Order algebras: a quantitative model of interaction

A quantitative model of concurrent interaction is introduced. The basic objects are linear combinations of partial order relations, acted upon by a group of permutations that represents potential non-determinism in synchronisation. This algebraic structure is shown to provide faithful interpretations of finitary process algebras, for an extension of the standard notion of testing semantics, leading to a model that is both denotational (in the sense that the internal workings of processes are ignored) and non-interleaving. Constructions on algebras and their subspaces enjoy a good structure that make them (nearly) a model of differential linear logic, showing that the underlying approach to the representation of non-determinism as linear combinations is the same.

💡 Research Summary

The paper introduces order algebras, a quantitative denotational model for concurrent interaction that unifies non‑interleaving semantics with a linear‑algebraic treatment of nondeterminism. The core idea is to represent the observable behaviour of a process as a linear combination of partial orders (posets) over events. Each partial order captures a possible causal structure (the “happens‑before” relation) of an execution, while the accompanying real coefficients quantify how strongly or frequently that structure occurs.

To model the inherent nondeterminism of synchronisation, the authors let a permutation group act on these linear combinations. A permutation corresponds to a renaming of event identifiers, which in turn represents the different ways a nondeterministic choice can be resolved. Two linear combinations that lie in the same orbit of the group are considered equivalent, thereby collapsing syntactically distinct but behaviourally identical executions. This group action provides a clean algebraic abstraction of nondeterministic branching without resorting to explicit probability distributions.

The authors extend the classic testing semantics used in process algebra. In the traditional setting, a test is a process that interacts with the system under observation, and the outcome is simply success or failure. Here, a test is itself a linear combination of partial orders, and the interaction between a system and a test is defined by a synchronisation product (denoted ★) that merges the two posets while respecting causality and the permutation action. The result of this product is again a linear combination, whose total coefficient (or a suitable functional thereof) yields a quantitative observation. Two processes are deemed equivalent if they produce identical quantitative observations for all tests. This quantitative testing refines standard observational equivalence: it distinguishes processes that differ only in the distribution of their causal structures, even when those differences are invisible to a purely Boolean test.

A major contribution is the faithful interpretation of finitary process algebras (such as CCS, CSP, and the π‑calculus) within order algebras. The usual operators—choice, parallel composition, restriction, and recursion—are mapped to algebraic operations on linear combinations: choice becomes vector addition, parallel composition becomes a tensor‑like product that merges independent posets, restriction corresponds to quotienting by a subgroup of permutations, and recursion is interpreted via least fixed points in the complete vector space. The authors prove full abstraction: the denotational semantics in the order algebra coincides exactly with the operational semantics of the source process algebra under the quantitative testing equivalence.

Beyond process algebra, the paper reveals a deep connection between order algebras and Differential Linear Logic (DILL). In DILL, the exponential modality is equipped with a differential operator that captures infinitesimal resource duplication and erasure. The authors show that the linear structure of order algebras (addition and scalar multiplication) mirrors the additive fragment of linear logic, while the permutation group action implements the structural rules of contraction and weakening. Moreover, a derivative on a linear combination of posets can be defined by “adding” a minimal event or “removing” an existing one, exactly mirroring the differential operator of DILL. Consequently, order algebras constitute (almost) a model of DILL: they satisfy the required coherence conditions, and the differential structure provides a quantitative account of how small changes in a concurrent system’s causal structure affect its overall behaviour.

The paper also analyses the algebraic properties of order algebras. The space of all finite linear combinations of posets is a complete vector space; subspaces generated by particular synchronisation patterns (e.g., purely sequential or purely parallel behaviours) are themselves order algebras, closed under the synchronisation product and the permutation action. This closure ensures that the model is robust under the usual compositional constructions of process theory.

In summary, the authors deliver a novel mathematical framework that simultaneously:

1. Captures non‑interleaving causality through partial orders,
2. Quantifies nondeterministic branching via linear coefficients,
3. Provides a compositional semantics for standard process algebras,
4. Extends testing semantics to a quantitative setting, and
5. Links concurrency theory with differential linear logic, showing that the linear‑algebraic handling of nondeterminism is not an ad‑hoc invention but a manifestation of a well‑studied logical principle.

The work opens several promising research directions: extending the model to infinite or continuous state spaces, integrating probabilistic or stochastic information by allowing coefficients from richer semirings, applying the quantitative testing framework to performance analysis and model checking, and exploring the interplay with other non‑interleaving models such as event structures or configuration structures. By unifying algebra, logic, and concurrency, order algebras provide a powerful lens for understanding the quantitative aspects of interaction in modern distributed systems.