Globally Convergent Simplicial Algorithm for Stationary Point Problems on Polytopes

摘要

The authors propose a simplicial variable dimension restart algorithm for the stationary point problem or variational inequality problem: given a convex polytope C of (R sup n) and a continuous function f: C -> Rn, find a point x bar of C such that f(x bar).x bar = or > f(x bar).x for any point x in C. The algorithm is globally and finitely convergent.