Optimal rank matrix algebras preconditioners

TL;DR

This paper develops an efficient algorithm for constructing optimal low-rank preconditioners for structured matrices like Toeplitz and Hankel, significantly improving iterative solver convergence.

Contribution

It extends the black-dot algorithm to a broader class of matrix algebras, enabling better preconditioning for Toeplitz-like systems.

Findings

The algorithm efficiently computes P in various low-complexity algebras.

It improves convergence rates of iterative methods for Toeplitz and Hankel matrices.

Theoretical results on the existence of decompositions for structured matrices.

Abstract

When a linear system Ax = y is solved by means of iterative methods (mainly CG and GMRES) and the convergence rate is slow, one may consider a preconditioner P. The use of such preconditioner changes the spectrum of the matrix defining the system and could result into a great acceleration of the convergence rate. The construction of optimal rank preconditioners is strongly related to the possibility of splitting A as A = P + R + E, where E is a small perturbation and R is of low rank. In the present work we extend the black-dot algorithm for the computation of such splitting for P circulant, to the case where P is in L, for several known low-complexity matrix algebras L. The algorithm so obtained is particularly efficient when A is Toeplitz plus Hankel like. We finally discuss in detail the existence and the properties of the decomposition A = P + R + E when A is Toeplitz, also extending to the phi-circulant and Hartley-type cases some results previously known for P circulant.