A fast direct solver for structured linear systems by recursive skeletonization

TL;DR

This paper introduces a fast direct solver for structured linear systems using recursive skeletonization, enabling efficient factorization and inversion especially for multiple right-hand sides or ill-conditioned problems.

Contribution

It extends the Martinsson/Rokhlin method to 2D and 3D, providing a scalable, structured approach for boundary integral equations with significant speedups over existing methods.

Findings

Achieves ~100x speedup in the solution phase compared to fast multipole methods.

Precomputation costs are high but acceptable for multiple solves or ill-conditioned systems.

Complexity is O(N) in 2D and O(N^{3/2}) in 3D for precomputation.

Abstract

We present a fast direct solver for structured linear systems based on multilevel matrix compression. Using the recently developed interpolative decomposition of a low-rank matrix in a recursive manner, we embed an approximation of the original matrix into a larger, but highly structured sparse one that allows fast factorization and application of the inverse. The algorithm extends the Martinsson/Rokhlin method developed for 2D boundary integral equations and proceeds in two phases: a precomputation phase, consisting of matrix compression and factorization, followed by a solution phase to apply the matrix inverse. For boundary integral equations which are not too oscillatory, e.g., based on the Green's functions for the Laplace or low-frequency Helmholtz equations, both phases typically have complexity O(N) in two dimensions, where $N$ is the number of discretization points. In our current implementation, the corresponding costs in three dimensions are $O(N^{3/2})$ and $O(N \log N)$ for precomputation and solution, respectively. Extensive numerical experiments show a speedup of $\sim 100$ for the solution phase over modern fast multipole methods; however, the cost of precomputation remains high. Thus, the solver is particularly suited to problems where large numbers of iterations would be required. Such is the case with ill-conditioned linear systems or when the same system is to be solved with multiple right-hand sides. Our algorithm is implemented in Fortran and freely available.