An Outer Approximation Algorithm for Solving General Convex Programs

Abstract

A new outer approximation algorithm is proposed for solving general convex programs. A remarkable advantage of the algorithm over existing outer approximation methods is that the approximation of the constraint set is not cumulative. That is, the algorithm solves at each iteration a quadratic program whose constraints depend only on the current estimate of an optimal solution. Convergence of the algorithm is proved and possible applications are discussed.