通过带位移逆迭代求解拉普拉斯算子的特征值和特征函数
谱理论
2012-08-02 v2 数值分析
摘要
本文提出了一种迭代方法,其灵感来自有限线性代数中的带位移逆迭代技术,旨在寻找任意有界域上具有齐次Dirichlet边界条件的拉普拉斯算子的特征值和特征函数。该方法采用直接的泛函分析方法,并不将拉普拉斯算子的特征值近似为有限线性算子的特征值。它基于在节面之外的一致收敛性,可以产生一种简单快速的算法,以最小的计算需求来计算特征值,而不是使用有限线性代数中普遍存在的Rayleigh商。此外,还引入了相关的无限维Sobolev空间中Rayleigh商的一种替代表达式,该表达式避免了梯度的积分,并被证明更有效。该方法也可用于产生任意给定函数的谱分解。
引用
@article{arxiv.1011.3266,
title = {Eigenvalues and eigenfunctions of the Laplacian via inverse iteration with shift},
author = {Rodney Josué Biezuner and Grey Ercole and Breno Loureiro Giacchini and Eder Marinho Martins},
journal= {arXiv preprint arXiv:1011.3266},
year = {2012}
}
备注
In this version the numerical tests in Section 6 were considerably improved and the Section 5 entitled "Normalization at each step" was introduced. Moreover, minor adjustments in the Section 1 (Introduction) and in the Section 7 (Fi nal Comments) were made. Breno Loureiro Giacchini was added as coauthor