An Infeasible Predictor-Corrector Algorithm with a New Neighborhood for Semidefinite Programming

摘要

In this paper, we present a new infeasible primal-dual MTY type predictor-corrector interior point method (IPM) for Semidefinite Programming (SDP). The proposed algorithm is based on a new neighborhood, which is an even wider neighborhood than a given F-norm neighborhood. The modified algorithm can be applied when the data matrices are block and diagonal. We verify its validity by numerical experiments for a given convergence tolerance and an initial residual.