Computation of the largest value of functions of several variables

From MaRDI portal





The author is concerned with the problem of finding the global maximum of a real function in an n-dimensional interval D. He defines a family w of nets of points that cover D. For each \(W\in w\) an 'error' \(\Delta\) is defined with respect to two suitable classes of functions \(C_ 1\), \(C_ 2\); that is, \[ \Delta(C_ j,W)=\sup_{f\in C_ j}\{f^*_ D- f^*_ W\},\quad j=1,2 \] where \(f^*_ D\) and \(f^*_ W\) are the largest values of f on D and W respectively. Then the author states four theorems concerning both the net that minimizes \(\Delta(C_ j,W)\) under the hypothesis that the number of nodes is bounded, and the net with the least number of nodes subject to the constraint \(\Delta(C_ j,W)<2\epsilon\), \(\epsilon\) fixed.











This page was built for publication: Computation of the largest value of functions of several variables

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3321846)