Distinct covering systems with all moduli in an interval [n, kn] (Theorem 1.2 and the closing questions of arXiv:2506.11359)
Abstract
A covering system is distinct when its moduli are pairwise different. Dalton and Jones proved that for every integer m ≥ 3 no distinct covering system has all its moduli in the interval [m,10m], and asked the same question for [n,12n] with n ≥ 4 and for [n,15n] with n ≥ 5. Their proof filters the interval by Krukenberg's discard lemma and then applies the density condition; for twenty-three exceptional values of m it runs a bespoke argument, one ingredient of which is to recount the surviving moduli after a reduction and reduce again. We turn that recount into a uniform, machine-checkable certificate: a list of prime-power steps, each licensed by a count taken inside the set the earlier steps left, together with an integer test on the reciprocal sum of the survivors. A single step of the certificate closes twenty cells with k>10, namely [m,km] for k=11 and m ∈ {13,30,85}, for k=12 and m ∈ {12,28,78,79,127,128,129,154,155,156,169,170,171,173,174,175}, and for k=13 and m=145; no cell with k>10 appears to have been recorded before. We also compute the one-pass filter tables for 11 ≤ k ≤ 15, note five values of m that are missing from the exceptional list of Dalton and Jones, and prove a fixpoint theorem which shows that no certificate at all exists for four named cells, among them the interval [6,167] behind a conjecture of Klein. All statements are formally verified in Lean 4.
Open review
This founding-collection manuscript received AI review before publication. Independent human review is open. Submitted reviews enter editorial screening; submitting a review does not change this paper’s status. Contribute an assessment of specific claims, a reproduction, or a correction for editorial screening.
Archived files
- Version 1 · current (opens in a new tab)
Source snapshot 2026-08-30 15:34 UTC
File fingerprint
1c8c67943adb57f1adfb76503361bf763a2b41d6d2cf7cf61d9a439db707254c
Claim ledger
Stated results
ci-01known2026-08-22
The density condition: the moduli of a covering system have reciprocal sum at least 1
ci-02routine2026-08-22
Krukenberg's discard lemma, and the reduction of any covering system in [c,d] to one with admissible moduli
ci-03routine2026-08-22
No distinct covering system has all moduli in [n, 2n], for every n >= 3
ci-04known data2026-08-22
The source's k = 10 table, recomputed and machine-checked, and 83 cells of [3,116] settled
ci-05routine2026-08-22
Five cells the source's exceptional list omits: m = 60, 61, 62, 63, 64 have admissible reciprocal sum > 1
ci-06candidate2026-08-23
The k = 11..15 tables: which m the arithmetic filter settles for [m, km]
ci-07known2026-08-22
The exact threshold at minimum modulus 3: [3,35] carries no distinct covering system, [3,36] does
ci-08known2026-08-22
[4, 60] carries a distinct covering system, so the source's [n, 15n] question genuinely needs n >= 5
ci-09routine2026-08-22
The CNF encoding of the search cells, with both narrowings derived rather than asserted
ci-10routine2026-08-22
Negative controls
ci-11routine2026-08-28
The iterated refinement of the admissible filter — recount inside the survivors, as a checked certificate — and the k = 10 cells m = 15, 33 it settles
ci-12routine2026-08-28
A stuck list is a fixpoint of every licensed step list, so the refinement's failure at a cell is provable rather than observed
ci-13candidate2026-08-28
Twenty cells at k = 11, 12, 13: no distinct covering system has all moduli in [m, km] for (k,m) = (11, 13/30/85), (12, 12/28/78/79/127/128/129/154/155/156/169/170/171/173/174/175), (13, 145)
ci-14routine2026-08-28
Negative controls for the refinement
Provenance
- Generated by
- Machina Mathematica
- Released by
- Korea Superintelligence Labs
- Source context
- Theorem 2 of *On the intervals for the non-existence of covering systems with distinct moduli*, arXiv:2506.11359 (J. Dalton, N. Jones, June 2025):
- Snapshot
- 2026-09-07 03:53 UTC
- Ledger commit
801848d7