Counter Example to a Strong Matroid Minor Conjecture

Shrawan Kumar · 2024

Read Counter Example to a Strong Matroid Minor Conjecture by Shrawan Kumar (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-04288.txt.

Introduction

Shrawan Kumar disproves a strong form of the Matroid Minor Conjecture in a short note from 2024. The claim, due to Draisma, concerned ideals in the ring of the infinite Grassmannian that stay stable under all finite swaps of axes. Kumar builds an explicit chain of such ideals, each strictly larger than the last, with no end. That chain breaks the finiteness trait the strong claim would have required. The weaker claims about the same space, he notes, remain open.

Catalog record

CREATORShrawan Kumar
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-04288

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Counter Example to a Strong Matroid Minor Conjecture Shrawan Kumar Abstract Abstract: The main result of this note asserts that a strong form of the Matroid Minor Conjecture due to J. Draisma is not true, i.e., there exist properly ascending chains of S_{\infty} -stable ideals in the affine coordinate ring of the affine infinite Grassmannian , where S_{\infty} is the infinite symmetric group. In fact, we explicitly construct such an ascending chain. His conjectures on topological noetherian property for the affine infinite Grassmannian remain open though. 1. Introduction Let k be any (including finite) field. For positive integers n,p , let V_{n,p} be the vector space over k with basis \{x_{-n},x_{-n+1},\dots,x_{-1},x_{1},x_{2},\dots,x_{p}\}. Let \{x^{*}_{-n},x^{*}_{-n+1},\dots,x^{*}_{-1},x^{*}_{1},x^{*}_{2},\dots,x^{*}_{p}\} be the dual basis of the dual vector space V_{n,p}^{*} . Let \Gr(p,V^{*}_{n,p}) be the Grassmannian of p -planes in V^{*}_{n,p} . Then, we have the Plücker embedding: \iota:\Gr(p,V^{*}_{n,p})\hookrightarrow\mathbb{P}\left(\wedge^{p}(V^{*}_{n,p})\right),\,\,A\mapsto\wedge^{p}(A),\,\,\text{for $A\in\Gr(p,V^{*}_{n,p})$}. Let \widetilde{\Gr}(p,V^{*}_{n,p}) be the corresponding affine cone (under the above embedding). Thus, we get an embedding \widetilde{\Gr}(p,V^{*}_{n,p})\hookrightarrow\wedge^{p}(V^{*}_{n,p}). This makes \widetilde{\Gr}(p,V^{*}_{n,p}) a closed irreducible (affine) subvariety of \wedge^{p}(V^{*}_{n,p}) . Define the surjective maps \xi=\xi^{n+1,p}:\wedge^{p}(V^{*}_{n+1,p})\twoheadrightarrow\wedge^{p}(V^{*}_{n,p}) induced from the standard inclusion of the bases and \theta=\theta^{n,p+1}:\wedge^{p+1}(V^{*}_{n,p+1})\twoheadrightarrow\wedge^{p}(V^{*}_{n,p}),\,\,\,\omega\mapsto i_{x_{p+1}}\omega, where i is the interior multiplication. Now, the maps \theta^{n+1,p+1} and \xi^{n+1,p+1} induce the restriction maps which are surjective: \widetilde{\Gr}(p,V^{*}_{n+1,p})\xleftarrow{\widetilde{\theta}}\widetilde{\Gr}(p+1,V^{*}_{n+1,p+1})\xrightarrow{\widetilde{\xi}}\widetilde{\Gr}(p+1,V^{*}_{n,p+1}). Define a partial order \leq on \mathbb{Z}_{\geq 1}\times\mathbb{Z}_{\geq 1} by (n,p)\leq(m,q) if n\leq m and p\leq q . The above maps \widetilde{\theta} and \widetilde{\xi} give rise to a surjective map between the affine varieties (see \lx@sectionsign 2 for more details): \widetilde{\Gr}(q,V^{*}_{m,q})\to\widetilde{\Gr}(p,V^{*}_{n,p})\,\,\,\text{for $(n,p)\leq(m,q)$}. Define the affine schemes: \widetilde{\Gr}(\infty/2,V^{*}_{\infty}):=\varprojlim_{(n,p)}\,\widetilde{\Gr}(p,V^{*}_{n,p}),\,\,\text{and}\,\,\wedge^{\infty/2}(V^{*}_{\infty}):=\varprojlim_{(n,p)}\,\wedge^{p}(V^{*}_{n,p}). Define the group \GL(\infty):=\varinjlim_{(n,p)}\,\GL(V_{n,p}),\,\,\,\text{ and its subgroup}\,\,\N(\infty):=\varinjlim_{(n,p)}\,\N(V_{n,p}), where \N(V_{n,p}) denotes the normalizer of the standard maximal