A primal-dual interior point method for nonlinear semidefinite programming (Q715092)

The authors propose a primal-dual interior point method for solving nonlinear semidefinite programming problems. Two types of iterations are used: outer iterations for finding a KKT point and inner iterations for computing an approximate barrier KKT point. The inner method uses a commutative class of Newton-like directions for generating line search directions. Then a new primal-dual merit function is proposed combining the primal barrier penalty function with the primal-dual barrier function. The global convergence of the resulting method is proven and numerical experiments are displayed to show the practical efficiency of the method.