[Paper Review] Questions about Boij-Söderberg theory

Daniel Erman, Steven V Sam|arXiv (Cornell University)|Jun 6, 2016

Commutative Algebra and Its Applications46 references3 citations

TL;DR

This paper investigates open questions in Boij–Söderberg theory, focusing on the classification of Betti tables and cohomology tables of graded modules and vector bundles via convex cones and duality. It establishes that every ray in the Betti and cohomology cones can be realized by equivariant objects over a field of characteristic zero, offering a structural explanation for the existence of such realizations and raising deeper questions about intrinsic geometric and representation-theoretic reasons behind them.

ABSTRACT

Boij-Söderberg theory focuses on the properties and duality relationship between two types of numerical invariants. One side involves the Betti table of a graded free resolution over the polynomial ring. The other side involves the cohomology table of a coherent sheaf on projective space. We discuss open questions and problems in Boij-Söderberg theory.

Motivation & Objective

• To investigate the structural reasons behind the existence of equivariant realizations of Betti and cohomology tables in Boij–Söderberg theory.
• To examine whether every extremal ray in the cone of Betti tables of finite length modules and cohomology tables of vector bundles can be realized by equivariant objects.
• To explore open directions in toric varieties, equivariant resolutions, monomial ideals, and asymptotic syzygy behavior.
• To identify fundamental open questions that could guide future research in Boij–Söderberg theory and its extensions.

Proposed method

• Analyzing the convex cone structure of Betti tables of modules of codimension ≥ c and cohomology tables of sheaves of complementary dimension.
• Using the duality between Betti tables and cohomology tables as a central organizing principle in the theory.
• Applying results from equivariant representation theory, particularly the existence of equivariant supernatural bundles over GL(n+1)-equivariant projective spaces.
• Leveraging known results on pure resolutions and their extremal rays to characterize the cone of Betti tables.
• Extending the theory to multigraded settings, such as toric varieties and multigraded Cox rings.
• Investigating asymptotic behavior of Veronese embeddings and their Boij–Söderberg decompositions as the degree d → ∞.

Experimental results

Research questions

• RQ1 Is there an intrinsic geometric or representation-theoretic reason why every Betti table of a finite length module and every cohomology table of a vector bundle can be realized (up to scalar multiple) by an equivariant object over a field of characteristic zero?
• RQ2 Can the Boij–Söderberg cone for toric varieties, such as P¹×P¹, be fully described using multigraded analogues of pure resolutions and duality?
• RQ3 Do similar stabilization phenomena in Betti tables occur for powers of monomial ideals beyond those already proven by Mayes-Tang for Engström’s conjecture?
• RQ4 What is the asymptotic behavior of the Boij–Söderberg decomposition of Betti tables for Veronese embeddings as the degree d → ∞?
• RQ5 Can the framework of Fløystad’s triple of homological data be extended meaningfully to equivariant or toric settings?

Key findings

• Every ray in the cone of Betti tables of finite length modules over a polynomial ring S = k[x₀,…,xₙ] can be realized by a GLₙ₊₁-equivariant module over a field of characteristic zero.
• Every ray in the cone of cohomology tables of vector bundles on Pⁿ can be realized by a GLₙ₊₁-equivariant vector bundle.
• The existence of equivariant realizations for all extremal rays is a consequence of the existence of equivariant supernatural bundles, which provide explicit constructions of pure resolutions.
• The theory of pure resolutions and their duality with cohomology tables underpins the classification of extremal rays in the Betti and cohomology cones.
• Asymptotic syzygy results suggest that for Veronese embeddings, the Betti tables stabilize in a way that depends only on the dimension of the variety, not on finer geometric data.
• Partial progress exists in multigraded settings, such as Z²-graded rings and toric varieties, but a complete description of the cones remains open.