A three-D filter line search method within an interior point framework

Abstract

Here we present a primal-dual interior point three-dimensional filter line search method for nonlinear programming. The three components of the filter aim to measure adequacy of feasibility, centrality and optimality of trial iterates. The algorithm also relies on a monotonic barrier parameter reduction and it includes a feasibility/centrality restoration phase. Numerical experiments with a set of well-known problems are carried out and a comparison with a previous implementation that differs on the optimality measure is presented.