PRIMAL-DUAL SOLUTION FOR THE SEMIDEFINITEPROGRAMMING PROBLEMS USING NEURALNETWORKS

In this paper we solve a wide rang of Semide nite Pro-gramming Problem (SDP) by using Recurrent Neural Networks(RNNs). SDP is an important numerical tools for analysis andsynthesis in systems and control theory. First we reformulate theproblem to a linear programming problem, second we reformu-late it to a rst order system of ordinary di erential equations.Then the RNN model is proposed to compute related primal anddual solutions simultaneously. Illustrative examples are includedto demonstrate the validity and applicability of the technique.