Nonuniform Deterministic Finite Automata over Finite Algebraic Structures

Abstract

Nonuniform deterministic finite automata (NUDFA) over monoids were invented by Barrington in to study boundaries of nonuniform constant-memory computation. Later, results on these automata helped to identify interesting classes of groups for which equation satisfiability problem (PolSat) is solvable in (probabilistic) polynomial time. Based on these results, we present a full characterization of groups, for which the identity checking problem (called PolEqv) has a probabilistic polynomial-time algorithm. We also go beyond groups, and propose how to generalise the notion of NUDFA to arbitrary finite algebraic structures. We study satisfiability of these automata in this more general setting. As a consequence, we present a full description of finite algebras from congruence modular varieties for which testing circuit equivalence CEqv can be solved by a probabilistic polynomial-time procedure. In our proofs we use two computational complexity assumptions: randomized Expotential Time Hypothesis and Constant Degree Hypothesis.