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
vector bundle, 2-vector bundle, (∞,1)-vector bundle
real, complex/holomorphic, quaternionic
In type theory, a display map [Taylor 1983-7, §4.3.2] is a morphism in a category which represents a dependent type under categorical semantics of dependent type theory.
That is, represents (interprets) a type dependent on a variable of type , usually written syntactically as . The intended intuition is that is the fiber of the map over .
More specifically, the syntactic category of a type theory is naturally equipped with a class of maps called “display maps” which come from its type dependencies, while models for such a theory can be studied in any other category equipped with a suitable class of maps called “display maps.” By the usual adjunction, such models are then equivalent to functors out of the syntactic category which preserve the relevant structure, including the display maps.
Since the most common dependent type theories include dependent product types, and any display map in a category that represents semantics for such a theory must be exponentiable, the term display map is sometimes used to mean any exponentiable morphism.
Sometimes all maps are display maps. This happens particularly in extensional type theory, such as versions of the internal logic of categories that include dependent types.
In the context of homotopy type theory display maps are fibrations, for instance in a type-theoretic model category.
For a category, a class of morphisms of is called a class of displays if
It follows automatically that the full subcategory on of the arrow category is replete; in other words, the property of -morphisms given by membership in respects the principle of equivalence.
The category with displays is called well rooted if it has a terminal object and all the morphisms to are display maps.
It follows that binary products exist and their projections are display maps.
Often (which simplifies matters) is closed under composition:
Every identity morphism (and hence every isomorphism) belongs to .
The composite of a composable pair of elements of belongs to .
An interpretation or model of a category with class of displays in another category with displays is a functor
which preserves display maps and their pullbacks.
This appears as (Taylor 1999, def. 8.3.2).
Examples of categories with displays, def. , include
One extreme: any locally cartesian category with the class of all morphisms. (This is well rooted —if has a terminal object— and closed under composition.)
Another extreme: any category with the empty class of no morphisms.
Modifying (2) for an extreme case closed under composition: any category with the class of isomorphisms.
Modifying (2) for an extreme well-rooted case: A category with finite products and the class of product projections . (This is also closed under composition.)
A locally cartesian category (such as a topos) equipped with its class of monomorphisms. (This is closed under composition.)
A category equipped with a singleton pretopology. (This is closed under composition.)
(…)
See also Taylor 1999, Exp. 8.3.6.
The notion is due to
Paul Taylor, §4.3.2 in: Recursive Domains, Indexed Category Theory and Polymorphism, Cambridge (1983-7) [pdf]
Paul Taylor, Section 8.3 in: Practical Foundations of Mathematics, Cambridge University Press (1999) [webpage]
Last revised on September 12, 2024 at 10:03:48. See the history of this page for a list of all contributions to it.