arXiv subjects
Publications and source records attributed to Matthew Wales.
For a graph $H$, let $c(H)=\inf\{c\,:\,e(G)\geq c|G| \mbox{ implies } G\succ H\,\}$, where $G\succ H$ means that $H$ is a minor of $G$. We show that if $H$ has average degree $d$, then $$ c(H)\le (0.319\ldots+o_d(1))|H|\sqrt{\log d} $$ where $0.319\ldots$ is an explicitly defined constant. This bound matches a corresponding lower bound shown to hold for almost all such $H$ by Norin, Reed, Wood and the first author.
Let $c(H)$ be the smallest value for which $e(G)/|G|\geq c(H)$ implies $H$ is a minor of $G$. We show a new upper bound on $c(H)$, which improves previous bounds for graphs with a vertex partition where some pairs of parts have many more edges than others -- for instance a complete bipartite graph with a small number of edges placed inside one class. We also show a tight matching lower bound for almost all such graphs. We apply these results to show $c(K_{ft/\log t,t}) = (0.638\dotsc+o_{f}(1))t\sqrt{f}$, for $f = o(\log t) = ω(1)$.
The Hadwiger number $h(G)$ is the order of the largest complete minor in $G$. Does sufficient Hadwiger number imply a minor with additional properties? In [2], Geelen et al showed $h(G)\geq (1+o(1))ct\sqrt{\ln t}$ implies $G$ has a bipartite subgraph with Hadwiger number at least $t$, for some explicit $c\sim 1.276\dotsc$. We improve this to $h(G) \geq (1+o(1))t\sqrt{\log_2 t}$, and provide a construction showing this is tight. We also derive improved bounds for the topological minor variant of this problem.
An independent transversal of a graph $G$ with a vertex partition $\mathcal P$ is an independent set of $G$ intersecting each block of $\mathcal P$ in a single vertex. Wanless and Wood proved that if each block of $\mathcal P$ has size at least $t$ and the average degree of vertices in each block is at most $t/4$, then an independent transversal of $\mathcal P$ exists. We present a construction showing that this result is optimal: for any $\varepsilon > 0$ and sufficiently large $t$, there is a family of forests with vertex partitions whose block size is at least $t$, average degree of vertices in each block is at most $(\frac14+\varepsilon)t$, and there is no independent transversal. This unexpectedly shows that methods related to entropy compression such as the Rosenfeld-Wanless-Wood scheme or the Local Cut Lemma are tight for this problem. Further constructions are given for variants of the problem, including the hypergraph version.
Multi-interaction systems abound in nature, from colloidal suspensions to gene regulatory circuits. These systems can produce complex dynamics and graph neural networks have been proposed as a method to extract underlying interactions and predict how systems will evolve. The current training and evaluation procedures for these models through the use of synthetic multi-relational datasets however are agnostic to interaction network isomorphism classes, which produce identical dynamics up to initial conditions. We extensively analyse how isomorphism class awareness affects these models, focusing on neural relational inference (NRI) models, which are unique in explicitly inferring interactions to predict dynamics in the unsupervised setting. Specifically, we demonstrate that isomorphism leakage overestimates performance in multi-relational inference and that sampling biases present in the multi-interaction network generation process can impair generalisation. To remedy this, we propose isomorphism-aware synthetic benchmarks for model evaluation. We use these benchmarks to test generalisation abilities and demonstrate the existence of a threshold sampling frequency of isomorphism classes for successful learning. In addition, we demonstrate that isomorphism classes can be utilised through a simple prioritisation scheme to improve model performance, stability during training and reduce training time.
A Gallai coloring is an edge coloring that avoids triangles colored with three different colors. Given integers $e_1\ge e_2 \ge \dots \ge e_k$ with $\sum_{i=1}^ke_i={n \choose 2}$ for some $n$, does there exist a Gallai $k$-coloring of $K_n$ with $e_i$ edges in color $i$? In this paper, we give several sufficient conditions and one necessary condition to guarantee a positive answer to the above question. In particular, we prove the existence of a Gallai-coloring if $e_1-e_k\le 1$ and $k \le \lfloor n/2\rfloor$. We prove that for any integer $k\ge 3$ there is a (unique) integer $g(k)$ with the following property: there exists a Gallai $k$-coloring of $K_n$ with $e_i$ edges in color $i$ for every $e_1\le\dots \le e_k$ satisfying $\sum_{i=1}^ke_i={n\choose 2}$, if and only if $n\ge g(k)$. We show that $g(3)=5$, $g(4)=8$, and $2k-2\le g(k)\le 8k^2+1$ for every $k\ge 3$.