Speaker：Junfeng Yin（Tongji University）

Time：2022-04-16, 10:00

Location：Tencent Meeting ID：507663044（No Passward）

Abstract:

Iteration methods play important roles in the solution of linear equations. A new algorithm is proposed to accelerate the surrounding algorithms based on a sequence of Householder reflections by restarted techniques for solving large linear equations. Theoretical analysis show that this method converges and the convergence rate is estimated. Numerical experiments verify the efficiency of restarted strategies, which can greatly decrease the number of iteration steps and save the elapsed CPU time.