On Coalgebraic Product Constructions for Markov Chains and Automata
By: Mayuko Kori, Kazuki Watanabe
Potential Business Impact:
Finds errors in computer programs faster.
Verifying traces of systems is a central topic in formal verification. We study model checking of Markov chains (MCs) against temporal properties represented as (finite) automata. For instance, given an MC and a deterministic finite automaton (DFA), a simple but practically useful model checking problem asks for the probability of traces on the MC that are accepted by the DFA. A standard approach to solving this problem constructs a product MC of the given MC and DFA, reducing the task to a simple reachability probability problem on the resulting product MC. In this paper, on top of our recent development of coalgebraic framework, we first present a no-go theorem for product constructions, showing a case when we cannot do product constructions for model checking. Specifically, we show that there are no coalgebraic product MCs of MCs and nondeterministic finite automata for computing the probability of the accepting traces. This no-go theorem is established via a characterisation of natural transformations between certain functors that determine the type of branching, including nondeterministic or probabilistic branching. Second, we present a coalgebraic product construction of MCs and multiset finite automata (MFAs) as a new instance within our framework. This construction addresses a model checking problem that asks for the expected number of accepting runs on MFAs over traces of MCs. The problem is reduced to solving linear equations, which is solvable in polynomial-time under a reasonable assumption that ensures the finiteness of the solution.
Similar Papers
DTMC Model Checking by Path Abstraction Revisited (extended version)
Formal Languages and Automata Theory
Helps computers predict if a goal is reachable.
Initial Algebra Correspondence under Reachability Conditions
Logic in Computer Science
Makes different computer programs work the same.
Generalised Arc Consistency via the Synchronised Product of Finite Automata wrt a Constraint
Formal Languages and Automata Theory
Makes computers solve hard problems faster.