scientific article; zbMATH DE number 4045501

From MaRDI portal
Publication:3783089





[For the entire collection see Zbl 0619.00012.] In former papers by the first author [Springer Lect. Notes Econ. Math. Systems 259, 79-89 (1985; Zbl 0575.90071)] and by the first author and \textit{P. Zencke} [Springer Lect. Notes Control Inf. Sci. 66, 132-155 (1985; Zbl 0561.90084)] a method has been proposed to compute approximations to eigenvalues and eigenfunctions of the Laplacean by defect-minimization. This method leads to a parametric (linear) semi- infinite programming problem P(\(\lambda)\), the parameter \(\lambda\) of which is to be determined such that a certain function \(\epsilon\) of the value v(\(\lambda)\) and the solution z(\(\lambda)\) of P(\(\lambda)\) has a local minimum. In general \(\epsilon\) depends in a nondifferentiable way on \(\lambda\). A method of \textit{R. Mifflin} [Math. Program. 28, 50-71 (1984; Zbl 0528.49024)] is taken to find the local minima of \(\epsilon\). As this requires the knowledge of generalized gradients of \(\epsilon\) differentiability of v(\(\lambda)\) and z(\(\lambda)\) is studied under appropriate regularity conditions. Some numerical examples are given.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3783089)