Stable Parallel Solvers for General Tridiagonal Linear Systems

摘要

We analyze the problem of solving tridiagonal linear systems on parallel computers. In the case where the coefficient matrix is symmetric positive definite and/or diagonally dominant, this problem has been extensively discussed. More difficult is the case where the coefficient matrix does not satisfy the above requirements. Here we present two algorithms which are able to handle general tridiagonal matrices, and compare them on some test problems carried out on a distributed memory parallel computer.