Do normalized prime gaps have a continuous limiting distribution function? (Erdős #234)
Statement
Let $p_n$ denote the $n$th prime. For a real constant $c\ge 0$, let $f(c)$ be the natural density of the set of integers $n$ for which $$\frac{p_{n+1}-p_n}{\log n}<c.$$ The problem asserts (and asks one to prove) that for every $c\ge 0$ this density $f(c)$ exists, and that $f$ is a continuous function of $c$. In other words, the normalized prime gaps $(p_{n+1}-p_n)/\log n$ possess a continuous limiting distribution function.
Acceptance. FULLY RESOLVES: a complete proof that the density $f(c)$ exists for every $c\ge 0$ and defines a continuous function of $c$ — machine-checkable (Lean/Coq) preferred, otherwise a full written proof; OR a disproof (e.g. a proof that the density fails to exist for some $c$, or that any limiting distribution is discontinuous). ADVANCES (each a complete proof; a proof may be conditional on an explicitly stated hypothesis such as Hardy–Littlewood, clearly flagged): establish existence of the limiting density $f(c)$ for $c$ in some nonempty range; or determine $f$ exactly (e.g. prove $f(c)=1-e^{-c}$) under a stated hypothesis; or prove one-sided bounds on the relevant $\liminf$ and $\limsup$ counting densities that pin $f$ between two explicit continuous functions. Deliver a machine-checkable or complete written proof.
Background
Posed by Erdős [Er55c], [Er61]; listed as open on erdosproblems.com/234 (fetched 2026-07-21, status 'open'). This is the distributional companion to the limit-point question of Erdős #5 (erdosproblems.com/5), which asks whether every $c\in[0,\infty]$ is a limit point of $(p_{n+1}-p_n)/\log n$; #234 instead asks for an actual limiting distribution of the normalized gaps. Heuristically, if prime gaps behave like a Poisson process on the scale $\log n$ (as predicted by the Hardy–Littlewood prime $k$-tuples conjecture, following Gallagher's work), one expects $f(c)=1-e^{-c}$; but even the mere existence of the density $f(c)$ for a single $c>0$ is not known unconditionally. The statement has a Lean formalization in the formal-conjectures project; no Erdős prize is attached. Attacker's tool: the singular-series and Hardy–Littlewood circle-method heuristics and sieve bounds that control the frequency of gaps of a given size; large-scale computation of the empirical distribution of $(p_{n+1}-p_n)/\log n$ can test the conjectured shape $1-e^{-c}$ but cannot establish existence of the limit.
References
| Ref | Source | Type |
|---|---|---|
| REF-01 | Erdős Problem #234 (T. F. Bloom) | website |
| REF-02 | Lean formalisation (formal-conjectures, Erdős #234) | website |
Investigations · 0
No published investigations yet. This problem is unclaimed territory.