Graded Monads in the Semantics of Nominal Automata

Nominal automata models serve as a formalism for data languages, and in fact often relate closely to classical register models. The paradigm of name allocation in nominal automata helps alleviate the pervasive computational hardness of register models in a tradeoff between expressiveness and computational tractability. For instance, regular nondeterministic nominal automata (RNNAs) correspond, under their local freshness semantics, to a form of lossy register automata. Unlike the full register automaton model, RNNAs allow for inclusion checking in elementary complexity. The semantic framework of graded monads provides a unified algebraic treatment of spectra of behavioural equivalences in the setting of universal coalgebra. In the present work, we extend the associated notion of graded algebraic theory to the nominal setting, and develop a nominal version of the notion of graded behavioural equivalence game. In the arising framework of graded nominal algebra, we conduct an extended case study, giving an algebraic theory capturing the local freshness semantics of RNNAs and the related nominal transition systems. Moreover, we instantiate the general behavioural equivalence game to this setting.

💡 Research Summary

The paper develops a unified categorical and algebraic framework for regular nondeterministic nominal automata (RNNAs), focusing on their local freshness semantics. After reviewing the basics of nominal sets—finite support, equivariant actions, abstraction, and powersets—the authors introduce bar strings, which are words over an extended alphabet that include bound names marked by a preceding bar. They define α‑equivalence on bar strings and two language extraction operators: N, which yields data languages from clean bar strings (global freshness), and D, which simply removes bars (local freshness). While N enforces that each freshly read name is distinct from all previously seen names, D only requires distinctness from names currently stored in registers, making it more permissive but also more complex to analyze.

RNNAs are presented as orbit‑finite state sets with α‑invariant transition relations, an initial state, and a set of final states. Their semantics is given by interpreting runs over bar strings and then applying either N or D to obtain ordinary data languages. The authors note that inclusion checking for RNNAs under global freshness is elementary (exponential‑space), whereas under local freshness it is more challenging.

The central technical contribution is the extension of graded monads to the category of nominal sets, yielding “graded nominal algebra.” By assigning depths to operations—depth 1 to transition labels and nondeterministic choice, depth 0 to name restriction (⟨a⟩·)—they construct a graded algebraic theory that captures the local freshness behavior of RNNAs. Crucially, the inclusion of name restriction as a depth‑0 operation resolves α‑renaming obstacles that arise when trying to model local freshness categorically. They prove that any depth‑1 graded nominal theory induces a depth‑1 graded monad, satisfying the prerequisites for the general results of graded semantics (logical characterizations, game characterizations, etc.).

Building on this, the paper adapts the graded behavioural equivalence game (a Spoiler‑Duplicator game) to the nominal setting. In each round, Spoiler selects a state and a name environment, and Duplicator must respond with a matching state while respecting the graded depth constraints. For global freshness the game reduces to the standard trace‑equivalence game, but for local freshness the presence of name restriction forces additional rules to handle α‑renamings. The authors verify that the winning region of Duplicator coincides with the graded behavioural equivalence defined by the graded monad, thereby providing an operational method to decide trace equivalence of RNNA states under both semantics.

Finally, the paper discusses how the graded framework yields logical characterizations and up‑to techniques for RNNA equivalence checking, offering a more tractable alternative to the notoriously hard inclusion problem for classical register automata. By integrating category theory, universal algebra, and game semantics, the work presents a powerful new toolbox for reasoning about nominal automata, especially those employing name allocation and local freshness constraints.