# Low-rank Matrix Optimization Using Polynomial-filtered Subspace Extraction

## TL;DR

This paper introduces a polynomial-filtered subspace extraction framework to accelerate first-order methods for low-rank matrix optimization, achieving comparable convergence with reduced computational cost.

## Contribution

The paper proposes a novel polynomial-filtered subspace extraction approach that reduces eigen-decomposition costs in low-rank matrix optimization algorithms, with theoretical guarantees and practical speedups.

## Key findings

- Achieves multi-fold speedups in numerical experiments.
- Maintains convergence speed with polynomial degree growing as O(log k).
- Constant polynomial degree suffices when warm-start property is used.

## Abstract

In this paper, we study first-order methods on a large variety of low-rank matrix optimization problems, whose solutions only live in a low dimensional eigenspace. Traditional first-order methods depend on the eigenvalue decomposition at each iteration which takes most of the computation time. In order to reduce the cost, we propose an inexact algorithm framework based on a polynomial subspace extraction. The idea is to use an additional polynomial-filtered iteration to extract an approximated eigenspace, and project the iteration matrix on this subspace, followed by an optimization update. The accuracy of the extracted subspace can be controlled by the degree of the polynomial filters. This kind of subspace extraction also enjoys the warm start property: the subspace of the current iteration is refined from the previous one. Then this framework is instantiated into two algorithms: the polynomial-filtered proximal gradient method and the polynomial-filtered alternating direction method of multipliers. We give a theoretical guarantee to the two algorithms that the polynomial degree is not necessarily very large. They share the same convergence speed as the corresponding original methods if the polynomial degree grows with an order $\Omega(\log k)$ at the $k$-th iteration. If the warm-start property is considered, the degree can be reduced to a constant, independent of the iteration $k$. Preliminary numerical experiments on several low-rank matrix optimization problems show that the polynomial filtered algorithms usually provide multi-fold speedups.