Erdős Problem #7
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: verifiable (not automatically the live result).
Subject areas: number theory, covering systems.
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
A covering system is a finite list of congruences
such that every integer \(n\) satisfies at least one congruence. The modulus \(m_i\) is the step between integers in the corresponding progression. In a distinct covering system the moduli \(m_i\) are pairwise different.
Erdős Problem #7, associated with Erdős and Selfridge, asks:
Can all the moduli in a distinct covering system be odd integers greater than \(1\)?
Precisely, does there exist a positive integer \(k\), pairwise distinct odd integers \(m_1,\ldots,m_k\geq3\), and residues \(a_i\) such that
This existence question remains open in the checked Erdős Problems entry and the available 2026 literature. It does not require the moduli to be square-free. That extra restriction defines an important stronger problem for which nonexistence has been proved.
The catalogue marks this question VERIFIABLE, meaning an affirmative answer could be witnessed by a finite, exactly checkable list. It does not mean that a witness exists, that exhaustive searches settle the negative answer, or that any logical independence/undecidability theorem has been established.
A small example
Consider the three distinct odd moduli \(3,5,7\), with residue classes
These do not cover all integers: for instance \(n=4\) belongs to none of them. In fact, the entire configuration repeats every
An independent integer enumeration of \(n=0,1,\ldots,104\) gives 57 covered and 48 uncovered residues.
There is also a short proof of the exact number 48, without relying on enumeration. Because \(3,5,7\) are pairwise coprime, the Chinese remainder theorem identifies a residue modulo \(105\) with one choice of residue modulo each of \(3,5,7\). To be uncovered, we must avoid one prescribed residue for each modulus. Hence there are exactly
uncovered classes, whatever three residues are chosen.
This does not exclude all possible odd covering systems: distinct odd moduli need not be pairwise coprime.
For comparison, an actual covering system with distinct but not all odd moduli is
Checking the 12 residues modulo \(12\) finds zero uncovered: the class modulo \(2\) covers all even residues; the remaining odd residues \(1,3,5,7,9,11\) are covered respectively by moduli \(4,3,4,12,3,6\). This positive control illustrates both what a finite certificate looks like and why the odd-moduli qualification cannot be dropped.
More generally, for any proposed finite list let \(L=\mathrm{lcm}(m_1,\ldots,m_k)\). Since coverage is periodic with period \(L\), the list covers every integer if and only if it covers all \(L\) residue classes \(0,\ldots,L-1\). A candidate yes certificate is therefore decidable by finite arithmetic. There is no known global bound on \(k\) or \(L\) making failure to find a certificate a proof of no.
What is known
- Necessary divisibility, not existence. Hough and Nielsen, Covering systems with restricted divisibility, Duke Mathematical Journal 168 (2019), 3261–3295, prove that every distinct covering system has at least one modulus divisible by \(2\) or \(3\). Therefore a hypothetical all-odd system must include a modulus divisible by \(3\). It remains hypothetical.
- The square-free subclass is impossible. Balister, Bollobás, Morris, Sahasrabudhe and Tiba, The Erdős–Selfridge problem with square-free moduli, Algebra & Number Theory 15 (2021), 609–626, rule out a covering whose moduli are simultaneously odd, square-free and distinct. The proof does not eliminate odd moduli containing a repeated prime factor.
- Further restrictions on a hypothetical solution. Balister and collaborators, On the Erdős covering problem: the density of the uncovered set, Inventiones Mathematicae 228 (2022), 377–414, provide additional covering-system estimates. Theorem 1.4 (page 4) of the authors' accessible preprint states that for a distinct covering system with \(L=\mathrm{lcm}(m_1,\ldots,m_k)\), one has \(2\mid L\), \(9\mid L\), or \(15\mid L\). For an all-odd system this reduces to the necessary condition \(9\mid L\) or \(15\mid L\). Importantly, \(15\mid L\) does not imply that any single modulus is divisible by 15: the paper explicitly warns against that inference. This establishes a constraint, not the existence of such a covering.
- Finite formalized exclusions. Mian and Siddique's July 2026 preprint, Kernel-Checked Exclusions for the Erdős–Selfridge Odd Covering Problem, reports a Lean-checked exclusion of candidates whose least common multiple is at most \(10{,}000\). This is the authors' reported formalization, not a kernel replay independently performed here and not a resolution of the unbounded existence question.
- Changing the hypotheses changes the question. Bispels, Cohen, Harrington, Lowrance, Pontes, Schaumann and Wong, A further investigation on covering systems with odd moduli, Discrete Mathematics 349 (2026), article 115013, study a variant permitting one odd modulus to occur more than once. That is not the original distinct-moduli question.
