Logic as a distributive law

TL;DR

This paper develops a categorical framework using monads and distributive laws to derive a spatial-behavioral type system from a formal calculus, connecting logic, type theory, and process calculi.

Contribution

It introduces a novel algorithm that constructs a type system from monadic presentations of calculi and logic, unifying Curry-Howard and realizability interpretations.

Findings

Derived a natural interpretation of formulae via monads and distributive laws.

Implemented a modal operator and related it to lambda calculus arrow types.

Encoded properties like confinement and liveness in a higher-order pi-calculus variant.

Abstract

We present an algorithm for deriving a spatial-behavioral type system from a formal presentation of a computational calculus. Given a 2-monad Calc: Catv$\to$ Cat for the free calculus on a category of terms and rewrites and a 2-monad BoolAlg for the free Boolean algebra on a category, we get a 2-monad Form = BoolAlg + Calc for the free category of formulae and proofs. We also get the 2-monad BoolAlg $\circ$ Calc for subsets of terms. The interpretation of formulae is a natural transformation $\interp{-}$: Form $\Rightarrow$ BoolAlg $\circ$ Calc defined by the units and multiplications of the monads and a distributive law transformation $\delta$: Calc $\circ$ BoolAlg $\Rightarrow$ BoolAlg $\circ$ Calc. This interpretation is consistent both with the Curry-Howard isomorphism and with realizability. We give an implementation of the "possibly" modal operator parametrized by a two-hole term context and show that, surprisingly, the arrow type constructor in the $\lambda$-calculus is a specific case. We also exhibit nontrivial formulae encoding confinement and liveness properties for a reflective higher-order variant of the $\pi$-calculus.