Latin trades in groups defined on planar triangulations

TL;DR

This paper investigates the algebraic and combinatorial properties of Latin trades derived from planar triangulations, establishing new results on their embeddings in abelian groups and providing algorithms for such embeddings.

Contribution

It introduces a novel algebraic framework for analyzing Latin trades on planar triangulations and proves conjectures related to their embeddings and group structures.

Findings

A group with free rank exactly two for such triangulations.

Equivalence of orders of torsion subgroups for black and white triangles.

Construction of Latin trades not embeddable in any abelian group for higher genus surfaces.

Abstract

For a finite triangulation of the plane with faces properly coloured white and black, let A be the abelian group constructed by labelling the vertices with commuting indeterminates and adding relations which say that the labels around each white triangle add to the identity. We show that A has free rank exactly two. Let A* be the torsion subgroup of A, and B* the corresponding group for the black triangles. We show that A* and B* have the same order, and conjecture that they are isomorphic. For each spherical latin trade W, we show there is a unique disjoint mate B such that (W,B) is a connected and separated bitrade. The bitrade (W,B) is associated with a two-colourable planar triangulation and we show that W can be embedded in A*, thereby proving a conjecture due to Cavenagh and Drapal. The proof involves constructing a (0,1) presentation matrix whose permanent and determinant agree up to sign. The Smith Normal Form of this matrix determines A*, so there is an efficient algorithm to construct the embedding. Contrasting with the spherical case, for each genus g>0 we construct a latin trade which is not embeddable in any group and another that is embeddable in a cyclic group. We construct a sequence of spherical latin trades which cannot be embedded in any family of abelian groups whose torsion ranks are bounded. Also, we show that any trade that can be embedded in a finitely generated abelian group can be embedded in a finite abelian group. As a corollary, no trade can be embedded in a free abelian group.