Time Integration and Conjugate Gradient Methods for the Incompressible Navier-Stokes Equations

摘要

This paper addresses one class of iterative methods that are promising for cost effective Navier-Stokes simulations. The approach is based on the conjugate gradient (CG) method for solving linear algebraic systems with a symmetric positive-definite (SPD) matrix. These methods have many important characteristics: monotone convergence (in an appropriate error norm), optimal error minimization, finite termination, the direct exploitation of sparseness in the matrix, and no need to estimate ''iteration parameters.'' 46 refs., 1 tab. (ERA citation 11:033487)