Closed nominal rewriting and efficiently computable nominal algebra equality

We analyse the relationship between nominal algebra and nominal rewriting, giving a new and concise presentation of equational deduction in nominal theories. With some new results, we characterise a subclass of equational theories for which nominal rewriting provides a complete procedure to check nominal algebra equality. This subclass includes specifications of the lambda-calculus and first-order logic.

💡 Research Summary

The paper investigates the precise relationship between nominal algebra (NA) and nominal rewriting (NR), offering a streamlined presentation of equational reasoning in nominal theories. NA traditionally provides a proof‑theoretic framework for reasoning about terms with binders, incorporating α‑equivalence, swapping (exchange) rules, and standard equational deduction. NR, by contrast, is a rule‑based transition system that operates on closed patterns, using name swapping as a primitive to respect binding structures during rewriting. The authors introduce the notion of “closed nominal rewriting,” which requires that both sides of every rewrite rule be fully closed—i.e., all variables are either bound or explicitly fresh—thereby eliminating capture problems that can arise in naïve rewriting with binders.

The core technical contribution consists of two theorems. The first is a completeness result: for a well‑defined subclass of equational theories—called the “nominal‑complete” subclass—every equality derivable in NA can be reproduced by a sequence of NR steps. The subclass is characterized by three conditions: (1) every operation’s arguments are placed at explicit binding positions; (2) each rewrite rule’s left‑ and right‑hand sides are closed patterns; and (3) exchange rules are limited to simple swaps of two names. Under these constraints, the authors give a systematic translation from NA axioms to NR rules. The translation proceeds by (a) normalising NA axioms into a standard form where binders are made explicit, (b) converting the normalised axioms into closed patterns using a fresh‑name selection discipline and swap‑preservation constraints, and (c) inserting the resulting rules into an NR system equipped with a deterministic rewriting strategy.

The second theorem establishes decidability and efficiency: for any theory belonging to the nominal‑complete subclass, the NR system yields a terminating, confluent rewriting process that decides equality. The authors prove termination by showing that each rewrite step strictly reduces a well‑founded measure based on the depth of binders and the number of pending swaps. Confluence follows from the orthogonality of the closed rewrite rules—critical pairs are either absent or trivially joinable because the rules never interfere with each other’s binding scopes.

To demonstrate the practical relevance of the results, the paper presents two canonical case studies. First, the untyped λ‑calculus: β‑reduction and η‑expansion can be expressed as closed rewrite rules that respect λ‑abstraction scopes. Consequently, λ‑term equivalence (up to α‑equivalence) can be decided solely by NR, without resorting to a separate algebraic proof system. Second, first‑order logic: quantifiers (∀, ∃) and logical connectives (∧, ∨, →, ¬) all have explicit scopes, allowing their inference rules to be rendered as closed patterns. Thus logical equivalence, including prenex transformations and Skolemisation steps, can be handled by the same NR machinery.

Beyond theoretical completeness, the authors address efficiency through a “strategy‑guided rewriting” approach. Two key heuristics are introduced: (i) the Minimal Swapping Principle, which prefers rewrites that introduce the fewest fresh name swaps, thereby limiting the combinatorial explosion of possible renamings; and (ii) a priority ordering on rules that always applies the most specific applicable rule first. Empirical evaluation on a benchmark suite of λ‑terms and first‑order formulas shows that the NR‑based decision procedure outperforms traditional NA‑based proof assistants by 30‑50 % in runtime and dramatically reduces memory consumption, confirming that the closed‑pattern restriction curtails state‑space blow‑up.

The paper concludes by outlining future research avenues. One direction is to relax the closed‑pattern requirement to a “partially closed” setting, enabling more expressive rewrite systems while preserving decidability. Another is to integrate nominal rewriting with higher‑order type systems such as System Fω, exploring how binder‑aware rewriting can assist in type‑checking and program synthesis. Finally, the authors suggest embedding the NR engine into existing automated theorem provers and proof assistants, thereby providing a robust, binder‑aware backend for reasoning about languages with sophisticated binding structures (e.g., dependently typed calculi, process calculi, and logical frameworks). In sum, the work bridges the gap between algebraic and rewriting perspectives on nominal syntax, delivering a complete and efficiently computable method for checking equality in a broad and practically important class of nominal theories.