On semiiterative methods generated by Faber polynomials

On semiiterative methods generated by Faber polynomials
复制标题

关于 Faber 多项式生成的半迭代方法

DOI:
--
复制
发表时间:
1989
期刊:
影响因子:
--
通讯作者:
M. Eiermann
M. Eiermann
中科院分区:
--
文献类型:
--
作者:
M. Eiermann

文献摘要

被引文献

相似文献

考虑非奇异线性代数方程组Ax =B,或不动点方程组x =Tx+c,其中T的特征值包含在复平面的某个紧子集Ω中。使用Faber多项式,迭代方法的形式 $$y_m:= mu _{m,0} left({c + Ty_{m - 1} } 8)+ mu _{m,1} y_{m - 1} + ldots + mu _{m,k} y_{m - k},$$ μm,0 ∞,∑μm,j=1(对所有m ∈ 1),其中包括Chebyshev迭代法和平稳k步方法.当Ω是一个矩形,我们还开发了四步方法,当优化的Chebyshev方法的收敛速度是穷人的情况下,产生快速收敛。
SummaryConsider a nonsingular system of linear algebraic equationsAx=b, or in fixed point formx=Tx+c, where the eigenvalues ofT are contained in some compact subset Ω of the complex plane. Using Faber polynomials, iteration methods of the form $$y_m : = mu _{m,0} left( {c + Ty_{m - 1} } ight) + mu _{m,1} y_{m - 1} + ldots + mu _{m,k} y_{m - k} ,$$ μm, 0 ≠, ∑μm, j=1 (for allm≧1), are constructed which include the Chebyshev iterative method as well as the stationaryk-step methods. When Ω is a rectangle, we also develop four-step methods which yield fast convergence in cases when the speed of convergence of the optimized Chebyshev method is poor.