A filter algorithm for nonlinear semidefinite programming (Q5962406)

Motivated by the local convergence properties obtained for the sequential semidefinite programming algorithm and the global properties obtained for the filter sequential quadratic programming, the aim of the paper is to exploit this recent filter methodology in order to propose a globally convergent algorithm for solving the NLSDP (nonlinear semidefinite programming problem). First the authors present in details the filter semidefinite programming algorithm for solving NLSDP. Then they prove the global convergence of this algorithm and finally they report some numerical tests on selected problems from the COMPleib library.