The Complexity of Enriched Mu-Calculi

The fully enriched μ-calculus is the extension of the propositional μ-calculus with inverse programs, graded modalities, and nominals. While satisfiability in several expressive fragments of the fully enriched μ-calculus is known to be decidable and ExpTime-complete, it has recently been proved that the full calculus is undecidable. In this paper, we study the fragments of the fully enriched μ-calculus that are obtained by dropping at least one of the additional constructs. We show that, in all fragments obtained in this way, satisfiability is decidable and ExpTime-complete. Thus, we identify a family of decidable logics that are maximal (and incomparable) in expressive power. Our results are obtained by introducing two new automata models, showing that their emptiness problems are ExpTime-complete, and then reducing satisfiability in the relevant logics to these problems. The automata models we introduce are two-way graded alternating parity automata over infinite trees (2GAPTs) and fully enriched automata (FEAs) over infinite forests. The former are a common generalization of two incomparable automata models from the literature. The latter extend alternating automata in a similar way as the fully enriched μ-calculus extends the standard μ-calculus.

💡 Research Summary

The paper investigates the decidability frontier of the fully enriched μ‑calculus, a powerful extension of the propositional μ‑calculus that adds three orthogonal features: inverse programs (allowing backward navigation in transition systems), graded modalities (quantifying over at least/at most k successors), and nominals (state‑naming constants). While recent work has shown that the combination of all three features yields an undecidable satisfiability problem, the authors ask whether dropping any single feature restores decidability, and if so, what the exact complexity is.

To answer this, they define three maximal fragments, each obtained by omitting one of the three extensions: (i) the graded‑plus‑nominal fragment (no inverse programs), (ii) the inverse‑plus‑nominal fragment (no graded modalities), and (iii) the inverse‑plus‑graded fragment (no nominals). These fragments are incomparable in expressive power—none is a subset of another—yet each is as expressive as possible while still lacking one of the three constructs.

The technical core of the paper is the introduction of two novel automata models that capture precisely the behaviours needed for these fragments. The first model, two‑way graded alternating parity tree automata (2GAPTs), operates on infinite trees and combines two‑way navigation (both parent‑to‑child and child‑to‑parent moves) with graded counting. 2GAPTs subsume both the previously studied two‑way alternating parity automata and graded alternating automata, providing a uniform framework for fragments that involve inverse programs and/or graded modalities.

The second model, fully enriched automata (FEAs), works over infinite forests rather than single trees. FEAs extend alternating parity automata with the ability to handle nominals (by labeling forest roots) and inverse programs (by allowing transitions that move against the usual parent‑child direction). A key technical contribution is a “forest compression” technique that reduces an infinite forest with nominal annotations to a finite-state representation without losing the essential acceptance conditions.

For each automaton class the authors prove that the emptiness problem is ExpTime‑complete. The upper bound for 2GAPTs is obtained by constructing a parity game whose positions encode both the current state of the automaton and the current grade counters; a winning strategy corresponds to an accepting run. The graded counters are handled by a ranking function that merges the parity priorities with the quantitative constraints, yielding an exponential‑time algorithm. For FEAs, the forest compression yields a finite alternating parity automaton whose emptiness can be decided by the classic algorithm for alternating parity tree automata, again within exponential time. The lower bound follows from the known ExpTime‑hardness of the standard μ‑calculus satisfiability problem, which embeds directly into each fragment.

Having established the automata emptiness complexities, the authors give a linear‑time translation from any formula of the three fragments into an equivalent automaton of the appropriate class. The translation respects the semantics of inverse programs, graded modalities, and nominals, ensuring that a formula is satisfiable iff the resulting automaton is non‑empty. Consequently, satisfiability for each fragment is decidable and ExpTime‑complete.

The paper therefore identifies a family of maximal, mutually incomparable, yet decidable extensions of the μ‑calculus. It delineates precisely which combination of features pushes the logic over the decidability border and shows that any proper subset of the three extensions remains tractable at the optimal exponential time bound. The results have immediate implications for the design of specification languages and model‑checking tools that need expressive navigation and counting while retaining algorithmic feasibility. The authors conclude with suggestions for future work, including the exploration of additional operators (e.g., temporal modalities), optimization of the automata constructions for practical implementations, and the investigation of the trade‑offs between expressiveness and succinctness within these maximal decidable fragments.