On the Use of Matrix Factorization Techniques in Penalty Function Methods for Structured Linear Programs

An algorithm converging to an optimal solution of a linear program in a finite number of steps is proposed. The algorithm is based on the use of smooth penalty functions as well as on matrix factorization techniques. It consists of finding corner points of the piece-wise linear unconstrained minima trajectory. The application of the algorithm to dynamic linear programs and block-angular programs is described