3
$\begingroup$

I'm dealing with this construction, available whenever you have two categories $\cal C,D$ and a profunctor $\varphi\colon {\cal C}\leadsto{\cal D}$ between them.

Define a category ${\cal C}\boxtimes_\varphi{\cal D}$ (the gluing of $\cal C,D$) having objects the triples $(C,D,\alpha)$, with $\alpha\in \varphi(C,D)$, and morphisms the triples $(u,v,\gamma)$ such that the square

enter image description here

is commutative, if we denote as a dotted arrow the "fake" (or "generalized") arrow $\alpha\in \varphi(C,D)$ and $\beta\in \varphi(C', D')$; profunctors carries a natural left-right action of hom-sets, so that "$\beta \cdot u$ and $v\cdot \alpha$ have meaning. The square above "commutes" if these two elements coincide in the same $\gamma\in \varphi(C, D')$.

Now, I have shown these interesting (albeit almost tautological) properties:

  1. ${\cal C}\boxtimes_\varphi{\cal D}$ is isomorphic to the category of elements of $\varphi$;
  2. There is an isomorphism $$ {\bf Cat}(\mathcal A, {\cal C}\boxtimes_\varphi{\cal D})\cong {\bf Cat}(\mathcal A, \mathcal C)\boxtimes_{\Phi}{\bf Cat}(\mathcal A, \mathcal D) $$ where $\Phi = {\bf Cat}(\mathcal A, \varphi)$ is the profunctor $(F, G)\mapsto \int_{A\in\cal A}\varphi(FA, GA)$, so that there is the suggestive formula $$\textstyle {\bf Cat}(\mathcal A, \text{Elts }\varphi)\cong \text{Elts }{\bf Cat}(\mathcal A, \varphi) $$
  3. If $\varphi$ is the identity profunctor $\hom\colon \mathcal C^\text{op}\times\mathcal C\to \bf Sets$, then $\text{Elts }\varphi$ is the arrow category ${\bf Cat}(\Delta[1], \mathcal C)$.

Have you ever seen this construction in other places, exhibiting any other interesting property? Does it have a special name? Any generalization to the enriched setting (where the category of elements may fail to exist, but there the bare definition can be verbatim given)?

I wonder if there a relation between the cograph of $\varphi$ and its category of elements? In some contexts it seems the answer is yes (more on this if necessary, but it's not my priority at the moment).

Improve this question
$\endgroup$
3
  • 2
    $\begingroup$ I would consider this just as the category of elements of $\varphi$, unless I’m missing something. How you would you see this as differing from any other construction of the category of elements? $\endgroup$
    – Peter LeFanu Lumsdaine
    Dec 25, 2014 at 0:53
  • $\begingroup$ My question was more on the side of "has anybody used the same construction under a different name?". The answer is yes, in at least a case: arxiv.org/abs/1212.6170 page 25. I wanted to see if there were others. And I'm also interested in the enriched extension, where Elts(F) does not exist, but I can give the same definition. Is it well-suited to cover all the other interesting properties of the category of elements? On a separate note, merry Xmas! $\endgroup$
    – fosco
    Dec 25, 2014 at 10:23
  • $\begingroup$ I would not think of this as a glueing: Consider my answer to your other question (mathoverflow.net/questions/190492/…). The Category of elements behaves more like a fibre product over a distributor A distributor also induces a category structure on the disjoint union of the objects $\mathrm{Ob}X\amalg \mathrm{Ob} Y$ and this one should be called the glueing as it exhibits the required adjointness property. $\endgroup$
    – Gerrit Begher
    Jan 17, 2015 at 10:22

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.