Searcharxiv⌕ Search

arXiv subjects

Paul Seymour

Publications and source records attributed to Paul Seymour.

At least 19 recordsLinked to original sources
arXiv2026

Asymptotic structure. I. Coarse tree-width

In this paper, we develop a coarse analogue of tree-width. We prove that a graph $G$ admits a tree-decomposition in which each bag is contained in the union of a bounded number of balls of bounded radius, if and only if $G$ admits a quasi-isometry to a graph with bounded tree-width. (The ``if'' half is easy, but the ``only if'' half is challenging.) This generalizes a recent result of Berger and Seymour, concerning tree-decompositions when each bag has bounded radius. We also prove a similar result for line-width, which is an extension of path-width to infinite graphs.

math.CO↗
arXiv2026

Asymptotic structure. III. Excluding a fat tree

Robertson and Seymour proved that for every finite tree $H$, there exists $k$ such that every finite graph $G$ with no $H$ minor has path-width at most $k$; and conversely, for every integer $k$, there is a finite tree $H$ such that every finite graph $G$ with an $H$ minor has path-width more than $k$. If we (twice) replace ``path-width'' by ``line-width'', the same is true for infinite graphs $G$. We prove a ``coarse graph theory'' analogue, as follows. For every finite tree $H$ and every $c$, there exist $k,L,C$ such that every graph that does not contain $H$ as a $c$-fat minor admits an $(L,C)$-quasi-isonetry to a graph with line-width at most $k$; and conversely, for all $k,L,C$ there exist $c$ and a finite tree $H$ such that every graph that contains $H$ as a $c$-fat minor admits no $(L,C)$-quasi-isometry to a graph with line-width at most $k$.

math.CO↗
arXiv2026

Asymptotic structure. V. The coarse Menger conjecture in bounded path-width

Menger's theorem tells us that if $S,T$ are sets of vertices in a graph $G$, then (for $k\ge0$) either there are $k+1$ vertex-disjoint paths between $S$ and $T$, or there is a set of $k$ vertices separating $S$ and $T$. But what if we want the paths to be far apart, say at distance at least $c$? One might hope that we can find either $k+1$ paths pairwise far apart, or $k$ sets of bounded radius that separate $S$ and $T$, where the bound on the radius is some $\ell$ that depends only on $k,c$ (the ``coarse Menger conjecture''). The last three authors showed in an earlier paper that this is false for all $k\ge 2$ and $c\ge3$, by constructing a sequence of finite graphs giving counterexamples for larger and larger values of $\ell$ with $k=2$ and $c=3$. These counterexamples contained subdivisions of uniform binary trees with arbitrarily large depth as subgraphs, and so had unbounded path-width. Here we show that, if $H$ is a graph that can be drawn in the plane such that each region shares a vertex with the infinite region, then the coarse Menger conjecture is true for all graphs not containing $H$ as a minor. Consequently, the conjecture is true for all graphs with bounded path-width (by taking $H$ to be a sufficiently large tree), and it is true for series-parallel graphs (by taking $H=K_4$). The first is somewhat surprising, since the conjecture is false for bounded tree-width.

math.CO↗
arXiv2026

Induced subgraph density. IV. New graphs with the Erdős-Hajnal property

Erdős and Hajnal conjectured that for every graph $H$, there exists $c>0$ such that every $H$-free graph $G$ has a clique or a stable set of size at least $|G|^c$ (a graph is $H$-free if it has no induced subgraph isomorphic to $H$). Alon, Pach, and Solymosi reduced the Erdős-Hajnal conjecture to the case when $H$ is {\em prime} (that is, $H$ cannot be obtained by vertex-substitution from smaller graphs); but until now, it was not shown for any prime graph with more than five vertices. We will provide infinitely many prime graphs that satisfy the conjecture. Let $H$ be a graph with the property that for every prime induced subgraph $G'$ with $|G'|\ge 3$, $G'$ has a vertex of degree one and a vertex of degree $|G'|-2$. We will prove that every graph $H$ with this property satisfies the Erdős-Hajnal conjecture, and infinitely many graphs with this property are prime. More generally, say a graph is {\em buildable} if every prime induced subgraph with at least three vertices has a vertex of degree one. We prove that if $H_1$ and $\overline{H_2}$ are buildable, there exists $c>0$ such that every graph $G$ that is both $H_1$-free and $H_2$-free has a clique or a stable set of size at least $|G|^c$. Our proof uses a new technique of ``iterative sparsification'', where we pass to a sequence of successively more restricted induced subgraphs. This approach also extends to ordered graphs and to tournaments. For ordered graphs, we obtain a theorem which significantly extends a recent result of Pach and Tomon about excluding monotone paths; and for tournaments, we obtain infinitely many new prime tournaments that satisfy the Erdős-Hajnal conjecture (in tournament form).

