Published August 11, 2026 | Version 1.0.0

A Counterexample to Prescribed-Cycle Recovery in Barnette Graphs

Authors/Creators

Description

Bekos, Kaufmann, and Pfister asked whether every prescribed Hamiltonian cycle of a Barnette graph can be recovered by choosing the dual spanning tree in the book-embedding algorithm of Alam et al. We answer this question negatively. An explicit 16-vertex Barnette graph has a Hamiltonian cycle whose complementary perfect matching meets all three edge-color classes in the simultaneous edge/face coloring used by the algorithm. Every output of the algorithm contains every edge designated green, independently of the spanning tree. The prescribed cycle omits a green edge under every global color labeling and therefore cannot be an output.

This record contains the venue-neutral preprint and LaTeX source, a dependency-free certificate verifier, the exact graph/embedding/coloring certificate, and a pinned Lean 4 companion. The Lean companion connects facial incidence to the induced edge colors, checks the displayed Hamiltonian order and exact cycle/matching partition, and certifies connectivity after all deletions of at most two vertices.

The Lean source contains no sorry declarations or custom axioms. Finite propositions using native_decide are explicitly disclosed and accompanied by complete #print axioms audits. The topological interpretation of the oriented incidence data and the published algorithmic Property 1 remain identified external inputs.

AI disclosure: OpenAI Codex assisted with literature and novelty searches, certificate verification, code testing, proof development, and manuscript drafting and editing. Anthropic Claude was used for independent adversarial review. The author independently checked the work, accepts full responsibility, and confirms that neither system is an author.

Files

preprint.pdf

Files (123.2 kB)

Name Size Download all
md5:cd014256fdb2a0085fff05bbeaec9dcd
17.7 kB Preview Download
md5:bcb3fe8dbf03606648f02f41007290c4
720 Bytes Download
md5:445e6f110cd1239ab4c44b919193ee74
1.6 kB Preview Download
md5:bb3b2f2dc7a822a83b26469a3a2194b5
1.1 kB Preview Download
md5:a6539d1612a3682adc9be31d2408fc17
1.9 kB Preview Download
md5:928d23fc0685fd9e9342ad18d8e82c15
18.4 kB Preview Download
md5:8d51908eab84bdd22b289aca17f7667c
79.9 kB Preview Download
md5:77f1cadc26a07b5d01439d6c14fe94c0
1.2 kB Preview Download
md5:e96587e6393ef8498a122f7079f96061
653 Bytes Download

Additional details

Software

Programming language
Python , Lean