Homotopical equivalence of combinatorial and categorical semantics of process algebra

It is possible to translate a modified version of K. Worytkiewicz’s combinatorial semantics of CCS (Milner’s Calculus of Communicating Systems) in terms of labelled precubical sets into a categorical semantics of CCS in terms of labelled flows using a geometric realization functor. It turns out that a satisfactory semantics in terms of flows requires to work directly in their homotopy category since such a semantics requires non-canonical choices for constructing cofibrant replacements, homotopy limits and homotopy colimits. No geometric information is lost since two precubical sets are isomorphic if and only if the associated flows are weakly equivalent. The interest of the categorical semantics is that combinatorics totally disappears. Last but not least, a part of the categorical semantics of CCS goes down to a pure homotopical semantics of CCS using A. Heller’s privileged weak limits and colimits. These results can be easily adapted to any other process algebra for any synchronization algebra.

💡 Research Summary

The paper establishes a precise homotopical correspondence between two semantic frameworks for Milner’s Calculus of Communicating Systems (CCS). The first framework is a combinatorial semantics based on labelled precubical sets, following a modified version of K. Worytkiewicz’s approach. In this setting, each n‑dimensional cube encodes a concurrent execution of n actions, and the faces of the cube represent partial orders among those actions. Labels attached to the cubes correspond directly to CCS actions, giving a concrete, low‑level description of process behaviour.

The second framework is a categorical semantics built from labelled flows. A flow is a topological model consisting of a set of states together with a space of execution paths (directed homotopies) between them. The categorical structure of flows admits a Quillen model structure, which provides cofibrant and fibrant replacements, homotopy limits, and homotopy colimits in a systematic way. Labels are again preserved as maps on the path spaces, so the observable actions of CCS are faithfully represented.

The central technical contribution is the construction of a geometric realization functor |‑| : Precub → Flow. This functor sends each n‑cube of a precubical set to an n‑dimensional globular cell (a directed n‑sphere) inside a flow, respecting source‑target boundaries and preserving labels. The authors prove two complementary theorems: (1) if two precubical sets X and Y are isomorphic, then their realizations |X| and |Y| are weakly equivalent as flows; (2) conversely, if |X| and |Y| are weakly equivalent, then X and Y are isomorphic as precubical sets. Hence the combinatorial isomorphism class of a precubical set coincides exactly with the homotopy class of its associated flow. No geometric information is lost in the passage from the combinatorial to the categorical world.

However, working directly in the category of flows still requires non‑canonical choices: cofibrant replacements, homotopy limits, and homotopy colimits are defined only up to weak equivalence. To obtain a truly canonical semantics, the authors move to the homotopy category Ho(Flow). In Ho(Flow) every object is already cofibrant, and the “privileged weak limits and colimits” introduced by A. Heller become available. Within this homotopical setting the semantics of CCS is defined purely in terms of weak equivalence classes of flows, eliminating any dependence on arbitrary model‑category constructions.

An important philosophical outcome is that the combinatorial machinery of precubical sets disappears once the homotopy category is adopted. The semantics of a CCS process can be read off from the weak equivalence class of its flow, without referring to the underlying cubes, faces, or boundary maps. This yields a cleaner, more abstract description of concurrency that is robust under homotopical deformation.

Finally, the authors argue that the whole construction is not specific to CCS. Any process algebra equipped with a synchronization algebra can be treated in exactly the same way: one defines a labelled precubical set semantics, builds the geometric realization into labelled flows, and then works in Ho(Flow) to obtain a homotopical semantics. The only ingredient that changes is the labeling rule that encodes the particular synchronization constraints of the algebra. Consequently, the paper provides a general blueprint for translating combinatorial models of concurrency into categorical, homotopical ones, and for doing so in a way that is canonical, lossless, and independent of the underlying combinatorial details.