Sparsifying Sums of Positive Semidefinite Matrices

TL;DR

This paper introduces a new theory for sparsifying sums of positive semidefinite matrices, providing instance-specific bounds and applying it to improve Cayley graph sparsification.

Contribution

It develops a novel PSD matrix sparsification framework based on the connectivity threshold, enabling tighter bounds and extending Cayley graph sparsification to general groups.

Findings

Achieves sparsifiers with O(ε^{-2} N^*(A) (log n)(log r)) matrices

Proves N^*(A) is necessary for sparsification with ε<0.99

Improves Cayley graph sparsification bounds to O(ε^{-2} log^4 N) generators

Abstract

In this paper, we revisit spectral sparsification for sums of arbitrary positive semidefinite (PSD) matrices. Concretely, for any collection of PSD matrices $\mathcal{A} = \{A_1, A_2, \ldots, A_r\} \subset \mathbb{R}^{n \times n}$, given any subset $T \subseteq [r]$, our goal is to find sparse weights $\mu \in \mathbb{R}_{\geq 0}^r$ such that $(1 - \epsilon) \sum_{i \in T} A_i \preceq \sum_{i \in T} \mu_i A_i \preceq (1 + \epsilon) \sum_{i \in T} A_i.$ This generalizes spectral sparsification of graphs which corresponds to $\mathcal{A}$ being the set of Laplacians of edges. It also captures sparsifying Cayley graphs by choosing a subset of generators. The former has been extensively studied with optimal sparsifiers known. The latter has received attention recently and was solved for a few special groups (e.g., $\mathbb{F}_2^n$). Prior work shows any sum of PSD matrices can be sparsified down to $O(n)$ elements. This bound however turns out to be too coarse and in particular yields no non-trivial bound for building Cayley sparsifiers for Cayley graphs. In this work, we develop a new, instance-specific (i.e., specific to a given collection $\mathcal{A}$) theory of PSD matrix sparsification based on a new parameter $N^*(\mathcal{A})$ which we call connectivity threshold that generalizes the threshold of the number of edges required to make a graph connected. Our main result gives a sparsifier that uses at most $O(\epsilon^{-2}N^*(\mathcal{A}) (\log n)(\log r))$ matrices and is constructible in randomized polynomial time. We also show that we need $N^*(\mathcal{A})$ elements to sparsify for any $\epsilon < 0.99$. As the main application of our framework, we prove that any Cayley graph can be sparsified to $O(\epsilon^{-2}\log^4 N)$ generators. Previously, a non-trivial bound on Cayley sparsifiers was known only in the case when the group is $\mathbb{F}_2^n$.