These statements have different logical strengths. A theorem about every covering system imposes a condition; a theorem about all square-free systems eliminates only a subclass; and a finite-period exclusion rules out only bounded periods. None of these steps by itself answers the unrestricted odd-covering question.
GCL's current state
The GCL source-audit return for #7 pinned the problem statement to the Formal Conjectures provider at commit 85f863718beeec7b58a3a1926ee92e3472bc2020 and examined the meaning of the strict covering-system definition indexed over \(\mathbb Z\). In particular, finite indexing, distinct moduli, oddness, and exclusion of trivial modulus \(1\) are important to matching the mathematics. The theorem file contains an open conjectural formulation, not a certified affirmative answer.
The source audit also established the elementary exact period-checking criterion, the Chinese-remainder obstruction for pairwise-coprime moduli, and concrete finite fixtures such as those above. The present article independently rechecked both small-example counts using integer residue enumeration. These are elementary local facts, not a proof for arbitrary odd moduli.
The recorded source return has disposition BOUNDED_TASK_DESIGNED. It did not include an independently replayed Lean proof of Erdős #7 or a MathCert-certified resolution. The GCL explanatory article is documentary work, not mathematical certification. Manuscript PR #2577 merged, and the genuine authored EC-ED02 RESULT/1 was projected into operational RETURNED by protected run #38139713290, recorded in the parent receipt. Neither merge nor RETURNED state is editorial acceptance. A deployed-site workflow succeeded, but page-specific rendered-mathematics readback and separate editorial acceptance were not established by that receipt.
What remains
The central yes/no question is still:
- Yes route: exhibit finitely many pairwise distinct odd moduli \(>1\) and corresponding residues, then verify all residues modulo their least common multiple.
- No route: prove that every such finite candidate leaves some integer uncovered, without assuming an unproved bound on the number or size of the moduli.
The square-free result shows that any hypothetical example must exploit nonsquare-free structure. Restrictions such as divisibility by \(3\), the Theorem 1.4 necessary condition that \(9\) or \(15\) divide the least common multiple, and finite formalized exclusions narrow possibilities but do not remove the unbounded search space.
For GCL specifically, the previously outstanding 2022 theorem location has now been independently checked against the author's preprint. Remaining checks include reconciliation of other cited primary-source interfaces as appropriate, the non-authoring editorial acceptance of the exact amended manuscript, and direct readback of GitHub, MkDocs and deployed HTML mathematical rendering. Protected manuscript merge and operational RETURNED transport are recorded, but do not discharge those checks.
Sources and claim boundaries
- T. F. Bloom, Erdős Problem #7, with revision history: catalogue statement, VERIFIABLE classification and open-status note (consulted 11 October 2026). The frozen issue metadata and external prize labels have differed; no prize entitlement is asserted here.
- GCL issue #1276, source-audit RESULT/1, 10 October 2026: exact provider snapshot, logic of verification, source-to-claim ledger and finite arithmetic fixtures. A returned source audit is not mathematical certification.
- R. D. Hough and P. P. Nielsen, Covering systems with restricted divisibility, Duke Math. J. 168 (2019), 3261–3295: the necessary divisibility by \(2\) or \(3\). The deep published proof was not independently replayed here.
- P. Balister, B. Bollobás, R. Morris, J. Sahasrabudhe and M. Tiba, The Erdős–Selfridge problem with square-free moduli, Algebra & Number Theory 15 (2021), 609–626: exclusion of the odd square-free subclass.
- P. Balister et al., On the Erdős covering problem: the density of the uncovered set, Invent. Math. 228 (2022), 377–414: Theorem 1.4, page 4, author's accessible preprint arXiv:1811.03547, gives \(2\mid L\) or \(9\mid L\) or \(15\mid L\). The exact theorem statement has been read at its source; its deep proof has not been replayed.
- I. Mian and S. Siddique, arXiv:2607.25628, 28 July 2026: preprint report of kernel-checked exclusions for least common multiple \(\leq 10{,}000\); not an independent GCL proof or a solution.
- C. Bispels et al., A further investigation on covering systems with odd moduli, Discrete Math. 349 (2026), article 115013: explicitly repeated-modulus variant, not the unrestricted distinct-modulus problem.
Scope of this document: mathematical exposition and finite reproducible examples only. No new solution, formal proof acceptance, independent reviewer signoff, prize claim, or publishing authority is asserted.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1276 — 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.