MAX-PLUS-ALGEBRAIC PROBLEMS AND THE EXTENDED LINEAR COMPLEMENTARITY PROBLEM ― ALGORITHMIC ASPECTS

摘要

Many fundamental problems in the max-plus-algebraic system theory for discrete event systems ― among which the minimal state space realization problem ― can be solved using an Extended Linear Complementarity Problem (ELCP). We present some new, more efficient methods to solve the ELCP. We show that an ELCP with a bounded feasible set can be recast as a standard Linear Complementarity Problem (LCP). Our proof results in three possible numerical solution methods for a given ELCP: regular ELCP algorithms, mixed integer linear programming algorithms, and regular LCP algorithms. We also apply these three methods to a basic max-plus-algebraic benchmark problem.