Contents
Idea
This is the generalization of the notion of ends from category theory to (∞,1)-category theory.
Definition
End of a space-valued functor
Given an (∞,1)-functor to ∞Groupoids from the product of a locally small (∞,1)-category with its opposite (∞,1)-category, we define its end to be the (∞,1)-categorical hom-space in the (∞,1)-category of (∞,1)-functors from the hom-functor of to :
Ends and coends as a weighted (co)limit
More generally, given an (∞,1)-functor to ∞Groupoids to any locally small (∞,1)-category , we define its (co)end, when it exists, as the (co)representing object of
When , this definition agrees with the previous.
As (co)limits over the twisted arrow category
Let be a functor between (∞,1)-categories. We can express the (co)end as (co)limits over the twisted arrow (∞,1)-category by
The second of these, for example, follows from the fact the projection is the left fibration classified by , and so we can write the hom-functor as the left Kan extension and take an adjoint transpose.
Properties
Proposition (Fubini)
Given a functor , if the relevant ends/coends exist, we have natural equivalences
Proof
For ends, using the fact preserves products we can compute
The same argument applies for coends.
Proposition
Given functors between two (∞,1)-categories, the space of natural transformations between them is given by the end
This is proposition 5.1 of Gepner-Haugseng-Nikolaus 15.
References