Iterative solution of linear systems with sparse matrices of big order arising from approximation of the linear differential equations of elasticity with large scale of Lame coefficients alteration. A comparative analysis of the conjugate gradient and conjugate residuals methods with and without incomplete factorization preconditioning is given. Performance results of the codes for these algorithms are presented for different formats of sparse matrix representation, for using of some special mathematical functions from different math software, and for different computer platforms. A parallelization of the algorithms is considered for the systems with distributed and shared memory.
Mail to Webmaster www@www-sbras.nsc.ru |
|Home Page| |English Part| |
Go to Home |