← All problems

Erdős's frangible-prime-gap question

Let pn<pn+1p_n<p_{n+1} be consecutive primes, let gn=pn+1−png_n=p_{n+1}-p_n, and let P−(m)P^-(m) denote the least prime divisor of an integer m>1m>1. Call the gap (pn,pn+1)(p_n,p_{n+1}) frangible if P−(m)<gnP^-(m)<g_n for every integer mm satisfying pn<m<pn+1p_n<m<p_{n+1}. If N(X)N(X) denotes the number of frangible gaps with pn+1≤Xp_{n+1}\le X, determine the frequency of such gaps; in particular, prove that N(X)=o(X)N(X)=o(X). Gafni and Tao proved the stronger unconditional bound N(X)≪X/log⁡2XN(X)\ll X/\log^2 X, and, assuming the Dickson--Hardy--Littlewood prime-tuples conjecture, obtained an asymptotic N(X)∼cX/log⁡2XN(X)\sim cX/\log^2 X for a constant cc.

References

Source: arXiv

Progress summary

Refreshed
Open · solution reported

The counting question has a claimed solution, and newer work mainly sharpens its predicted numerical constant rather than changing the conclusion.

Erdős’s problem asks how frequently consecutive-prime gaps contain no integer whose least prime factor is at least the gap length. Tao and Gafni’s work identifies this as Erdős’s problem number 682682 and gives both an unconditional density-zero result and a conditional asymptotic.

October 2026 quantitative refinement

Mossinghoff and Trudgian extend the analysis to gaps containing fewer than any fixed number rr of such integers, proving Nr(X)≪rX/log⁡2XN_r(X)\ll_r X/\log^2 X unconditionally and obtaining conditional asymptotics. For the original case, enumeration gives the rigorous bound c≥2.7413688…c\ge 2.7413688\ldots; supplementary computation suggests c≈2.7416c\approx2.7416, and the survey reaches 101710^{17}.

Current status (as of October 2026): The original counting question is claimed solved by an unconditional upper bound and a conditional asymptotic, while the sharper constant remains conditional and the preprints are unrefereed.

Sources

Solutions 0

No solutions have been posted yet.