An SMT-Based Perfect Sampling Algorithm for Stochastic Petri Nets

Abstract

This paper proposes a perfect sampling algorithm for stochastic Petri nets (SPN). The perfect sampling is a technique to draw samples exactly following the stationary distribution. The paper develops the enveloped perfect sampling algorithm (EPSA) for SPN. The main idea behind our approach is to formulate the mathematical programming to obtain lower and upper bounds of system states which are required by EPSA, and the problem is solved by using SMT (satisfiability modulo theories) solver. The presented SMT-based EPSA for SPN expands the applicability of perfect sampling algorithm for SPNs.