arXiv · 2608.21648
Certified Countermodels in Profile and Incidence Fibres of Diamond-Induced Edge Partitions
Abstract
Let $G=(V,E)$ be a finite loopless directed graph. Each directed two-step diamond identifies its two pairs of opposite edges, and the connected components of the resulting auxiliary graph on $E$ define a canonical partition $\Pi_{\mathrm{opp}}(G)$. We ask whether a finite relational certificate built from the blocks of an edge partition identifies this canonical partition. We prove the functoriality and a universal coarsening property of $\Pi_{\mathrm{opp}}(G)$, count its complete fixed-profile fibre and its exact radius-one transposition neighbourhood, and reduce the parity condition in the certificate to an odd closed walk in a directed $\mathbb Z_2$-gain graph. In a fixed catalogue, four labelled target-positive rows, representing three isomorphism types, each admits a noncanonical positive partition obtained by a single cross-block transposition. In one row a stronger positive comparison partition also preserves every per-vertex, fixed-role incoming and outgoing count; it arises from a 12-edge alternating trade. The corresponding incidence fibre is a singleton in the other three rows. A deterministic 64-index family of comparison partitions yields 78 bounded positives among 256 nonidentity partitions. On the same profile fibre, one member has a complete 885-element relation semigroup and no write-preserve-use certificate, giving an exact negative. Thus the certificate has genuine but intermediate selectivity: it is neither determined by block sizes nor specific to the canonical target. The claims are finite statements in computational combinatorics and are supported by explicit, independently checkable certificates.
Explore related subjects
Keep this discovery
Ivan Khalamendyk. 2026-08-21. Certified Countermodels in Profile and Incidence Fibres of Diamond-Induced Edge Partitions. https://arxiv.org/abs/2608.21648
Cite the original work for its findings. Save a collection to share your selection of sources.