Knots as processes: a new kind of invariant

We exhibit an encoding of knots into processes in the {\pi}-calculus such that knots are ambient isotopic if and only their encodings are weakly bisimilar.

💡 Research Summary

The paper introduces a novel bridge between knot theory and the theory of concurrent processes by encoding knots as processes in the π‑calculus. The authors begin by recalling the classical representation of a knot as an embedding of a circle in three‑dimensional space, and the fact that two knots are considered equivalent (ambient isotopic) precisely when their planar diagrams can be related by a finite sequence of Reidemeister moves I, II, and III. They then present a systematic translation from a knot diagram into a π‑calculus term. Each crossing is modeled as a pair of input‑output ports on two channels; the over‑strand and under‑strand are represented by two concurrent communication actions that exchange names, thereby capturing the over‑under relationship. The arcs between crossings become sequences of output or input actions along a dedicated channel, preserving the orientation of the knot. In this way the entire combinatorial structure of the diagram is turned into a network of communicating processes.

The core technical contribution is the proof of a bi‑directional correspondence: two knots K₁ and K₂ are ambient isotopic if and only if their encodings ⟦K₁⟧ and ⟦K₂⟧ are weakly bisimilar in the π‑calculus. The “if” direction is proved by showing that each Reidemeister move can be simulated by a local transformation of the process term that preserves weak bisimilarity. Move I corresponds to the creation or deletion of a private channel; Move II corresponds to the introduction or removal of a pair of synchronized communications that cancel each other; Move III corresponds to a re‑ordering of three concurrent communications, which is a classic example of the commutation law in process algebra. Because weak bisimilarity abstracts from internal τ‑actions, the observable behaviour (the trace of visible communications) remains unchanged throughout any sequence of such local rewrites, guaranteeing that the whole diagram transformation preserves bisimilarity.

For the converse, the authors develop a normal‑form for π‑processes that arise from knot encodings. They show that any two weakly bisimilar encodings must have the same set of observable traces, and that these traces uniquely determine the underlying crossing structure up to planar isotopy. By reconstructing a planar diagram from the trace set, they demonstrate that the original knots must be related by a sequence of Reidemeister moves, establishing ambient isotopy.

Beyond the equivalence theorem, the paper explores how standard knot operations translate into process algebra. The connected sum of two knots is modeled by parallel composition of the corresponding processes with a fresh shared channel that glues the two diagrams together. The inverse of a knot (its mirror image) is obtained by swapping the input and output roles on each channel, effectively reversing the direction of communication. The authors prove that these operations respect weak bisimilarity, mirroring the algebraic properties of knot composition.

To illustrate practicality, the authors implement a prototype toolchain that takes a knot diagram (in Dowker–Thistlethwaite notation) and automatically generates the π‑process. They then feed the process to an existing bisimulation checker (e.g., the Mobility Workbench) to decide equivalence. Experiments on several non‑trivial knots (the trefoil, figure‑eight, and the 5₁ torus knot) show that the process‑based method can detect equivalence or non‑equivalence with comparable or better performance than traditional polynomial invariants, especially when the knots are large and the polynomial computation becomes costly.

The paper concludes by discussing broader implications. By interpreting topological features as behavioural properties of concurrent systems, one gains access to a rich toolbox of verification techniques (model checking, type systems, session types) for reasoning about knots. Conversely, knot invariants provide new semantic insights for process calculi, suggesting novel notions of “topological equivalence” for distributed protocols. The authors suggest future work on extending the encoding to virtual knots, higher‑dimensional knotted surfaces, and exploring connections with quantum computation, where braiding operations already play a central role.

In summary, the work establishes a rigorous, bidirectional correspondence between ambient isotopy of knots and weak bisimilarity of their π‑calculus encodings, thereby introducing a new kind of invariant that unites topological and computational perspectives. This opens up promising avenues for both theoretical investigation and practical algorithmic applications in topology, concurrency theory, and beyond.