Erdős Problem #678
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: proved (not automatically the live result).
Subject areas: number theory.
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.
The question in plain language
Sometimes the least common multiple of a block of \(k\) consecutive integers exceeds the lcm of a later block containing \(k+1\) integers. Erdős Problem #678 asks whether there are infinitely many such triples of starting points and lengths. The answer is yes, established by Stijn Cambie in 2024. The length \(k\) is allowed to increase across the family of examples. It is not true that each fixed sufficiently large \(k\) has infinitely many qualifying pairs of starts.
A small example
The original 1979 referee example gives \(M(96,7)=8\,321\,670\,749\,700\) and \(M(104,8)=3\,803\,928\,503\,760\), where \(M(n,k)=\mathrm{lcm}(n+1,\ldots,n+k)\). Thus \(M(96,7)>M(104,8)\) with separated intervals because \(104\geq96+7\). Cambie's paper provides another exact example, \(M(52,7)=143\,282\,767\,320 > M(62,8)=75\,522\,807\,360\). All four numbers were checked by exact integer gcd/lcm recurrence. Two examples alone do not establish infinitude.
What is known
The directly inspected Cambie Theorem 1 is stronger than the existential conjecture: for every constant \(C\geq1\) and every sufficiently large length \(k\), one can find disjoint blocks with \(k\) and \(k+1\) terms whose lcm ratio exceeds \(C\). This supplies infinitely many triples as \(k\) varies. The distinction from a fixed \(k\) infinite family is mathematical, not editorial pedantry. For fixed \(k\), one has \(M(n,k)\leq(n+k)^k\), while the consecutive-lcm binomial estimate gives \(M(m,k+1)\geq\binom{m+k+1}{k+1}\). With \(m\geq n+k\), the second grows as \(n^{k+1}\) for fixed \(k\), so \(n\) must remain bounded if the desired reversed inequality holds. Then \(m\) must also be bounded. The fixed \(k\) infinite-pairs version is therefore false.
GCL's current state
The authenticated issue-bound audit records STATUS_CONFLICT for the formal target, not for Cambie's theorem. The GCL-pinned 678.lean asserts that for every sufficiently large fixed \(k\), the set of qualifying start-point pairs is infinite. That is not the source theorem, and the pinned proof body is sorry. External Lean work reported by the catalogue has a PNT assumption and has not been replayed by GCL. The historically ambiguous original Er79/Er92e pages were not directly checked.
What remains
Correct the formal statement through the protected upstream governance process, retaining the source statement's order of quantifiers and exact interval endpoints. Reproduce the elementary fixed \(k\) finiteness proof and check the binomial lower bound in the intended Lean library. Then audit the external Lean proof's theorem target, prime-counting assumption, dependency hashes and proof holes. A statement file or reported external proof is not in itself internal certification.
Sources and claim boundaries
Stijn Cambie, Resolution of an Erdős’ problem on least common multiples (2024), Theorem 1 directly read; maintained #678; GCL-pinned Lean #678; authenticated return. Original Erdős paper pages and the full external formal proof environment uninspected. The manuscript was published as a protected, browser-verified explanatory draft in October 2026. The external Lean work is not GCL-certified; exact-hash V2 editorial admission is a separate controlled decision.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1949 — 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.