论文信息 - An incomplete inverse as a preconditioner for the conjugate gradient method

An incomplete inverse as a preconditioner for the conjugate gradient method

Abstract This paper introduces a new preconditioner with a super convergence for the conjugate gradient method for solving a system of linear equations in a banded structure. The idea is based upon an innovative method of the author to decompose a symmetric matrix into its inverse, and the preconditioner is the result of an incomplete decomposition, i.e., an approximation to the inverse. By the presented preconditioner, the rate of convergence can be significantly improved. Examples to demonstrate the improvement of rate of convergence and the speedup implemented on an Alliant/FX8 computer are included.

Jenn-Ching Luo | J. Luo

[1] Isaac Fried. A gradient computational procedure for the solution of large problems arising from the finite element discretization method , 1970 .

[2] Rami Melhem,et al. Toward Efficient Implementation of Preconditioned Conjugate Gradient Methods On Vector Supercomputers , 1987 .

[3] G. Golub,et al. Block Preconditioning for the Conjugate Gradient Method , 1985 .

[4] Charbel Farhat,et al. A general approach to nonlinear FE computations on shared-memory multiprocessors , 1989 .

[5] Y. Saad,et al. Practical Use of Polynomial Preconditionings for the Conjugate Gradient Method , 1985 .

[6] J. Luo. Fictitious array techniques for storing and accessing a symmetric matrix , 1991 .

[7] Thomas J. R. Hughes,et al. New alternating direction procedures in finite element analysis based upon EBE approximate factorizations. [element-by-element] , 1983 .

[8] Gene H. Golub,et al. Some History of the Conjugate Gradient and Lanczos Algorithms: 1948-1976 , 1989, SIAM Rev..

[9] A note on parallel processing , 1992 .

[10] Thomas J. R. Hughes,et al. Solution algorithms for nonlinear transient heat conduction analysis employing element-by-element iterative strategies , 1985 .

[11] P. Ducksbury. Parallel array processing , 1986 .