Optimality conditions and dual formulations for programming problems over cone domains
Date
Authors
Major Professor
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Abstract
Let P be an open set in E('n), and C be an arbitrary cone in E('n). LetF and g be functions from P to E('l) and E('n) respectively;Consider the following minimization problem, Problem P, find anx('0) (epsilon) E('n), if it exists, such that;F(x('0)) = min F(x); x (epsilon) X; x('0) (epsilon) X;where,;X = x : x (epsilon) P (L-HOOK) E('n), g(x) (epsilon) C (L-HOOK) E('m);and F is differentiable at x('0);Associated with the minimization problem is a modified Kuhn-Tucker stationary point problem over cone domains. Find anx('0) (epsilon) P (L-HOOK) E('n) and u('0) (epsilon) -C* (L-HOOK) E('n) (C* is the polar cone of C) such that;(DEL)'F(x('0)) + u('0)'(DEL)g(x('0)) = 0; u('0)'g(x('0)) = 0; g(x('0)) (epsilon) C;Necessary and sufficient optimality conditions are established between Problem P and the modified Kuhn-Tucker stationary point problem for a certain class of nonlinear programming problems over arbitrary cone domains;These results are used to prove a modified Farkas Lemma over degenerate and nondegenerate cone domains which uses only a "partial" linear duality theorem;A quadratic programming problem is considered. Its dual problem is constructed in a natural way over degenerate and nondegenerate cone domains and quadratic duality is established between the two problems.