Local Termination: theory and practice

The characterisation of termination using well-founded monotone algebras has been a milestone on the way to automated termination techniques, of which we have seen an extensive development over the past years. Both the semantic characterisation and most known termination methods are concerned with global termination, uniformly of all the terms of a term rewriting system (TRS). In this paper we consider local termination, of specific sets of terms within a given TRS. The principal goal of this paper is generalising the semantic characterisation of global termination to local termination. This is made possible by admitting the well-founded monotone algebras to be partial. We also extend our approach to local relative termination. The interest in local termination naturally arises in program verification, where one is probably interested only in sensible inputs, or just wants to characterise the set of inputs for which a program terminates. Local termination will be also be of interest when dealing with a specific class of terms within a TRS that is known to be non-terminating, such as combinatory logic (CL) or a TRS encoding recursive program schemes or Turing machines. We show how some of the well-known techniques for proving global termination, such as stepwise removal of rewrite rules and semantic labelling, can be adapted to the local case. We also describe transformations reducing local to global termination problems. The resulting techniques for proving local termination have in some cases already been automated. One of our applications concerns the characterisation of the terminating S-terms in CL as regular language. Previously this language had already been found via a tedious analysis of the reduction behaviour of S-terms. These findings have now been vindicated by a fully automated and verified proof.

💡 Research Summary

The paper “Local Termination: Theory and Practice” extends the well‑established semantic characterisation of termination—based on well‑founded monotone algebras—to the setting of local termination, i.e., termination of a selected set of terms rather than all terms of a term rewriting system (TRS). The authors achieve this by allowing the underlying algebras to be partial: the interpretation of function symbols may be undefined for some arguments, and the ordering relation need not be globally well‑founded. This flexibility enables a semantic framework that can capture termination properties of non‑terminating TRSs when restricted to sensible inputs, such as combinatory logic (CL) or encodings of Turing machines.

Key contributions are:

1. Partial Σ‑Algebras and Partial Models – The paper defines partial Σ‑algebras where each operation is a partial function, together with a binary relation ≻. A partial model requires that whenever the left‑hand side of a rule is defined, its interpretation strictly exceeds that of the right‑hand side. This asymmetry reflects the fact that rewriting may turn an undefined (non‑terminating) term into a defined (terminating) one, but not vice‑versa.

2. Monotone Partial Algebras – The authors introduce the notions of closed and monotone partial operations with respect to ≻. A monotone partial Σ‑algebra is a partial algebra whose operations satisfy these properties. When ≻ is well‑founded the algebra yields a complete characterisation of local termination: a TRS is locally terminating on a set T iff there exists a well‑founded monotone partial algebra that serves as a partial model for the TRS on T.

3. Local Relative Termination – The framework is extended to relative termination: given two TRSs R₁ and R₂, R₁ terminates relative to R₂ on a set T if a suitable extended monotone partial algebra can be built where rules of R₁ are strictly decreasing and rules of R₂ are non‑increasing.

4. Adaptation of Global Techniques – Classical global termination techniques are re‑engineered for the local setting:

   • Rule removal: rules may be eliminated only if they are weakly decreasing with respect to the partial model, ensuring that the reachable term set is unchanged.
   • Semantic labeling: by labeling symbols according to a partial model, the authors obtain a transformed TRS whose global termination implies the original system’s local termination on the target set.
   • Transformation to global termination: a systematic reduction of a local termination problem to a global one is presented, allowing the reuse of existing automated provers.

5. Automation via Regular Languages – When the set of normalising terms N of a TRS is regular (e.g., for orthogonal, non‑erasing systems), the authors construct a tree automaton (partial model) that recognises exactly N. This automaton can be used for semantic labeling, turning the local problem into a global one that is amenable to tools such as TTT2 and CeTA. The paper demonstrates this on the S‑combinator and δ‑combinator of CL, automatically deriving the regular language of terminating S‑terms and certifying termination with existing provers.

6. Application to Global Termination – The authors show that their local techniques can be turned around: a global termination problem can be encoded as a local termination problem on the set of right‑hand sides of forward closures, then transformed back to a global problem via the methods of Section 6. This approach solves a previously open benchmark from the 2008 termination competition using simple linear polynomial interpretations.

7. Partial Quasi‑Models Combined with Monotone Algebras – By merging deterministic tree automata (quasi‑models) equipped with a state ordering ≥ with monotone algebras, the paper obtains partial monotone algebras that are both closed under rewriting and support a well‑founded ordering. An implementation of this combined framework is reported, confirming its practical viability.

Overall, the work provides a complete semantic characterisation of local termination, supplies transformations that bridge local and global termination, and demonstrates that many existing global techniques can be reused after suitable adaptation. The authors validate their theory with substantial automation, handling classic non‑terminating systems (CL, Turing machines) and solving open termination benchmarks. This establishes local termination as a robust, automatable analysis tool, extending the reach of termination methods beyond the traditional global setting.