A New Effective Iterative Method for Sparse linear Systems and Remarks on Parallel Computing
This paper introduces an efficient Iterative method based on decomposition of subspace for solving large sparse linear systems. The key to the efficiency of this method is to make full use of the numerical discrete schemes, which determine the matrix characteristics of the linear systems involved. This method changes the solution of the whole sparse linear systems into some small-scale problems. A simple proof for convergence of the new iterative method is described. Numerical experiments show that the method is more effective and faster than the Krylov subspace counterparts (generalized conjugate residual GCR). In parallel computing, the new method only needs local communication, thus can be natural parallelled. By reducing most of matrix-vector multiplications, the new method is promising for mass-scale parallel computing.
Iterative Method Sparse Linear Systems Parallel Computing
Gong XiPing Zhang LiLun Wu JianPing Zhao Wen Tao
School of Computer Science National University of Defence Technology Changsha, Hunan, China
国际会议
成都
英文
1917-1920
2010-12-17（万方平台首次上网日期，不代表论文的发表时间）