[論文レビュー] Singular Hodge theory for combinatorial geometries

Tom Braden, June Huh+3 more|arXiv (Cornell University)|Oct 13, 2020

Advanced Combinatorial Mathematics参考文献 34被引用数 30

ひとこと要約

本論文は matroid の intersection cohomology モジュールを定義し、Poincaré duality、hard Lefschetz、および Hodge–Riemann relations を証明する。これにより matroid の Kazhdan–Lusztig 多項式の非負性と単調性（およびそれらの equivariant バージョン）が得られる。

ABSTRACT

We introduce the intersection cohomology module of a matroid and prove that it satisfies Poincaré duality, the hard Lefschetz theorem, and the Hodge-Riemann relations. As applications, we obtain proofs of Dowling and Wilson's Top-Heavy conjecture and the nonnegativity of the coefficients of Kazhdan-Lusztig polynomials for all matroids.

研究の動機と目的

• Motivate and formalize a Hodge-theoretic framework for matroids via intersection cohomology methods.
• Prove Poincaré duality, hard Lefschetz, and Hodge–Riemann relations for the intersection cohomology module of a matroid.
• Derive nonnegativity of Kazhdan–Lusztig polynomials and unimodality of Z-polynomials in both ordinary and equivariant settings.
• Connect combinatorial matroid theory with geometric and representation-theoretic structures to resolve conjectures on Top-Heavy distributions.

提案手法

• Define the graded Möbius algebra H(M) and the augmented Chow ring CH(M) of a matroid.
• Construct the intersection cohomology module IH(M) as a canonical indecomposable summand of CH(M).
• Prove that IH(M) satisfies the Kähler package: Poincaré duality, hard Lefschetz, and Hodge–Riemann relations (with respect to a suitable ample class).
• Use a stratification-based spectral sequence and degeneracy arguments to relate IH(M) to local data from contractions/localizations M_F.
• Deduce nonnegativity of the Kazhdan–Lusztig polynomial P_M(t) and unimodality of Z_M(t); extend to equivariant versions P_M^Γ(t), Z_M^Γ(t).
• Provide a combinatorial and geometric framework that unifies Kazhdan–Lusztin–Stanley theory with matroid theory.

実験結果

リサーチクエスチョン

• RQ1 Can the intersection cohomology of a matroid be realized as a canonical, representation-theoretically meaningful module?
• RQ2 Do Poincaré duality, hard Lefschetz, and Hodge–Riemann relations hold for IH(M) in full generality for all matroids?
• RQ3 Do Kazhdan–Lusztig polynomials of matroids have nonnegative coefficients, and are Z-polynomials unimodal, including in equivariant settings?
• RQ4 How does equivariance under a group action Γ influence the structure and positivity of P_M^Γ(t) and Z_M^Γ(t)?
• RQ5 Can the topology of realizable cases illuminate non-realizable matroids via a purely combinatorial/representation-theoretic approach?

主な発見

• IH(M) satisfies the Poincaré duality pairing and the hard Lefschetz isomorphism for degrees up to d/2.
• IH(M) fulfills the Hodge–Riemann relations, yielding positive-definite forms on primitive components.
• P_M(t) has nonnegative coefficients (equivariant version P_M^Γ(t) has honest representations as coefficients).
• Z_M(t) is unimodal with coefficients corresponding to subrepresentations, and Z_M^Γ(t) is unimodal in the equivariant setting.
• Equivariant statements extend the nonnegativity/unimodality to Γ-actions, with coefficients interpreted as honest representations of Γ.
• Foundations imply nonnegativity of inverse Kazhdan–Lusztig polynomial Q_M(t) as well, with coefficients as honest representations.
• Monotonicity: P_M^Γ(t) − P_{M_F}^{Γ}(t) has coefficients that are honest representations of the stabilizer Γ_F when F is a nonempty flat fixed by Γ.