Erdős Problem #310
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: proved (not automatically the live result).
Subject areas: number theory, unit fractions.
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 #310 asks whether a dense set of small positive integers must contain a small-denominator rational made from distinct unit fractions. A unit fraction is a reciprocal such as \(1/6\). The source question starts with \(\alpha>0\), an integer \(N\geq1\), and a set \(A\subseteq\{1,\ldots,N\}\) satisfying \(|A|\geq\alpha N\). It asks for a subset \(S\subseteq A\) and positive integers \(a,b\) such that
with \(b=O_\alpha(1)\). The last notation means that, after fixing the density \(\alpha\), one bound for \(b\) works independently of \(N\) and of the particular set \(A\). The positivity condition matters: the empty subset would otherwise give the uninformative sum \(0\).
The live Erdős Problems entry for #310 gives this formulation and identifies the historical reference as Erdős and Graham, Old and new problems and results in combinatorial number theory (1980). It also records the later quantitative result described below. The original GCL issue #1580 binds this source inquiry to assignment ERDOS-CATALOGUE-0310-S01.
A small example
Take \(N=6\) and \(A=\{2,3,6\}\). Then \(|A|=3=(1/2)N\), so this is a concrete set of density \(\alpha=1/2\). Choose \(S=A\). Exact integer arithmetic gives
The calculation was checked with Python's fractions.Fraction, which returned sum = 1 and verified = True; no floating-point approximation is involved. It illustrates the target shape and shows that the denominator can be \(1\). It does not establish the theorem: one finite example says nothing about every dense \(A\), about sufficiently large \(N\), or about a bound uniform over all such sets.
What is known
Thomas F. Bloom's paper, On a density conjecture about unit fractions, states in Theorem 2 of the introduction that every set of positive upper density in the positive integers contains a finite subset whose reciprocal sum is exactly \(1\). Its more quantitative Theorem 3 gives a finite-interval criterion in terms of the reciprocal mass \(\sum_{n\in A}1/n\). Liu and Sawhney explain that Bloom's result already implies the original Problem #310 conclusion with \(b=O_\alpha(1)\).
The sharper source is Yang P. Liu and Mehtaab Sawhney, On further questions regarding unit fractions, arXiv version 1 dated 2024-04-10. Proposition 1.4 in the introduction states that, for every \(\varepsilon>0\) and sufficiently large \(N\) depending on \(\varepsilon\), if
then there are integers \(1\leq s\leq t\leq\exp(C/\alpha)\) and a subset \(B\subseteq A\) with \(\sum_{n\in B}1/n=s/t\). The paper explicitly says that applying Bloom's Proposition 1 with standard estimates gives the original \(t=O_\alpha(1)\) conclusion, while Proposition 1.4 records the stronger quantitative range and the exponential denominator bound. It also explains why that exponential dependence is essentially sharp. These are external published or posted mathematical results; this article does not independently reprove them.
GCL's current state
The frozen catalogue bytes are teorth/erdosproblems data/problems.yaml at commit 536408716724a2402bdfa7c1d764bf83d64a8c89; the relevant Git blob is 4e60263a80af6207aca494d2e1388e44fa372889, and the downloaded file hash is be65b0db56fe13becec3178f2937685ee7e6a2f5df9325712474722c279c4182. Its entry records informal status proved from 2025-08-31, reported formal status Lean, and later generated fields marking the status proved (Lean) and the statement formalized. Those fields are catalogue evidence, not GCL mathematical certification.
The exact upstream FormalConjectures file is present at commit 9c333fac16ecba3bfe65a7495a11849b3ba55747, file blob 519642a3fd73c940cb7a750859127ddda64a66c7, with source hash bcc1604be816226a78367152aa5c93fb76327a2dbde5eb6956f4d0d6e32180b3. It formalizes both the bounded-denominator statement and the Liu–Sawhney variant, but both theorem declarations end with by sorry. The file links to plby/lean-proofs commit 8822f7ddef30fadbd92e1c6ab4ed897af356af5e, whose referenced source was read but not replayed here. Thus the formal source provides a useful interface and provenance trail, not a result certified by this audit.
What remains
The external mathematical status is resolved by Bloom and Liu–Sawhney, but a careful GCL record still needs to preserve the exact semantic chain. In particular, the original statement uses a constant depending only on fixed \(\alpha\), while Proposition 1.4 adds an admissible range for \(\alpha\) that depends on \(N\), together with the explicit bound \(t\leq\exp(C/\alpha)\). A formalization review should check positive numerators, nonempty subsets, natural-versus-real coercions, the interval convention \([1,N]\), and the meaning of “sufficiently large.”
A bounded successor could compare the FormalConjectures statement line by line with Proposition 1.4 and Bloom's Proposition 1, then record whether the linked plby theorem is definitionally or only mathematically equivalent. Acceptance would require a source-level correspondence table and exact source identities; falsification would be a quantifier, domain, or denominator mismatch. A stopping rule is reached when that comparison is complete or when a missing primary passage prevents it. No Lean build or mathematical certification is supplied by this documentary article.
Sources and claim boundaries
- Original GCL issue #1580, including the pinned source packet. The original SOURCE RESULT/1 citation is authenticated original RESULT/1; no future comment link is fabricated here.
- Live Erdős Problems #310, a maintained summary and bibliography, not the sole authority for the proof.
- Erdős and Graham (1980), cited as
ErGr80, the historical primary source named by the live entry. - Bloom, arXiv:2112.03726, PDF hash
16130b55cf4f1e33d6958954f08f334becf24c70886560d76665d33f78dccd09; Theorem 2 and Theorem 3 are the relevant primary statements. - Liu and Sawhney, arXiv:2404.07113, version 1, PDF hash
29f6ea8f73bb17a9084bec115af4b3d8015d9b6d5e3c803a3caa8f79a287d91b; Proposition 1.4 is the primary quantitative statement. - FormalConjectures #310, exact commit and blob recorded above; its
sorrybodies are an explicit claim boundary.
The catalogue label, live-page summary, formal statement, and external papers are different evidence classes. V2 ledger state alone decides accepted compendium coverage. This article reports the source inquiry and its limits; it does not certify the mathematics, certify a Lean replay, promote publication, or replace the authenticated original RESULT/1.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1580 — 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.