math.CO↗
arXiv2026

Minors of plane digraphs

A digraph $H$ is a ``semi-strong minor'' of another, $G$, if a subdivision of $H$ can be obtained from a subdigraph of $G$ by contracting strongly-connected subdigraphs to single vertices. We will define a width measure of ``plane'' digraphs (that is, drawn in the plane) based on a kind of branch-composition, and show that for every plane digraph $H$, all plane digraphs not containing $H$ as a semi-strong minor have bounded width, while plane digraphs in general have unbounded width.

math.CO↗
arXiv2026

Induced subgraph density. VII. The five-vertex path

We prove the Erdős-Hajnal conjecture for the five-vertex path $P_5$; that is, there exists $c>0$ such that every $n$-vertex graph with no induced $P_5$ has a clique or stable set of size at least $n^c$. This completes the verification of the Erdős-Hajnal property of all five-vertex graphs. Our methods combine probabilistic and structural ideas with the iterative sparsification framework introduced in the third and fourth papers in the series.

math.CO↗
arXiv2026

When all directed cycles have the same weight

A digraph $G$ is weightable if its edges can be weighted with real numbers such that the total weight in each directed cycle equals 1. There are several equivalent conditions: that $G$ admits a 0/1-weighting with the same property, or that $G$ contains no subdivided "double-cycle" as a subdigraph, or that for every triple of vertices, all directed cycles containing all three pass through them in the same cyclic order. And there is quite a rich supply of such digraphs: for instance, any digraph drawn in the plane such that each of its directed cycles rotates clockwise around the origin is weightable (let us call such digraphs "circular"), and there are weightable planar digraphs with much more complicated structure than this. Until now the general structure of weightable digraphs was not known, and that is our objective in this paper. We will show that: - there is a construction that builds every planar weightable digraph from circular digraphs; and - there is a (different) construction that builds every weightable digraph from planar ones. We derive a poly-time algorithm to test if a digraph is weightable.

math.CO↗
arXiv2025

Asymptotic structure. II. Path-width and additive quasi-isometry

We show that if a graph $G$ admits a quasi-isometry $ϕ$ to a graph $H$ of bounded path-width, then we can assign a non-negative integer length to each edge of $H$, such that the same function $ϕ$ is a quasi-isometry to this weighted version of $H$, with error only an additive constant.

math.CO↗
arXiv2025

Line-width and path-width

For finite graphs, path-width is an interesting and useful concept, but if we extend it to infinite graphs in the most obvious way (by making the indexing path infinite), it does not work nicely. The simplest extension that works nicely is to allow the indexing set to be any totally-ordered set, and then the corresponding decomposition is called a ``line-decomposition'', and the maximum bag size needed is called ``line-width''. In particular, the indexing set need not be a well-order; but the corresponding decomposition would be easier to use if it was. We show that if a graph has line-width at most $k$, it admits a well-ordered line-decomposition with width at most $2k$, and this is best possible.

math.CO↗
arXiv2025

Induced subgraph density. VI. Bounded VC-dimension

We confirm a conjecture of Fox, Pach, and Suk, that for every $d>0$, there exists $c>0$ such that every $n$-vertex graph of VC-dimension at most $d$ has a clique or stable set of size at least $n^c$. This implies that, in the language of model theory, every graph definable in NIP structures has a clique or anti-clique of polynomial size, settling a conjecture of Chernikov, Starchenko, and Thomas. Our result also implies that every two-colourable tournament satisfies the tournament version of the Erdős-Hajnal conjecture, which completes the verification of the conjecture for six-vertex tournaments. The result extends to uniform hypergraphs of bounded VC-dimension as well. The proof method uses the ultra-strong regularity lemma for graphs of bounded VC-dimension proved by Lovász and Szegedy and the method of iterative sparsification introduced by the authors in an earlier paper.

math.CO↗
arXiv2025

Asymptotic structure. VI. Distant paths across a disc

Menger's theorem says that, for $k\ge0$, if $S, T$ are sets of vertices in a graph $G$, then either there are $k + 1$ vertex-disjoint paths between $S$ and $T$, or there is a set X of at most $k$ vertices such that every $S$-$T$ path passes through $X$. The ``coarse Menger conjecture'' proposed a generalization of Menger's theorem for paths that are far apart: for all $k, c$ there exists $\ell$, such that for every graph $G$ and subsets $S, T \subset V (G)$, either there are $k + 1$ paths between $S$ and $T$, pairwise with distance more than $c$, or there is a set $X \subset V (G)$ of at most $k$ vertices such that every $S$-$T$ path has distance at most $\ell$ from $X$. This is known to be false, but may be true if $G$ is planar. Here we show that it is true if $G$ is planar and all vertices in $S \cup T$ are on the infinite region. In this case, we also obtain a linear-time algorithm to test for the existence of $k+ 1$ paths between $S$ and $T$, pairwise with distance more than $c$.

