A NOTE ON THE RELATIVE ERROR FOR THE EIGENVALUES OF THE STURM-LIOUVILLE PROBLEM

DOI10.1515/DEMA-1999-0213MaRDI QIDQ4258346zbMATH OpenOpenAlexFDO

Authors Przemysław Kosowski

Publication date 28 July 2000

Published in Demonstratio Mathematica (Search for Journal in Brave)

Copyright license Creative Commons Attribution-NonCommercial 4.0 International

Full work available at URL https://doi.org/10.1515/dema-1999-0213

zbMATH Keywords

AAdHP correctionCourant-Fischer-Poincaré minimax theoremeigenvaluesfinite element methodrelative errorSturm-Liouville problem

Mathematics Subject Classification ID

Sturm-Liouville theory (34B24) Perturbation theory of linear operators (47A55) Linear symmetric and selfadjoint operators (unbounded) (47B25)

Review text

Referring to \textit{P. Kosowski} [Linear Operators (\textit{J. Janas, F. H. Szafraniec} and \textit{J. Zemánek}, eds.), Banach Center Publications 38, Warsaw: Polish Academy of Sciences, Inst. of Mathematics (1997; Zbl 0863.00036), 183-186 (1997; Zbl 0878.34017)] and \textit{J. D. Pryce} [Numerical solution of Sturm-Liouville problems, Oxford: Clarendon Press (1993; Zbl 0795.65053)] the author presents the following theorem: If \(\|q_1-q_2\|_{\infty}\leq\epsilon\), then \(|\lambda_k^{(1)}- \lambda_k^{(2)}|\leq\epsilon\) for each \(k=1,2,3,\dots\) Here, \(\lambda_k^{(j)}\) stand for eigenvalues of the Sturm-Liouville operator \(L_j\): \[ L_ju=-u+q_ju, \] \[ D(L_1)=D(L_2)=\{u\in H^2(a,b):\alpha u(a)+\alphau(a)=0 ,\;\beta u(b)+\betau(b)=0 \}, \] \(q_j\) are real-valued and belong to \(C([a,b])\), \(\alpha\), \(\alpha\), \(\beta\), \(\beta\) are real constants and \(\alpha^2 +\alpha^{\prime 2} >0\), \(\beta^2+\beta^{\prime 2}>0\). The author gives a numerical (calculated by the finite element method with AAdHP correction) counterexample to the following statement of \textit{F. de Hoog} and \textit{J. W. Paine} [J. Aust. Math. Soc., Ser. B 21, 365-383 (1980; Zbl 0417.34046)]: \[ |\lambda_k^{(1)}-\lambda_k^{(2)}|\leq\|q_1-q_2\|_2. \] The sequel is devoted to estimating the relative error \[ \left|{\lambda_k^{(1)}-\lambda_k^{(2)}\over\lambda_k^{(1)}}\right| \] via the spectral radius of the operator \(L_1^{-1}F\), where the operator \(F\) acts in \(L^2(a,b)\) according to the formula \(Fu=(q_1-q_2)u\). Results of numerical calculations are presented.

Recommendations

• The relative error in the Pruess method for Sturm-Liouville problems
• Guaranteed Error Bounds for Eigenvalues of Singular Sturm-Liouville Problems
• An asymptotic estimate of the error of eigenvalues in the method of finite elements for the Sturm-Liouville problem
• scientific article; zbMATH DE number 3948772
• Error estimate of eigenvalues of perturbed second-order discrete Sturm-Liouville problems
• Erratum: Eigenvalue and eigenfunction asymptotics for regular Sturm- Liouville problems
• On a correction of Numerov-like eigenvalue approximations for Sturm- Liouville problems
• Constructing of asymptotic estimates for eigenvalues of the Sturm-Liouville operator with arbitrary accuracy
• Accurate Approximation of Eigenvalues and Zeros of Selected Eigenfunctions of Regular Sturm-Liouville Problems
• An Inequality for Eigenvalues of Sturm-Liouville Problems

Cited in

(1)

• The relative error in the Pruess method for Sturm-Liouville problems

This page was built for publication: A NOTE ON THE RELATIVE ERROR FOR THE EIGENVALUES OF THE STURM-LIOUVILLE PROBLEM

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4258346)