A Many-Sorted Variant of Japaridze's Polymodal Provability Logic

Abstract

We consider a many-sorted variant of Japaridze's polymodal provability logic $\mathsf{GLP}$. In this variant, which is denoted $\mathsf{GLP}^\ast$, propositional variables are assigned sorts $\alpha \leq \omega$, where variables of finite sort $n < \omega$ are interpreted as $\Pi_{n+1}$-sentences of the arithmetical hierarchy, while those of sort $\omega$ range over arbitrary ones. We prove that $\mathsf{GLP}^\ast$ is arithmetically complete with respect to this interpretation. Moreover, we relate $\mathsf{GLP}^\ast$ to its one-sorted counterpart $\mathsf{GLP}$ and prove that the former inherits some well-known properties of the latter, like Craig interpolation and PSpace decidability. We also study a positive variant of $\mathsf{GLP}^\ast$ which allows for an even richer arithmetical interpretation---variables are permitted to range over theories rather than single sentences. This interpretation in turn allows the introduction of a modality that corresponds to the full uniform reflection principle. We show that our positive variant of $\mathsf{GLP}^\ast$ is arithmetically complete.