We present a primal-dual interior point method for nonlinearoptimization that relies on a line search filter strategy to allowconvergence from poor starting points. The filter technique hasalready been adapted to interior point methods in different ways.Our filter relies on three components. Each entry in the filterincludes the feasibility measure, the centrality measure and thebarrier objective function value as the optimality measure.Numerical experiments carried out with a set of engineering designproblems show that our filter approach is effective in reaching thesolution. A comparison with other well-known methods is alsoreported.
展开▼