Answers are grounded in this paper and cite their sources.
Computational analysis reveals major structural invariants in three Petersen coloring counterexamples, indicating survival of related classical graph conjectures.
Jaeger's Petersen coloring conjecture (1988), which asserted that every bridgeless cubic graph admits an edge map onto the Petersen graph sending every vertex star onto a vertex star, was refuted in August 2026: Putman exhibited two nonisomorphic 112-vertex counterexamples with SAT-solver certificates, Jooken gave a human-checkable proof, and Goedgebeur, Jooken, Macajova, Mattiolo and Mazzuoccolo found a 52-vertex counterexample together with infinite families, placing the smallest counterexample between 38 and 52 vertices. The conjecture had been the strongest known sufficient condition for the Berge-Fulkerson conjecture and the 5-cycle double cover conjecture. This preprint audits, by exact proof-carrying computation, what survives on the three retrievable counterexamples. Independent encodings sharing no variable scheme with the public ones refute Petersen colorings and normal 5-edge-colorings of all three graphs with proofs checked by drat-trim; five colorable controls are accepted, and Putman's four public proofs verify under the same checker. Every one of the three graphs admits a Berge-Fulkerson cover, a Berge cover by five perfect matchings, a Fan-Raspaud triple, a 5-cycle double cover and a nowhere-zero 5-flow, all given as explicit witnesses re-verified from the graph alone; none admits a nowhere-zero 4-flow. Their perfect matching index is exactly 4, one below the Petersen graph's value. The two 112-vertex graphs have oddness 4 and resistance 3; the 52-vertex graph has oddness 2 and resistance 2, every value certified from both sides. All three admit normal and strong normal 6-edge-colorings, so their normal chromatic index is exactly 6. Defining the Petersen defect as the least number of vertices at which the star condition must fail, the 52-vertex graph has defect exactly 2: no single vertex absorbs the obstruction (52 checked refutations), while every one of its 1,326 vertex pairs does (explicit witnesses). A parity theorem explains the first half for every cubic graph: for any edge map into the Petersen graph the number of bad vertices is never exactly one, so every counterexample has defect at least 2, and the three known counterexamples attain it; moreover every one of the 6,216 vertex pairs of each 112-vertex graph is critical as well (universal 2-criticality on all three graphs, 13,758 pair witnesses). A short proof shows that no counterexample consists solely of copies of the Petersen 4-pole F from which all known counterexamples are built, and a counterexample-guided search over compositions of F with free vertices below 52 vertices is reported with its exact scope. Hypotheses were declared before each run; a refuted prediction (oddness 2 for the 112-vertex graphs) is preserved. The paper decides nothing about the general conjectures and says so. Code, declared hypotheses, artifacts, verdicts and verification instructions: https://github.com/fsantibanezleal/CAOS_RESEARCH (problems/combinatorics/petersen-coloring).
No takes yet. Share an insight, caveat, or question.
Felipe Santibañez-Leal (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: