Skip to content

Erdős Problem #252

Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: number theory, 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

For a positive integer \(n\), let \(\sigma_k(n)\) be the sum of the powers indexed by \(k\) of the positive divisors of \(n\). Thus \(\sigma_k(1)=1\) and \(\sigma_k(6)=1^k+2^k+3^k+6^k\). The problem asks whether the rapidly convergent real number

\[ \alpha_k=\sum_{n\geq 1}\frac{\sigma_k(n)}{n!} \]

is irrational for every integer \(k\geq 1\). The quantifier is important: this is one assertion for each positive integer \(k\), not merely a question about one fixed numerical series. The factorial makes the tail small, but it does not make irrationality automatic; a convergent series with rational terms can still have a rational sum.

The normalized source statement above follows the pinned catalogue packet and the papers. The current maintained page writes the displayed sum without an explicit lower limit. Its surrounding discussion and cited literature make the intended range \(n\geq1\) clear, but the missing bound is a source-fidelity defect worth recording rather than silently repairing.

A small example

For \(k=4\), direct divisor enumeration gives

\[ \begin{aligned} (\sigma_4(1),\ldots,\sigma_4(7)) &=(1,17,82,273,626,1394,2402),\\ \sum_{n=1}^{7}\frac{\sigma_4(n)}{n!} &=\frac{106271}{2520} \mathrel{\approx}42.171031746. \end{aligned} \]

For example, \(\sigma_4(6)=1+16+81+1296=1394\). The values and the fraction were independently reproduced with exact integer arithmetic:

from fractions import Fraction
from math import factorial

def sigma(k, n):
    return sum(d**k for d in range(1, n + 1) if n % d == 0)

values = [sigma(4, n) for n in range(1, 8)]
partial = sum((Fraction(sigma(4, n), factorial(n))
               for n in range(1, 8)), Fraction())
assert values == [1, 17, 82, 273, 626, 1394, 2402]
assert partial == Fraction(106271, 2520)

This is a finite check, not an irrationality proof. It neither controls the infinite tail nor distinguishes a rational sum from an irrational one. It is useful only as a reproducible sanity check on the normalization and the arithmetic used in the source arguments.

What is known

The classical unconditional result is now exactly the finite range \(1\leq k\leq4\).

  • For \(k=1\), Erdős and Straus's 1971 theorem on series with divisor-function numerators applies to the factorial specialization: their Theorem 2.26, p. 642, assumes a monotone integer sequence \(a_n\) with \(a_n\geq n^{1/2}\) eventually; taking \(a_n=n\) gives the denominator \(1\cdot2\cdots n=n!\). Their later 1974 paper, Theorem 1.1, p. 85, restates the earlier growth criterion in a broader rationality framework. These are borrowed published results, not a new proof here.
  • For \(k=2\), the historical record is Problem 4518 of Erdős and Kac, American Mathematical Monthly 60 (1953), p. 47, followed by Breusch's solution in volume 61 (1954), pp. 264–265. The bibliography in the pinned formal file conflates the problem and solution metadata by citing “61 (1954), Problem 4518”; the separate records are the safer source boundary.
  • For \(k=3\), Schlage-Puchta's theorem in The Irrationality of a Number Theoretical Series (2006), p. 455, proves irrationality unconditionally and also proves the implication for every positive integer \(k\) from Schinzel's Hypothesis H. Friedlander, Luca, and Stoiciu independently prove the unconditional \(k=3\) case in Theorem 1, p. 2, and their Theorem 2, p. 3, proves the analogous implication from the conjecture asserting simultaneous primality of \(k\) affine forms.
  • Pratt proves \(\alpha_4\) irrational in “The Irrationality of a Divisor Function Series of Erdős and Kac,” arXiv:2209.11124v1 (2022), Theorem 1, p. 2. Pratt explicitly says that proving the cases \(k\geq5\) appears to require new ideas.

Thus the unconditional open remainder can be stated precisely as

\[ \forall k\in\mathbb Z,\quad k\geq5\Longrightarrow\alpha_k\notin\mathbb Q. \]

The conditional statements are different claims. Schinzel's Hypothesis H implies irrationality for every positive integer \(k\) in Schlage-Puchta's theorem. The prime-tuple conjecture implies the same conclusion in Friedlander–Luca–Stoiciu. A conditional theorem is not a classical unconditional solution, and neither source is a formal proof in the provider snapshot.

GCL's current state

The authenticated original issue #1522 remains open. Its complete pinned packet identifies teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89, catalogue blob 4e60263a80af6207aca494d2e1388e44fa372889, and the protected formal-provider commit 85f863718beeec7b58a3a1926ee92e3472bc2020. The catalogue snapshot says informal status open and formal status unformalized, both dated 31 August 2025; its separate formalized: yes field dated 3 January 2026 refers to statement metadata, not a proof.

The issue's authenticated RESULT/1 return has disposition STATUS_CONFLICT. It records that the maintained problem page, edited 22 January 2026, says \(1\leq k\leq4\) is known and records both conditional implications for every positive integer \(k\). The protected catalogue's broad open label is therefore stale as a whole-problem label; it should not be read as denying the four classical cases. The safe reconciliation is “known for \(1\leq k\leq4\), open for the unconditional \(k\geq5\) remainder.”

The pinned Lean file states the main theorem, the cases \(k=0,1,2,3,4\), the \(k\geq5\) residual, and the two conditional variants, but every listed theorem body contains the placeholder sorry. Its tsum ranges over natural numbers, unlike the source series starting at \(n=1\). The exact value and harmlessness of the \(n=0\) term require an explicit check of the arithmetic-function definition. The formal prime-tuple statement also restricts offsets to natural numbers, while the paper allows integer offsets; the Schinzel encoding uses finite polynomial sets and natAbs. Equivalence of these encodings has not been certified here. External Lean links are pointers for future replay, not GCL certification.

What remains

Mathematically, the central open task is to prove or refute irrationality for every fixed \(k\geq5\), with the first unresolved case \(k=5\). The published proofs use a rationality assumption, factorial denominator clearing, and carefully selected prime or almost-prime patterns. For \(k=4\), Pratt's argument illustrates the depth: sieve estimates and exponential-sum input are combined to contradict the near-integrality forced by rationality. That method is a named published lemma chain, not an original argument supplied by this article.

For formalization, the smallest immediate obligation is semantic rather than heroic: prove that the source series and the provider's natural-number tsum denote the same real number, including the \(n=0\) summand. A second obligation is to state the prime-tuple and Schinzel hypotheses in a form demonstrably equivalent to the papers' integer-polynomial formulations. Until those checks and an actual kernel replay are complete, the provider status should remain “statement scaffold with unproved declarations.”

Sources and claim boundaries

  • Original statement and live status: Erdős Problems #252, LaTeX source, and history. The live page is current relative to this audit; the pinned catalogue remains the protected historical snapshot.
  • Erdős–Straus (1971): P. Erdős and E. G. Straus, “Some Number Theoretic Results,” Pacific Journal of Mathematics 36 (1971), 635–646, Theorem 2.26, p. 642, DOI, author-hosted scan.
  • Erdős–Straus (1974): P. Erdős and E. G. Straus, “On the Irrationality of Certain Series,” Pacific Journal of Mathematics 55 (1974), 85–92, Theorem 1.1, p. 85, DOI, scan.
  • Erdős–Kac and Breusch: Problem 4518, Monthly 60 (1953), p. 47 and Breusch's solution, Monthly 61 (1954), pp. 264–265. The catalogue/formal citation conflates these two records.
  • Schlage-Puchta (2006): J.-C. Schlage-Puchta, “The Irrationality of a Number Theoretical Series,” Ramanujan Journal 12 (2006), 455–460, Theorem, p. 455, DOI, arXiv text.
  • Friedlander–Luca–Stoiciu (2007): “On the Irrationality of a Divisor Function Series,” Integers 7 (2007), Paper A31, Theorems 1–2, pp. 2–3, PDF, EuDML record.
  • Pratt (2022): Kyle Pratt, “The Irrationality of a Divisor Function Series of Erdős and Kac,” arXiv:2209.11124v1, Theorem 1, p. 2, HTML, arXiv DOI.
  • GCL evidence: Issue #1522, its authenticated RESULT/1, the pinned catalogue entry, and the pinned formal declaration.

The article claims only source-reported classical results, the explicitly labelled conditional implications, and the independently reproduced finite calculation. It does not claim a new proof, a formal proof, a current solution for every positive integer \(k\), a GCL certificate, or editorial acceptance.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1522 — 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.

Back to catalogue · Compendium introduction