Deformations, Derived Categories, and Multiparameter Persistence: A Theoretical Framework

TL;DR

This paper introduces a geometric framework using deformation theory and derived categories to analyze multiparameter persistence modules, addressing their algebraic complexity and stability issues.

Contribution

It unifies topological data analysis with deformation theory, providing explicit calculations and conjectures linking interleaving distances to derived convolution metrics.

Findings

Explicit calculations of Ext^1 groups reveal diverse deformation behaviors.

Obstruction classes in Ext^2 vanish in small cases but are inevitable in larger grids.

A conjecture relates interleaving distance to derived convolution metrics, suggesting bilipschitz equivalence.

Abstract

Multiparameter persistent homology has emerged as a powerful generalization of topological data analysis, capable of encoding multivariate filtrations. However, the algebraic complexity of multiparameter persistence modules, marked by wild representation type, poses fundamental obstacles to classification, stability, and interpretability. In this paper, we propose a unifying theoretical framework that brings together deformation theory and derived categories to study multiparameter persistence from a geometric perspective. A central contribution is a comprehensive conceptual dictionary (Table 1) bridging topological data analysis and deformation theory, which interprets perturbations as deformations and stability as smoothness of moduli spaces. We present explicit calculations of extension groups \(Ext^1\) for concrete multiparameter modules over small posets, revealing diverse behaviors ranging from unexpected rigidity to large families of deformations. We further investigate obstruction classes in \(Ext^2\); while these vanish in our specific examples over the square poset, we demonstrate their inevitability in larger grids (e.g., \(3 \times 3\)) via global dimension arguments, highlighting a qualitative transition in the geometry of moduli spaces. Finally, we formulate a unified conjecture relating the interleaving distance to derived convolution metrics, establishing a bilipschitz equivalence at the level of the derived category of persistence modules. Together, these results shift the perspective on multiparameter persistence from static classification to the geometry of families, opening new avenues for invariants, stability theorems, and moduli-based analysis.