Formal P-Category Theory and Normalization by Evaluation in Rocq
By: David G. Berry, Marcelo P. Fiore
Potential Business Impact:
Makes computers prove math rules faster.
Traditional category theory is typically based on set-theoretic principles and ideas, which are often non-constructive. An alternative approach to formalizing category theory is to use E-category theory, where hom sets become setoids. Our work reconsiders a third approach - P-category theory - from \v{C}ubri\'c et al. (1998) emphasizing a computational standpoint. We formalize in Rocq a modest library of P-category theory - where homs become subsetoids - and apply it to formalizing algorithms for normalization by evaluation which are purely categorical but, surprisingly, do not use neutral and normal terms. \v{C}ubri\'c et al. (1998) establish only a soundness correctness property by categorical means; here, we extend their work by providing a categorical proof also for a strong completeness property. For this we formalize the full universal property of the free Cartesian-closed category, which is not known to have been performed before. We further formalize a novel universal property of unquotiented simply typed lambda-calculus syntax and apply this to a proof of correctness of a categorical normalization by evaluation algorithm. We pair the overall mathematical development with a formalization in the Rocq proof assistant, following the principle that the formalization exists for practical computation. Indeed, it permits extraction of synthesized normalization programs that compute (long) beta-eta-normal forms of simply typed lambda-terms together with a derivation of beta-eta-conversion.
Similar Papers
First-Order Modal Logic via Logical Categories
Logic in Computer Science
Makes computers understand logic better.
Realization of relational presheaves
Category Theory
Models computer actions for faster, clearer programs.
Universal Quantitative Abstraction: Categorical Duality and Logical Completeness for Probabilistic Systems
Logic in Computer Science
Makes AI learn better with guaranteed results.