Demonstrates that the covering number C(12,6,4) is 41, indicating a precise bound in combinatorial design.
A t-(v, k, λ) covering is a collection of k-subsets (blocks) of a v-set such that everyt-subset of points lies in at least λ blocks; the covering number Cλ(v, k, t) is the leastnumber of blocks in such a collection, and one writes C(v, k, t) when λ = 1. The recordedbounds for C(12, 6, 4) have been 40 ≤ C(12, 6, 4) ≤ 41. We show that no 4-(12, 6, 1)covering with 40 blocks exists, and hence that C(12, 6, 4) = 41. A counting argumentshows that in a hypothetical 40-block covering every point lies in exactly 20 blocks, thelink of every point is an optimal 3-(11, 5, 1) covering with a forced degree sequence, andthe six pairs of points of degree 10 form a perfect matching; an exhaustive case analysisover the orbits of a group of order 3840, carried out by satisfiability solving, then showsthat no optimal 3-(11, 5, 1) covering occurs as such a link. Each of the 81 formulas inthe primary proof has an unsatisfiability certificate checked by drat-trim and by theformally verified checker cake_lpr; two additional cross-encoding certificates are checkedby the same pipeline. The lower-bound argument uses no tabulated covering number: itsonly numerical input, C(10, 4, 2) ≥ 9, is itself certified. As a by-product the certificatesyield a self-contained certified proof that the optimal 3-(11, 5, 1) covering is unique up toisomorphism. Equivalently, the Turán number T (12, 8, 6) is 41; the new value propagatesto improved lower bounds for C(13, 7, 5), C(14, 8, 6), C(15, 9, 7) and C(16, 10, 8).
No takes yet. Share an insight, caveat, or question.
Charlie Krug (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: