Skip to content

Erdős Problem #407

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

Erdős Problem 407 asks how many ways a positive integer can be written in the form

\[ n=2^a+3^b+2^c3^d, \]

where \(a,b,c,d\) are nonnegative integers. If \(w(n)\) is this number of representations, is there one absolute constant that bounds \(w(n)\) for every \(n\)? The question is a problem about exponential Diophantine equations: the three summands are built only from the primes \(2\) and \(3\), but their exponents vary.

There is an important convention to make explicit. The displayed equation naturally counts ordered exponent quadruples \((a,b,c,d)\). The quantitative paper most directly connected with the problem instead counts representations as distinct only when the sets of three numerical summands differ. Those are not automatically the same count, because the summands can be permuted or can coincide numerically. The original question, its later quantitative bounds, and the current formal statement must therefore be compared before their numbers are combined.

A small example

The case \(n=37\) can be checked with exact integer arithmetic. Positivity gives \(a,c\leq 5\) and \(b,d\leq 3\), so testing those finite ranges is exhaustive. The ordered solutions are

\[ \begin{aligned} &(0,2,0,3),(0,3,0,2),(1,1,5,0),(1,3,3,0),\\ &(2,0,5,0),(2,2,3,1),(2,3,1,1),(3,3,1,0),\\ &(4,1,1,2),(4,2,2,1),(5,0,2,0),(5,1,1,0). \end{aligned} \]

For example, the first three numerical checks include \(1+9+27=37\), \(1+27+9=37\), and \(2+3+32=37\); every tuple in the list gives the same total by direct evaluation of powers. Thus the ordered count is \(12\). If two representations are identified when their unordered sets of summands are equal, the same list has \(8\) distinct term sets. This is a reproducible finite check, not a proof of a global bound: it illustrates exactly why the counting convention matters.

What is known

The maintained Erdős Problems page, last edited 18 November 2025, records the question as solved and attributes the qualitative boundedness result to Evertse, Győry, Stewart, and Tijdeman. The page cites the original formulation to Erdős and Graham, Old and new problems and results in combinatorial number theory, p. 80. A direct copy of the 1988 Evertse–Győry–Stewart–Tijdeman source was not available in this bounded inquiry, so that attribution is reported through the maintained page and the later primary paper rather than presented as an independently checked quotation from the 1988 article.

The accessible primary source is Bajpai and Bennett, Effective S-unit equations beyond 3 terms: Newman's Conjecture, arXiv:2308.05162v1 (submitted 9 August 2023; PDF SHA-256 be15d92f9e0edd914b2525f4c9c888e3d83859028a03ccba3596b8feba9b91be). Their Theorem 2, p. 2, records the earlier eventual bound for the distinct-term-set count. Their Theorem 3, also p. 2, gives the effective statement

\[ \omega(N)\leq 4\text{ for }N\geq 131082, \qquad \omega(N)\leq 9\text{ for }N\geq 1, \]

along with sharper thresholds. Here \(\omega\) counts distinct tuples only after comparing the sets \(\{2^a3^b,2^c,3^d\}\) in the paper's equivalent ordering of the three summands. This establishes a strong external resolution of that normalized version, while not silently changing the definition of the ordered \(w\) in the formal file.

The paper's technical framework is an equation in S-units: a sum of a fixed number of terms whose prime factors lie in a prescribed finite set. In this application the relevant primes are \(2\) and \(3\), and the paper treats both nonvanishing and vanishing subsums. That distinction matters because identities among the three summands create repeated numerical representations without contradicting the finiteness argument for genuinely distinct term sets. The article therefore supports the boundedness and explicit thresholds it states, but it does not by itself certify that the ordered ncard in the newer Lean file has been transported through a formally checked equivalence.

GCL's current state

The source packet in issue #1677 pins the catalogue at teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89, with the 407 entry at blob 4e60263a80af6207aca494d2e1388e44fa372889. That snapshot says proved, last updated 31 August 2025, and reports the formal status as unformalized.

The live page now links a newer FormalConjectures file, whose blob is c4fb43cfd6dbe4e786e2e35d528677af0633450f. It defines \(w(n)\) by the ordered quadruple count and states boundedness with by sorry; its metadata links to an external Lean-proofs repository, but no replay or certification is supplied here. The issue's admitted GCL provider snapshot FC-GDM-85F86371 did not contain this statement, so the newer file is evidence of drift, not a replacement for the pinned source lock. GCL's existing triage class is OUTSIDE_STRICT_OPEN_TRIAGE. The V2 ledger, not this article, decides accepted coverage.

What remains

The immediate research need is a normalization audit. It should align the original ordered quadruple count, the paper's distinct-term-set count, and any intended quotient by permutations or repeated numerical summands. The \(n=37\) computation is a useful regression fixture for that audit. A second bounded task is to obtain and inspect the primary Evertse–Győry–Stewart–Tijdeman text, recording its exact theorem statement and hypotheses rather than relying on secondary attribution. Neither task should be described as a new proof of the parent problem.

Sources and claim boundaries

This article reports source evidence and a finite calculation only. It does not provide GCL mathematical certification, Lean replay, publication authority, or an independent verification of the inaccessible 1988 primary proof.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1677 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
  • Formal-statement coverage: No statement was confirmed in GCL's pinned formal-provider 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