Simple non-extensive sparsification of the hierarchical matrices (Q2190667)

It is well known that a number of problems arising in the discretization of boundary integral equations lead to matrices that can be well-approximated by hierarchical block low-rank (\(H\) matrices) as well as the fact that the development of the \(H\) matrices is the \(H^2\) matrices, which are the hierarchical block low-rank matrices with nested bases. In this paper the authors propose a new representation of \(H^2\) matrices. It is shown that the \(H^2\) factorization of a matrix \(A\in \mathbb{R}^{N\times N}\) is equivalent to the factorization \(A = USV^T\), where \(S\in \mathbb{R}^{N\times N}\) is a sparse matrix. Once the factorization is built, the authors substitute a solution of the system \(Ax = b\), by a solution of the system with the sparse matrix: \(Sy = U^Tb\), where \(x = Vy\).