Excluding surfaces as minors in graphs
Dimitrios M. Thilikos, Sebastian Wiederrecht
TL;DR
The paper addresses the constructive GMST problem by achieving a tight bound on Euler genus while preserving the global-structure, almost-embeddable framework, and it introduces a grid-encoding approach (Dyck-walls/Dyck-grids) to unify surface embeddings with minor-universality. The core method weaves together Dyck-walls, Σ-decompositions, and a local-to-global strategy to refine the local GMST to feature a tight genus bound and a minor-universal description for graphs excluding a fixed-genus surface minor. The main contributions include a tight Euler-genus bound compatible with a constructive GMST, a local GMST version tailored to grid-like, minor-universal patterns, and a polynomial-time algorithmic implementation that computes the relevant walls, decompositions, and minor models. Altogether, this advances constructive graph minor theory by providing explicit, scalable tools for decomposing H-minor-free graphs with genus-constrained embeddings and paves the way for practical algorithms targeting surface-embedded graph classes.
Abstract
The Graph Minors Structure Theorem (GMST) of Robertson and Seymour states that for every graph $H,$ any $H$-minor-free graph $G$ has a tree-decomposition of bounded adhesion such that the torso of every bag embeds in a surface $Σ$ where $H$ does not embed after removing a small number of \textsl{apex vertices} and confining some vertices into a bounded number of \textsl{bounded depth} vortices. However, the functions involved in the original form of this statement were not explicit. In an enormous effort Kawarabayashi, Thomas, and Wollan proved a similar statement with explicit (and single-exponential in $|V(H)|$) bounds. However, their proof replaces the statement "a surface where $H$ does not embed'' with "a surface of Euler-genus in $\mathcal{O}(|H|^2)$''. In this paper we close this gap and prove that the bounds of Kawarabayashi, Thomas, and Wollan can be achieved with a tight bound on the Euler-genus. Moreover, we provide a more refined version of the GMST focussed exclusively on excluding, instead of a single graph, grid-like graphs that are minor-universal for a given set of surfaces. This allows us to give a description, in the style of Robertson and Seymour, of graphs excluding a graph of fixed Euler-genus as a minor, rather than focussing on the size of the graph.
