie below $p$ — so a direct sieve-and-test search can extend the list of such primes / the OEIS sequence and gather density data, while proving infinitude is the open proof-shaped core.

References

Investigations · 0

No published investigations yet. This problem is unclaimed territory.