torus in \SL(V_{n,p}) . The standard action of \GL(V_{n,p}) on \wedge^{p}(V^{*}_{n,p}) gives rise to an action of \GL(\infty) on \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) (cf. \lx@sectionsign 2 for more details). Define the infinite symmetric group S_{\infty}:=\varinjlim_{n}\,S_{n},\,\,\,\text{where $S_{n}$ is the symmetric group on the symbols $\{-n,-(n-1),\dots,-1,1,2,\dots,n\}$}. Then, S_{\infty} canonically embeds in \GL(\infty) via the permutation matrices. Jan Draisma kindly told us his following conjecture. Conjecture 1.1 . Let k be any field. The affine infinite Grassmannian \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) is topologically S_{\infty} -noetherian, i.e., every descending chain of S_{\infty} -stable Zariski-closed subsets of \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) stabilizes. A slightly weaker form of the conjecture states that \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) is topologically \N(\infty) -noetherian. A stronger form of the conjecture states that any ascending chain of S_{\infty} -stable ideals in the affine coordinate ring of \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) stabilizes. The main result of this note is the following (cf. Theorem 3.3 ). Theorem 1.2 . The strong form of the above conjecture is false, i.e., there exists an ascending chain of S_{\infty} -stable ideals in the affine coordinate ring of \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) which does not stabilize. In fact, we give such an example explicitly. Before we can explain the significance of Draisma’s Conjecture 2.6 to some important results in Graph Theory, we need to briefly explain some of the very significant results in Matroid Minors Theory . We begin by recalling the following conjecture due to Rota [ Ro ] . Conjecture 1.3 . For each finite field k , there are, up to isomorphism, only finitely many excluded minors for the class of F -representable matroids. Rota’s conjecture is reminiscent of the classical Generalized Kuratowski Theorem [ Kur ] . As part of the Graph Minors Project , Neil Robertson and Paul Seymour were able to further generalize the Generalized Kuratowski Theorem to obtain the WQO ( Well Quasi Ordering ) Theorem stated below (cf. [ RS ] ). Their results were published in a series of twenty three journal papers totaling more than 700 pages from 1983 to 2004. Diestel, in his book on graph theory [ Di ] , says that this theorem dwarfs any other result in graph theory and may doubtless be counted among the deepest theorems that mathematics has to offer. Theorem 1.4 . (WQO). Each minor-closed class of graphs has only finitely many excluded minors. Equivalently, in any infinite set S of graphs, there must be a pair of graphs one of which is a minor of the other. Then, Robertson and Seymour proposed ideas for extending their Graph Minors Project to matroids. The challenge was taken up by Jim Geelen, Bert Gerards, and Geoff Whittle. Though it is not true that the WQO Theorem extends to all matroids. However, after extensive work for several years, they (Geelen et al.) announced the following slightly weaker theorem (cf. [ GGW , Theorem 6] ) significantly extending the WQO theorem to matroids. Theorem 1.5 . (Matroid WQO Theorem). For each finite field k and each minor-closed class of k -representable matroids, there are only finitely many k -representable excluded minors. According to their article [ GGW ] , to quote them: ‘We are now immersed in the lengthy task of writing up our results. Since that process will take a few years, we have written this article offering a high-level preview of the proof.’ The following result (communicated to us by J. Draisma), which is fairly easy to prove, provides a direct bridge between Conjecture 2.6 and Thorem 1.5 once we observe that \theta^{n,p+1}:\wedge^{p+1}(V_{n,p+1}^{*})\to\wedge^{p}(V_{n,p}^{*}) and \xi^{n+1,p}:\wedge^{p}(V_{n+1,p}^{*})\to\wedge^{p}(V_{n,p}^{*}) , at the level of matroids, correspond to contraction and deletion respectively. Theorem 1.6 . Let k be a finite field and assume that the set of k -points of the affine infinite Grassmannian \widetilde{\Gr}(\infty/2,V^{*}_{\infty}) , equipped with the Zariski topology, is S_{\infty} -noetherian. Then, matroids representable over k are well quasi ordered by the minor order. Acknowledgements: We are indebted to Jan Draisma for explaining to us his Conjecture 2.6 , showing its significance via his Theorem 1.6 to the Matroid WQO Theorem 1.5 , providing all the references [ Di ] , [ GGW ] , [ Kur ] , [ RS ] , [ Ro ] . This work was completed while the author was visiting the Institut des Hautes Études Scientifiques (Bures-sur-Yvette, France) during the fall semester of 2023, hospitality of which is gratefully acknowledged. 2. Infinite Grassmannian and the Matroid Minor Conjecture The base field in this note is any (including finite) field. Definition 2.1 . For positive integers n,p , let V_{n,p} be the vector space over k with basis \{x_{-n},x_{-n+1},\dots,x_{-1},x_{1},x_{2},\dots,x_{p}\}. Let \{x^{*}_{-n},x^{*}_{-n+1},\dots,x^{*}_{-1},x^{*}_{1},x^{*}_{2},\dots,x^{*}_{p}\} be the dual basis of the dual vector space V_{n,p}^{*} . Define the linear maps \eta=\eta^{n+1,p}:\wedge^{p}(V_{n,p})\hookrightarrow\wedge^{p}(V_{n+1,p}) induced from the standard inclusion of the bases and \beta=\beta^{n,p+1}:\wedge^{p}(V_{n,p})\hookrightarrow\wedge^{p+1}(V_{n,p+1}),\,\,\,\omega\mapsto\omega\wedge x_{p+1}. Dually, we get surjective maps \xi=\xi^{n+1,p}:\wedge^{p}(V^{*}_{n+1,p})\twoheadrightarrow\wedge^{p}(V^{*}_{n,p}) and \theta=\theta^{n,p+1}:\wedge^{p+1}(V^{*}_{n,p+1})\twoheadrightarrow\wedge^{p}(V^{*}_{n,p}),\,\,\,\omega\mapsto i_{x_{p+1}}\omega, where i is the interior multiplication. Let \Gr(p,V^{*}_{n,p}) be the Grassmannian of p -planes in V^{*}_{n,p} (For generalities on Grassmannians, see [ EH , \lx@sectionsign III.2.7] .) . Then, we have the Plücker embedding: \iota:\Gr(p,V^{*}_{n,p})\hookrightarrow\mathbb{P}\left(\wedge^{p}(V^{*}_{n,p})\right),\,\,A\mapsto\wedge^{p}(A),\,\,\text{for $A\in\Gr(p,V^{*}_{n,p})$}. Let \widetilde{\Gr}(p,V^{*}_{n,p}) be the corresponding affine cone (under the above embedding). Thus, we get an embedding \widetilde{\Gr}(p,V^{*}_{n,p})\hookrightarrow\wedge^{p}(V^{*}_{n,p}), where the image consists of the decomposable vectors (including the vector 0 ). This makes \widetilde{\Gr}(p,V^{*}_{n,p}) a closed irreducible (affine) subvariety of \wedge^{p}(V^{*}_{n,p}) . Now, the maps (1) \wedge^{p}(V_{n+1,p}^{*})\xleftarrow{\theta^{n+1,p+1}}\wedge^{p+1}(V_{n+1,p+1}^{*})\xrightarrow{\xi^{n+1,p+1}}\wedge^{p+1}(V_{n,p+1}^{*}) induce the restriction maps which are surjective: (2) \widetilde{\Gr}(p,V^{*}_{n+1,p})\xleftarrow{\widetilde{\theta}}\widetilde{\Gr}(p+1,V^{*}_{n+1,p+1})\xrightarrow{\widetilde{\xi}}\widetilde{\Gr}(p+1,V^{*}_{n,p+1}). To prove the existence of \widetilde{\theta} , we can write v_{1}\wedge\dots\wedge v_{p+1}=v^{\prime}_{1}\wedge\dots\wedge v^{\prime}_{p}\wedge(v^{\prime}_{p+1}+\alpha x^{*}_{p+1}),\,\,\text{where $v^{\prime}_{i}(x_{p+1})=0\forall 1\leq i\leq p+1$}, for some \alpha\in k . Thus, i_{x_{p+1}}(v_{1}\wedge\dots\wedge v_{p+1})=\pm\alpha v^{\prime}_{1}\wedge\dots\wedge v^{\prime}_{p} and hence \theta induces the map \widetilde{\theta} on the corresponding cones of Grassmannians. The existence of \widetilde{\xi} is trivial to

Read the complete text