Optimal Algorithms for Linear Algebra in the Current Matrix Multiplication Time

TL;DR

This paper presents near-optimal algorithms for fundamental linear algebra problems that operate in time proportional to the current matrix multiplication exponent, leveraging advanced subspace embeddings and randomized transforms.

Contribution

It introduces a constant-factor subspace embedding with optimal size and time complexity, enabling the first optimal runtime algorithms for key linear algebra tasks.

Findings

Achieves $d^{ ext{omega}}$ time complexity for core problems

Develops a new subspace embedding using stacked SRHTs and semidefinite programming

Provides algorithms for basis finding, regression, and leverage score sampling with optimal runtime

Abstract

We study fundamental problems in linear algebra, such as finding a maximal linearly independent subset of rows or columns (a basis), solving linear regression, or computing a subspace embedding. For these problems, we consider input matrices $\mathbf{A}\in\mathbb{R}^{n\times d}$ with $n > d$. The input can be read in $\text{nnz}(\mathbf{A})$ time, which denotes the number of nonzero entries of $\mathbf{A}$. In this paper, we show that beyond the time required to read the input matrix, these fundamental linear algebra problems can be solved in $d^{\omega}$ time, i.e., where $\omega \approx 2.37$ is the current matrix-multiplication exponent. To do so, we introduce a constant-factor subspace embedding with the optimal $m=\mathcal{O}(d)$ number of rows, and which can be applied in time $\mathcal{O}\left(\frac{\text{nnz}(\mathbf{A})}{\alpha}\right) + d^{2 + \alpha}\text{poly}(\log d)$ for any trade-off parameter $\alpha>0$, tightening a recent result by Chepurko et. al. [SODA 2022] that achieves an $\exp(\text{poly}(\log\log n))$ distortion with $m=d\cdot\text{poly}(\log\log d)$ rows in $\mathcal{O}\left(\frac{\text{nnz}(\mathbf{A})}{\alpha}+d^{2+\alpha+o(1)}\right)$ time. Our subspace embedding uses a recently shown property of stacked Subsampled Randomized Hadamard Transforms (SRHT), which actually increase the input dimension, to "spread" the mass of an input vector among a large number of coordinates, followed by random sampling. To control the effects of random sampling, we use fast semidefinite programming to reweight the rows. We then use our constant-factor subspace embedding to give the first optimal runtime algorithms for finding a maximal linearly independent subset of columns, regression, and leverage score sampling. To do so, we also introduce a novel subroutine that iteratively grows a set of independent rows, which may be of independent interest.