The Bisimulation class The following class implements a basic normal bisimulation partition algorithms and derives a class for partitioning w.r.t. delayed bisimulation and weak bisimulation . All these algorithms are introduced in "Computing Maximal Weak and Other Bisimulations" from Alexandre Boulgakov et. al. (2016). The utilised normal bisimulation algorithm originates from the "change tracking algorithm" from Stefan Blom and Simona Orzan (see ref. of cited paper). Besides, the current paper uses topological sorted states to avoid computing saturated transitions explicitly when computing delayed and weak bisimulation
The TopoSort class perform a topological sort on states of a given automaton. if s1 is before s2 in the sort then there is no path from s2 to s1. The algorithm can be found in https://en.wikipedia.org/wiki/Topological_sorting under depth-first search, which is considered as first invented by R. Tarjan in 1976