Universality of Confluent, Self-Loop Deterministic Partially Ordered NFAs is Hard

TL;DR

This paper investigates the computational complexity of the universality problem for a specific class of automata called confluent, self-loop deterministic poNFAs, revealing it remains computationally hard even under restrictions.

Contribution

The paper proves that universality for confluent, self-loop deterministic poNFAs is PSpace-complete with a growing alphabet, establishing a lower bound for related problems.

Findings

Universality is PSpace-complete for polynomially growing alphabets.

Complexity drops to coNP-complete for fixed alphabets.

Universality remains as hard as for general NFAs despite restrictions.

Abstract

An automaton is partially ordered if the only cycles in its transition diagram are self-loops. The expressivity of partially ordered NFAs (poNFAs) can be characterized by the Straubing-Th\'erien hierarchy. Level 3/2 is recognized by poNFAs, level 1 by confluent, self-loop deterministic poNFAs as well as by confluent poDFAs, and level 1/2 by saturated poNFAs. We study the universality problem for confluent, self-loop deterministic poNFAs. It asks whether an automaton accepts all words over its alphabet. Universality for both NFAs and poNFAs is a PSpace-complete problem. For confluent, self-loop deterministic poNFAs, the complexity drops to coNP-complete if the alphabet is fixed but is open if the alphabet may grow. We solve this problem by showing that it is PSpace-complete if the alphabet may grow polynomially. Consequently, our result provides a lower-bound complexity for some other problems, including inclusion, equivalence, and $k$-piecewise testability. Since universality for saturated poNFAs is easy, confluent, self-loop deterministic poNFAs are the simplest and natural kind of NFAs characterizing a well-known class of languages, for which deciding universality is as difficult as for general NFAs.