math.CO↗
arXiv2025

Asymptotic structure. IV. A counterexample to the weak coarse Menger conjecture

Coarse graph theory concerns finding 'coarse' analogues of graph theory theorems, replacing disjointness with being far apart. One of the most interesting open questions is to find a coarse analogue of Menger's theorem, which characterizes when there are $k$ vertex-disjoint paths between two given sets $S,T$ of vertices of a graph. We showed in an earlier paper that the most natural such analogue is false, but a weaker statement remained as a popular open question. Here we show that the weaker statement is also false. More exactly, suppose that $S,T$ are sets of vertices of a graph $G$, and there do not exist $k$ paths between $S,T$, pairwise at distance at least $c$. To make an analogue of Menger's theorem, one would like to prove that there must be a small set $X\subseteq V(G)$ such that every $S-T$ path of $G$ passes close to a member of $X$: but how small and how close? In view of Menger's theorem, one would hope for $|X|<k$ and 'close' some function of $k,c$ (and indeed, this was conjectured by Georgakopoulos and Papasoglu, and independently, by Albrechtsen, Huynh, Jacobs, Knappe and Wollan); but we showed that this is false, even if $c=3$ and $k=3$. Here we upgrade the counterexample: we show that, even if $c=k=3$, no pair of constants (for 'small' and 'close') work. For all $\ell, m$, there is a graph $G$ and $S,T\subseteq V(G)$, such that there do not exist three $S-T$ paths pairwise with distance at least three, and yet there is no $X$ with $|X|\le m$ such that every $S-T$ path passes within distance at most $\ell$ of $X$.

math.CO↗
arXiv2025

Trees and near-linear stable sets

When $H$ is a forest, the Gyárfás-Sumner conjecture implies that every graph $G$ with no induced subgraph isomorphic to $H$ and with bounded clique number has a stable set of linear size. We cannot prove that, but we prove that every such graph $G$ has a stable set of size $|G|^{1-o(1)}$. If $H$ is not a forest, there need not be such a stable set. Second, we prove that when $H$ is a ``multibroom'', there {\em is} a stable set of linear size. As a consequence, we deduce that all multibrooms satisfy a ``fractional colouring'' version of the Gyárfás-Sumner conjecture. Finally, we discuss extensions of our results to the multicolour setting.

math.CO↗
arXiv2025

The vertex sets of subtrees of a tree

Let $\mathcal{F}$ be a set of subsets of a set $W$. When is there a tree $T$ with vertex set $W$ such that each member of $\mathcal{F}$ is the set of vertices of a subtree of $T$? It is necessary that $\mathcal{F}$ has the Helly property and the intersection graph of $\mathcal{F}$ is chordal. We will show that these two necessary conditions are together sufficient in the finite case, and more generally, they are sufficient if no element of $W$ belongs to infinitely many infinite sets in $\mathcal{F}$.

math.CO↗
arXiv2025

Subdivisions and near-linear stable sets

We prove that for every complete graph $K_t$, all graphs $G$ with no induced subgraph isomorphic to a subdivision of $K_t$ have a stable subset of size at least $|G|/{\rm polylog}|G|$. This is close to best possible, because for $t\ge 7$, not all such graphs $G$ have a stable set of linear size, even if $G$ is triangle-free.

math.CO↗
arXiv2025

The minimal nonplanar strong digraphs

Kuratowski's theorem says that the minimal (under subgraph containment) graphs that are not planar are the subdivisions of $K_5$ and of $K_{3,3}$. Here we study the minimal (under subdigraph containment) strongly-connected digraphs that are not planar. We also find the minimal strongly-connected non-outerplanar digraphs and the minimal strongly-connected non-series-parallel digraphs.

math.CO↗
arXiv2025

A counterexample to the coarse Menger conjecture

Menger's well-known theorem from 1927 characterizes when it is possible to find $k$ vertex-disjoint paths between two sets of vertices in a graph $G$. Recently, Georgakopoulos and Papasoglu and, independently, Albrechtsen, Huynh, Jacobs, Knappe and Wollan conjectured a coarse analogue of Menger's theorem, when the $k$ paths are required to be pairwise at some distance at least $d$. The result is known for $k\le 2$, but we will show that it is false for all $k\ge 3$, even if $G$ is constrained to have maximum degree at most three. We also give a simpler proof of the result when $k=2$.

math.CO↗