Skip to content

Erdős Problem #1138

Editorial depth: Authored explanatory DRAFT — NOT editorially accepted.
Snapshot classification: disproved (not automatically the live result).
Subject areas: number theory, primes.

Read this status correctly

The status above is reproduced from GCL's immutable external-catalogue snapshot. It is not a new claim of current mathematical resolution, a source-semantic review, an independent proof check, or GCL certification. A published proof and a GCL replay are separate events.

UNACCEPTED DRAFT — mathematical editorial review pending

This material is a provisional, source-qualified manuscript published for line-by-line scrutiny and real HTML/LaTeX testing. It has NOT received separate GCL editorial acceptance or independent mathematical certification. Its presence on this website is not evidence of a solution, reviewed coverage, or final publication.

Review the underlying draft on GitHub.

The question in plain language

Suppose we know the largest prime gap whose preceding prime is below a large number \(x\). Is an interval of a few times that length, starting at any nearby point \(y\), guaranteed to contain approximately the expected number of primes?

Erdős Problem #1138 asks, for \(x/2<y<x\), \(C>1\) and

\[ d(x)=\max_{p_n<x}(p_{n+1}-p_n), \]

whether the uniform asymptotic

\[ \pi(y+Cd(x))-\pi(y)\sim\frac{Cd(x)}{\log y} \]

holds as \(x\to\infty\).

The maintained catalogue now labels this statement DISPROVED (LEAN), although its older explanation was not updated to reflect the new result.

A small example

For \(x=5000\), \(y=4000\) and \(C=2\), direct primality checks yield \(d(x)=34\). The actual number of primes between \(4000\) and \(4068\) is \(10\), while the proposed main term is

\[ \frac{68}{\log4000}. \]

A finite discrepancy does not disprove an asymptotic. A disproof must identify an obstruction persisting along arbitrarily large choices of \(x\), with the correct quantifiers over \(y\) and \(C\).

What is known

A 2026 disproof by Sunder, Kumrawat and Cheri, reported in the problem discussion, uses strict record prime gaps and a contradiction from two nearby constants \(1<C_1<C_2\). The source discussion reports a Lean 4 verification of the central contradiction and a separate Aristotle formalization.

The current formal-conjectures source labels the original conjecture false and points to an external proof commit. However, the local declaration ends in a proof placeholder, so its linked implementation and dependencies must be checked before GCL can claim a kernel replay.

The older catalogue explanation correctly observes that the proposed interval length is far shorter than ranges accessible using traditional prime-counting estimates. That background does not supersede the newer reported counterexample.

GCL's current state

Authenticated source return #2409 records the reported record-gap obstruction, its external formal proof pointer, and the finite prime-count calculation.

The full primary disproof and Lean dependency graph have not been independently reconstructed.

What remains

Read and independently verify the record-gap proof, precisely check its uniform asymptotic and constant quantifiers, and source-pin and kernel-replay the external Lean theorem without relying on a local placeholder.

Sources and claim boundaries

Erdős #1138; discussion of 2026 disproof; current Lean pointer; GCL audit. Externally disproved, independent source and kernel certification pending. Unaccepted draft.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #2409 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
  • Formal-statement coverage: A provider statement was included in the pinned snapshot. Presence does not mean the statement or proof has been checked.
  • External mathematical status: consult current problem page and its dated bibliography.

Back to catalogue · Compendium introduction