LP-Form Inclusion Functions for Global Optimization

Abstract

A new class of methods is presented for finding the global extrema of real-valued functions using point values and gradient inclusions. These methods construct polyhedral envelopes for each subfunction within regions generated by a subdivision strategy. The range of the envelopes determines bounds on the functions and can be computed by linear programs. Partial monotonicity conditions are presented that reduce the effective dimension or yield exact bounds on finite regions.

Cite this paper

@inproceedings{Mentzer2001LPFormIF, title={LP-Form Inclusion Functions for Global Optimization}, author={Stuart G. Mentzer}, year={2001} }