Binding bigraphs as symmetric monoidal closed theories

Milner’s bigraphs are a general framework for reasoning about distributed and concurrent programming languages. Notably, it has been designed to encompass both the pi-calculus and the Ambient calculus. This paper is only concerned with bigraphical syntax: given what we here call a bigraphical signature K, Milner constructs a (pre-) category of bigraphs BBig(K), whose main features are (1) the presence of relative pushouts (RPOs), which makes them well-behaved w.r.t. bisimulations, and that (2) the so-called structural equations become equalities. Examples of the latter include, e.g., in pi and Ambient, renaming of bound variables, associativity and commutativity of parallel composition, or scope extrusion for restricted names. Also, bigraphs follow a scoping discipline ensuring that, roughly, bound variables never escape their scope. Here, we reconstruct bigraphs using a standard categorical tool: symmetric monoidal closed (SMC) theories. Our theory enforces the same scoping discipline as bigraphs, as a direct property of SMC structure. Furthermore, it elucidates the slightly mysterious status of so-called links in bigraphs. Finally, our category is also considerably larger than the category of bigraphs, notably encompassing in the same framework terms and a flexible form of higher-order contexts.

💡 Research Summary

The paper revisits Milner’s theory of (binding) bigraphs and shows how it can be faithfully represented within the framework of symmetric monoidal closed (SMC) theories. A bigraphical signature K, consisting of a set of controls together with binding and free arities, is first translated into an SMC signature Σ_K over two sorts, t (terms) and v (names). The translation equips t with a commutative monoid structure (parallel composition ⊗, unit 0) and v with a binary operation c and unit w, mirroring the built‑in structural operations of bigraphs (parallel composition, name restriction, linking). For each control k the signature introduces a logical operation (v ⊗ B(k) ⊸ x) ⊗ v ⊗ F(k) → k, where x is I if k is atomic and t otherwise.

From any SMC signature one can construct the free SMC category S(Σ) whose objects are I‑MLL formulas and whose morphisms are proof‑nets (graphs of cells and wires) subject to the Danos‑Regnier correctness criterion. Wires connect negative ports (inputs) to positive ports (outputs); the bijection on ports of each sort guarantees that variables are bound exactly once, thereby enforcing the “no‑escape” scoping condition intrinsic to bigraphs. The authors further refine the construction to handle commutative monoid objects directly, collapsing the explicit m (and e) cells into the wiring structure.

Having built the free SMC category S(T_K) for the theory T_K (Σ_K together with the monoid equations), the authors define a functor T : Bbg(K) → S(T_K). On objects (interfaces) the functor is essentially injective: two interfaces with the same image are isomorphic. On morphisms it is faithful but not full; it deliberately excludes those morphisms that would violate the scoping discipline (e.g., a name escaping its scope). Nevertheless, for closed bigraphs—those without free names or sites—the functor is full and induces an isomorphism \