natural deduction metalanguage, practical foundations
type theory (dependent, intensional, observational type theory, homotopy type theory)
computational trinitarianism =
propositions as types +programs as proofs +relation type theory/category theory
There are in general two ways to define a dependent type theory:
One approach assumes a single separate type judgment and defines types via the single type judgment.
A second approach assumes an infinite number of universe levels and then defines an infinite number of type universes and, in the case for Coquand universes, type judgments indexed by , whereby types are then terms of the universes for Russell universes or identified with terms of the universes for Coquand universes via judgmental isomorphisms.
These two ways to define a dependent type theory leads to many different ways to make a dependent type theory impredicative. We assume throughout that there are some notion of dependent sum types, dependent product types, and identity types/path types.
For dependent type theories in the first sense, an impredicative dependent type theory is a dependent type theory with a type of all propositions defined via natural deduction inference rules. These dependent type theories are higher-order logics and are the internal logics of (the -categorical analogues of) elementary toposes. In general, (the -categorical analogues of) elementary toposes do not have any object classifiers.
For dependent type theories in the second sense, there are multiple notions of impredicative dependent type theory:
For impredicative dependent type theory in the sense of a dependent type theory for which the base universe satisfies impredicative polymorphism; these impredicative type theories are incommensurable with the notion of impredicative dependent type theory using a single type judgment, since one can prove that the type of all -small propositions is not -small.
For impredicative dependent type theory in the sense of a dependent type theory which has the propositional resizing axiom; these impredicative dependent type theories are stronger than the notion of impredicative dependent type theory using a single type judgment since these are the internal logics of -categories with an infinite number of inaccessible object classifiers.
Last revised on December 20, 2024 at 17:20:54. See the history of this page for a list of all contributions to it.