Decision Diagrams for Petri Nets: A Comparison of Variable Ordering Algorithms

摘要

The efficacy of decision diagram techniques for state space generation is known to be heavily dependent on the variable order. Ordering can be done a-priori (static) or during the state space generation (dynamic). We focus our attention on static ordering techniques. Many static decision diagram variable ordering techniques exist, but it is hard to choose which method to use, since only fragmented performance information is available. In the work reported in this paper we used the models of the Model Checking Contest 2017 edition to conduct an extensive comparison of 18 different algorithms, in order to better understand their efficacy. Comparison is based on the size of the decision diagram of the reachable state space, which is generated using the Saturation method provided by the Meddly library.