An improved numerical method for solving a class of continuous-time linear programming problems

Abstract

In this paper we consider a class of infinitedimensional linear programming problems, called continuoustime linear programming problems (CLP ). In our previous paper [16], we presented a discrete approximation method for solving (CLP ) and established the estimation for error bounds of searched numerical solutions. However, the computation times involved can be very large as a result of solving a very large finite-dimensional linear program. We will focus on a special class of (CLP ) and provide improved methods to overcome the computational bottleneck.