Monoidal computer I: Basic computability by string diagrams

We present a new model of computation, described in terms of monoidal categories. It conforms the Church-Turing Thesis, and captures the same computable functions as the standard models. It provides a succinct categorical interface to most of them, free of their diverse implementation details, using the ideas and structures that in the meantime emerged from research in semantics of computation and programming. The salient feature of the language of monoidal categories is that it is supported by a sound and complete graphical formalism, string diagrams, which provide a concrete and intuitive interface for abstract reasoning about computation. The original motivation and the ultimate goal of this effort is to provide a convenient high level programming language for a theory of computational resources, such as one-way functions, and trapdoor functions, by adopting the methods for hiding the low level implementation details that emerged from practice. In the present paper, we make a first step towards this ambitious goal, and sketch a path to reach it. This path is pursued in three sequel papers, that are in preparation.

💡 Research Summary

The paper introduces a novel model of computation called a “monoidal computer,” built on the mathematical framework of strong monoidal categories. The authors begin by recalling that classical models—Turing machines, λ‑calculus, and recursive function theory—are equivalent in the sense of the Church‑Turing thesis: they all capture the same class of computable (partial) functions. Their contribution is to recast this equivalence categorically, using a single‑object monoidal category C whose sole object A represents the entire data universe, while morphisms A → A represent programs or computational steps.

In this setting the monoidal tensor ⊗ models parallel composition, and categorical composition (∘) models sequential execution. Two distinguished morphisms are introduced: a comultiplication (δ : A → A⊗A) that copies data, and a counit (ε : A → I) that discards it. Together they give A a comonoid structure, satisfying the usual co‑associativity, co‑unit, and commutativity equations. A universal morphism u : I → A embeds basic data (e.g., natural numbers, strings, booleans) into the object A, thereby allowing the encoding of any conventional datatype inside the categorical world.

The graphical language of string diagrams is then employed as a concrete syntax for reasoning about morphisms. Objects become wires, morphisms become boxes, the tensor product is depicted by parallel wires, and composition by connecting boxes end‑to‑end. The copying and discarding morphisms appear as wire splits and merges, respectively. The authors stress that diagrammatic manipulation—rewiring, sliding, and applying the monoidal axioms—constitutes a sound and complete proof system: two morphisms are equal in the category if and only if their diagrams can be transformed into one another using a finite set of local rewrite rules (swap, associativity, unit insertion/removal). This “sound‑and‑complete graphical calculus” provides an intuitive, implementation‑agnostic interface for abstract computation.

To capture computability, the paper defines a universal program construction: for any partial recursive function f : ℕ ⇀ ℕ, there exists a morphism φ_f : A → A built from a finite diagram of the basic generators (δ, ε, u, and composition/tensor) that behaves exactly like f when the input wire is interpreted via the encoding u. Conversely, every morphism A → A corresponds to a partial recursive function under the same interpretation. This bidirectional correspondence establishes the completeness of the monoidal computer with respect to the Church‑Turing thesis.

Two central theorems are proved. The Completeness Theorem shows that the class of morphisms in the monoidal computer coincides with the class of Turing‑computable (partial) functions. The Soundness (or Full‑Abstraction) Theorem demonstrates that categorical equality of morphisms is exactly captured by diagrammatic equivalence, guaranteeing that the graphical language is not merely a heuristic but a rigorous reasoning tool.

Beyond the foundational results, the authors outline a longer‑term research agenda: leveraging the monoidal computer as a high‑level language for computational resource theory, particularly for cryptographic primitives such as one‑way functions and trapdoor permutations. By abstracting away low‑level implementation details in the same way that modern programming languages hide machine code, the monoidal computer aims to enable reasoning about resource constraints (time, space, algebraic structure) directly at the diagrammatic level. The present paper constitutes the first step—establishing the basic computational power and the diagrammatic calculus—while three sequel papers (in preparation) will develop resource‑sensitive extensions, cost models, and concrete language designs.

In summary, the work provides a categorical reformulation of classical computability, equips it with a rigorous and intuitive graphical syntax, proves that this framework is computationally equivalent to all standard models, and sets the stage for a new high‑level, resource‑aware programming paradigm grounded in monoidal category theory.