Gauss-Newton and inverse Gauss-Newton methods for coefficient identification in linear elastic systems

From MaRDI portal
(Redirected from Publication:411464)





This paper is concerned with an iterative solution of some type of parameter identification problems in linear elastic systems. As shown in the first section, such problems can be formulated by using suitable Hilbert spaces as equations with the form \[ C_j(B^*KB)^{-1} \; F_j = Z_j,\quad j=1, \dots ,J,\tag{1} \] where \(K\) are the identification parameters to be determined in some optimal sense and the remaining operators in (1) are obtained by observations or data of the problem. Here the author studies the so called case 2 in which \((B^*KB)\) exists as a bounded operator but \(B\) and \(B^*\) are not invertible. For solving the nonlinear least squares problems associated to (1) the author proposes an inverse Gauss-Newton iterative method, studying the convergence of this process. Some computational considerations for the effective solution are given. Finally, a discretized version of a linear second order boundary value problem is taken as a test problem to compare the iterations with standard Newton and inverse Newton iterations.











This page was built for publication: Gauss-Newton and inverse Gauss-Newton methods for coefficient identification in linear elastic systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q411464)