Erdős Problem #373
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: number theory, factorials.
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
Erdős Problem 373 asks about products of factorials. Find positive integers \(n\) and a finite nonincreasing list \(a_1,\ldots,a_k\) such that
where every \(a_i\) is at least \(2\) and \(a_1<n-1\). The question is whether only finitely many such lists exist. “Finitely many” concerns the complete data \((n,a_1,\ldots,a_k)\): it is stronger than saying that a search has found no examples up to a large bound.
The strict inequality is essential. If the largest factor were allowed to be \((n-1)!\), then any factorisation of \(n\) into factorials could give a solution after multiplying by \((n-1)!\). The problem deliberately removes this trivial source. The ordering is only a normalization, so the same factors are not counted again in different orders.
The live statement and the pinned formal statement agree on this core shape. The formal provider represents a solution by a natural number paired with a list, requires the list to be pairwise descending, and uses headI < n - 1 together with \(1<a\) for every list member. That encoding is a formal interface, not itself a proof.
A small example
The largest solution listed in the source material is an exact identity:
It satisfies \(15>14\geq5\geq2\geq2\). A second familiar identity is
Both calculations use ordinary integer multiplication, so they can be reproduced without floating-point arithmetic. For example, this exact check succeeds:
from math import factorial, prod
n, factors = 16, (14, 5, 2)
assert factorial(n) == prod(factorial(a) for a in factors)
assert n - 1 > factors[0] >= factors[1] >= factors[2] >= 2
I also ran a recursive exact-integer enumeration for \(2\leq n\leq16\), allowing nonincreasing factors from \(2\) through \(n-2\) and pruning only when the partial product exceeded \(n!\). It returned precisely
This finite check confirms the displayed examples and gives no evidence about larger \(n\). It cannot establish finiteness, because a finite search cannot rule out a later solution.
What is known
The primary source in the packet is P. Erdős, “Problems and results on number theoretic properties of consecutive integers and related questions,” Proceedings of the Fifth Manitoba Conference on Numerical Mathematics (1976), pp. 25–44. On printed p. 28, Theorem 2 says that if
then, for sufficiently large \(n\), the factorial equation has only the paper’s “trivial” solutions. Here \(P(m)\) means the largest prime factor of \(m\). Since Problem 373 excludes those trivial solutions, an eventual proof of this prime-factor inequality would imply the requested finiteness. Erdős states this as a conditional route; the paper does not prove the inequality and therefore does not prove Problem 373 unconditionally.
The maintained page reports a second conditional route involving \(P(n(n+1))/\log n\to\infty\), as well as later conditional and quantitative work. Those reports are useful leads, but they do not change the status of the exact problem without a source-level argument matching its quantifiers and hypotheses. The live entry also records conjectured small solutions, including the four identities reproduced above. A conjectured complete list is not a proof that the list is complete.
GCL's current state
The frozen catalogue source is teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89, file data/problems.yaml, Git blob 4e60263a80af6207aca494d2e1388e44fa372889. Its entry for 373 says informal_status: open, last updated 2025-08-31; formal_status is unformalized; and formalized is yes. The catalogue’s status ledger, rather than a prose article, determines accepted coverage.
The live Erdős Problems #373 page, observed in the source inquiry, is also OPEN and says that the problem cannot be resolved by finite computation. Its displayed statement has the same strict bound and ordering. Direct HTML retrieval was access-denied during this audit, so the live-page observation is recorded as a dated external status observation, not treated as a substitute for the primary paper.
The pinned formal provider is FormalConjectures/ErdosProblems/373.lean at commit 85f863718beeec7b58a3a1926ee92e3472bc2020, blob e4d534288b0e5eade9169f8b5d23fb22b1d3b286. It declares erdos_373 : S.Finite, plus conditional variants and conjectures about the maximal and two-factor solutions. Each theorem body is by sorry. This confirms statement formalization and exposes useful interfaces; it is not Lean replay or GCL mathematical certification.
What remains
A bounded successor task is to compare the printed definition of “trivial” in Erdős’s p. 28 discussion with the formal predicate S, then record a line-by-line normalization of \(n\), the ordered list, and the strict inequality. The acceptance test is a source citation for every quantifier, a successful exact-arithmetic check of the finite examples, and a formal statement whose admitted hypotheses are neither stronger nor weaker by accident. A mismatch in the empty-list convention, ordering, or meaning of “sufficiently large” falsifies the proposed normalization and stops the task. The task should stop after that comparison and a bounded finite check; it should not launch a full Lean build or promote the conditional prime-factor route to a proof.
Sources and claim boundaries
- Original assignment: MATH-PROGRAMME issue #1643. The original SOURCE RESULT/1 citation is authenticated original RESULT/1.
- Primary source: Erdős’s 1976 paper, Theorem 2 on printed p. 28; the downloaded PDF had SHA-256
2c098f2c3e5079ef09a44f693c7db43a29428bb1c8bf7df7ab57b020306a9ad0. - Frozen catalogue: the pinned
problems.yaml, commit536408716724a2402bdfa7c1d764bf83d64a8c89, blob4e60263a80af6207aca494d2e1388e44fa372889. - Formal declaration: the pinned Lean file, commit
85f863718beeec7b58a3a1926ee92e3472bc2020, blobe4d534288b0e5eade9169f8b5d23fb22b1d3b286.
The primary paper supplies a conditional implication, the finite calculation supplies only finite evidence, and the formal file supplies declarations with sorry. None is a GCL mathematical certificate, publication decision, or award claim. Acceptance and coverage remain decisions of the protected V2 ledger.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1643 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
- GCL triage: PROTECTED_READY_BACKLOG.
- 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.