Numerical method for solving the continuous-time linear programming problems with time-dependent matrices and piecewise continuous functions

Abstract

The numerical method is proposed in this paper to solve a general class of continuous-time linear programming problems in which the functions appeared in the coefficients and the time-dependent matrices are assumed to be piecewise continuous. In order to make sure that all the subintervals of time interval will not contain the discontinuities of the involved functions, a methodology for not equally partitioning the time interval is proposed. The main issue of this paper is to obtain an analytic formula of the error bound, where the strong duality theorem for the primal and dual pair of continuous-time linear programming problems with time-dependent matrices and piecewise continuous functions is a by-product. We shall propose two kinds of computational procedure to evaluate the error bounds. One needs to solve the dual problem of the discretized linear programming problem, and another one does not need to solve the dual problem. The detailed differences between these two computational procedures will be also presented. Finally we present a numerical example to demonstrate the usefulness of the numerical method.