Erdős Problem #258
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: proved (not automatically the live result).
Subject areas: irrationality.
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.
Documentary explanation. Editorial acceptance is determined by the protected V2 admissions ledger and the exact manuscript source hash. This article does not confer mathematical certification.
The question in plain language
Let \(a_1,a_2,\ldots\) be positive integers whose values eventually exceed every fixed bound; in symbols, \(a_n\to\infty\). Consider the positive series
where \(\tau(n)\) is the number of positive divisors of \(n\). The problem asks whether \(S(a)\) must always be irrational. The quantifier is important: this is one assertion about every positive-integer sequence tending to infinity, not merely about a convenient sequence such as \(a_n=n+1\).
The condition \(a_n\to\infty\) makes the series converge. After some index all \(a_n\) are at least \(2\), while \(\tau(n)\) grows much more slowly than an exponential; the finite initial segment cannot affect convergence or the rational-versus-irrational nature of the resulting value. The sequence need not be monotone: it may go up and down, provided it eventually dominates every fixed integer.
A small example
For the simple monotone choice \(a_n=n+1\), the first four terms are
This is only a beginning of an infinite sum; a decimal approximation or a finite partial sum cannot decide irrationality.
The audited RESULT/1 also records a finite check that illustrates the divisor estimate used by the reported unrestricted argument. At \(N=36\), direct factorisation verifies
For example, \(37\) is prime, so \(\tau(37)=2\); \(38=2\cdot19\), so \(\tau(38)=4\); \(40=2^3\cdot5\), so \(\tau(40)=8\); and \(60=2^2\cdot3\cdot5\), so \(\tau(60)=12\leq2^{24}\). The whole finite assertion is independently reproducible with:
from math import isqrt
def tau(n):
return sum(1 if d*d == n else 2
for d in range(1, isqrt(n) + 1) if n % d == 0)
assert all(tau(36 + k) <= 2**k for k in range(1, 31))
This check is deliberately modest. It does not establish the required assertion for infinitely many starting points \(N\), and it is not a substitute for the analytic theorem below.
What is known
There are three different layers of status.
First, the classical fixed-base case is proved. Erdős showed in 1948 that the Lambert series
is irrational for every integer \(t\) with \(|t|>1\). The positive case \(t\geq2\) is the part relevant here.
Second, Erdős and Straus proved the monotone case in Theorem 2.23 of their 1971 paper: if \(2\leq a_1\leq a_2\leq\cdots\), then \(S(a)\) is irrational. On the following pages they explicitly stated the unrestricted \(a_n\to\infty\) assertion as Conjecture 2.24. Thus the 1971 paper is a classical proof of a substantial special case, not a proof of the full quantifier in Problem #258.
Third, the unrestricted case now has an external proof report. Tao and Teräväinen’s Theorem 1.1 gives an absolute constant \(C>0\) and infinitely many positive integers \(N\) such that
where \(\Omega\) counts prime factors with multiplicity. Since \(\tau(m)\leq2^{\Omega(m)}\), these same \(N\) satisfy \(\tau(N+k)\leq2^{Ck}\). Chojecki’s April 2026 note observes that this is exactly the tail control needed: if a rational value were assumed, clearing the first \(N\) denominators would produce a positive integer smaller than \(1/2\) once the later \(a_n\) are sufficiently large. The note states the resulting theorem for every positive sequence with \(a_n\to\infty\).
Accordingly, the original unrestricted question is not described here as currently unresolved in the external literature. The reported proof is not a GCL-original result. Related Erdős–Straus questions involving \(\phi(n)\) or \(\sigma(n)\) under monotonicity remain separate conjectural questions and should not be folded into #258.
GCL's current state
The authenticated source audit is the RESULT/1 comment on MATH-PROGRAMME issue #1528. It records disposition SOURCE_LOCK_READY, compares the original positive-sequence statement with the stronger all-terms-at-least \(2\) formal encoding, and identifies Erdős (1948), Erdős–Straus (1971), Tao–Teräväinen, and Chojecki as the relevant evidence. The issue itself remains an open documentary work item; the comment is evidence, not a protected theorem disposition.
The pinned catalogue snapshot at teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89 records #258 as proved, Lean, and formalized: yes as of 21 April 2026. The live official problem page, last edited 28 May 2026, says “PROVED (LEAN)” and credits the Chojecki/Tao–Teräväinen route. Those labels report external database status; they do not certify a GCL proof.
The pinned Formal Conjectures file has theorem statements whose main bodies are by sorry. The linked Lean gist supplies a longer formal development, but declares the Tao–Teräväinen input as an axiom; the audit did not compile it. No GCL theorem, MathCert certificate, or independent replay is asserted here.
What remains
The mathematical resolution reported by the external note depends on accepting Tao–Teräväinen’s deep analytic theorem and the note’s exact tail argument. A bounded successor could compile the gist in its declared environment, inspect its axiom report, and check the index shift and finite-prefix normalization. That would audit a formal interface, not prove Tao–Teräväinen inside GCL.
The documentary work also remains distinct from mathematical status: the manuscript must pass source and rendering checks, display the precise draft-or-accepted status determined by the protected admissions ledger, rather than claim acceptance from the manuscript itself. A changed external proof, source correction, or formal replay would require a deliberate update rather than silent status inference.
Sources and claim boundaries
- Erdős (1948), On arithmetical properties of Lambert series, pp. 63–66: fixed integer base \(|t|>1\).
- Erdős–Straus (1971), Some number theoretic results, Theorem 2.23 and Conjecture 2.24, pp. 641–642: monotone theorem and historical unrestricted conjecture.
- Tao–Teräväinen, arXiv:2512.01739, version 2 dated 25 April 2026, Theorem 1.1: the consecutive-integer \(\Omega\) bound. This deep result was not reproved here.
- Chojecki, A short note on Erdős Problem #258, 14 April 2026, Lemma 1 and Theorem 2: external application of Tao–Teräväinen to the unrestricted series. No journal certification is inferred.
- RESULT/1 audit on issue #1528: source reconciliation, finite check, and formal-status caveats. It is not independent mathematical certification.
For identity retention, the audit registry records issue-body SHA-256 44840183141b361ab38b39f3bd7920e02c271dcca284316becd09fd60c353c71. The catalogue entry is pinned to Git blob 4e60263a80af6207aca494d2e1388e44fa372889 at commit 536408716724a2402bdfa7c1d764bf83d64a8c89; the formal provider is pinned to Git blob 6fa501e017e494bf3866f64c3465f2e2204224b7 at commit 85f863718beeec7b58a3a1926ee92e3472bc2020. These identities identify the inspected records; they do not turn catalogue metadata, a sorry-bearing file, or an axiom-conditional gist into a certified proof. This file itself has no editorial-admission hash or publication authority.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1528 — 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.