Symbolic powers and integral closures via extremal ideals
Trung Chau, Art Duval, Sara Faridi, Thiago Holleben, Susan Morey, Liana Şega
TL;DR
The paper develops a universal framework to study powers and symbolic powers of square-free monomial ideals by embedding them into the powers and symbolic powers of a single highly symmetric extremal ideal $\\mathcal{E}_q$ via a ring homomorphism $\\psi_I$. It proves that $\\overline{I^r} =\\psi_I(\\overline{\\mathcal{E}_q^r})R$ and $I^{(r)} =\\psi_I(\\mathcal{E_q^{(r)}})R$, providing test elements $g(r,q)$ and $h(r,q)$ to decide equality with ordinary powers and to detect non-equality in many cases. The framework yields explicit descriptions and generators for $I^{(2)}$, sharp bounds for integral defects, symbolic defects, and resurgences, and translates many computations into linear programming problems on $\\mathcal{E}_q$, while preserving LCM structures to transfer Betti-number information from the extremal ideal to any $I$. This reduces infinitely many algebraic questions to a finite, symmetric model, enabling concrete algorithmic and geometric approaches to longstanding containment and homological problems.
Abstract
This paper demonstrates that extremal ideals can be used to great effect to compute integral closures of powers and symbolic powers of square-free monomial ideals. We show that the generators of these powers are images of the generators of the corresponding powers of extremal ideals under a specific ring homomorphism. Extremal ideals provide sharp bounds for a variety of invariants widely studied in the literature, including resurgence, asymptotic resurgence, and symbolic defect, as well as Betti numbers of symbolic powers and of integral closures of powers of square-free monomial ideals. When restricted to the class of extremal ideals, algebraic computations are reduced to problems of discrete geometry and linear programming, allowing the use of a wide variety of techniques. As a result, in situations where computations are feasible for extremal ideals, we provide concrete sharp bounds for many of these invariants. Our methods reduce finding homological invariants and algebraic constructions for infinitely many ideals to computations for a single highly symmetric ideal, based solely on the number of generators.
