Bounds for eigenvalues of second-order elliptic differential operators (Q1186489)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Bounds for eigenvalues of second-order elliptic differential operators |
scientific article |
Statements
Bounds for eigenvalues of second-order elliptic differential operators (English)
0 references
28 June 1992
0 references
The article is concerned with the elliptic eigenvalue problem \[ L(u)=(1/q)[-\nabla^ TA\nabla u+cu]=\lambda u\quad\text{ in }\Omega\subset\mathbb{R}^ n, \] \(u\in{\mathcal D}=\{v\in H_ 2(\Omega):v=0\) on \(\Gamma_ 0\); \(\nabla^ TvA\nu+\gamma v=0\) on \(\Gamma_ 1\}\); \(\Omega\) is bounded, \(\partial\Omega=\Gamma_ 0\cup\Gamma_ 1\) \((\Gamma_ 0\cap\Gamma_ 1=\emptyset)\) is Lipschitz, \(\nu\) is the outer unit normal field on \(\partial\Omega\). Upper bounds for the eigenvalues of the weak problem are obtained by the Rayleigh-Ritz-method. To get lower bounds a generalization of Kato's inequality [\textit{T. Kato}, J. Phys. Soc. Japan 4, 334-339 (1949)] is presented and a homotopy algorithm connecting simple eigenvalue problems with the stated one is developped. A numerical procedure using a finite- element discretization and an inverse (subspace-) iteration process is described and applied to compute approximate eigenpairs and bounds for them.
0 references
weak problem
0 references
Rayleigh-Ritz-method
0 references
Kato's inequality
0 references
homotopy algorithm
0 references
finite-element discretization
0 references
iteration process
0 references
0 references
0 references
0 references