1
$\begingroup$

It is probably a trivial question. But I don't see the answer (and I didn't find anywhere).

Given a (complete and cocomplete) category X and an object A of X, we can define the "undercategory" A/X. See http://ncatlab.org/nlab/show/under+category

I have already noticed that the coproduct of a set {i_l: A to X} is the "natural injection" of A in the colimit of the obvious diagram defined by the set.

I'm trying to understand how the product in A/X looks like, in terms of colimits, limits, products or coproducts of X.

I appreciate any help. Thank you very much!

$\endgroup$
2
  • $\begingroup$ Instinctive guess: pushout? $\endgroup$
    – Yemon Choi
    Nov 27, 2011 at 8:43
  • $\begingroup$ If K={i_l: A to X} is a family of objects in A/X. The "natural injection" morphism A to colimit of K is the coproduct of the objects in K. In particular, if K is a family of only two objects, then the natural morphism A to Pushut (K) is the coproduct... $\endgroup$
    – Fernando
    Nov 27, 2011 at 9:12

2 Answers 2

5
$\begingroup$

For a family of objects $A \to B_i$ in $A/X$, their coproduct is usually called the pushout of the morphisms $A \to B_i$ in $C$. It represents the functor $X \to \mathrm{Set}$, which maps $P$ to the set of families of morphisms $B_i \to P$ which "coincide" on $A$. Pushouts in $X$ may be constructed via coproducts and coequalizers in $X$. The idea is to take $\coprod_i B_i$ and then identify $A \to B_i$ with $A \to B_j$.

The products in $A/X$ are more easy: The forgetful functor $A/X \to X$ preserves (also creates) them. This means that for a family of objects $f_i : A \to B_i$ the product is given by $f : A \to \prod_i B_i$, where $f$ is defined by $\mathrm{pr}_i f = f_i$.

$\endgroup$
0
4
$\begingroup$

Here is another perspective: the undercategory $A \downarrow \mathcal{X}$ is the category of algebras (Eilenberg-Moore category) for the monad whose underlying functor takes an object $X$ to $A \sqcup X$. (The unit of the monad is the coproduct inclusion $X \hookrightarrow A \sqcup X$; the multiplication is $\nabla_A \sqcup X: A \sqcup A \sqcup X \to A \sqcup X$.) Then we can just quote the result that a monadic functor

$$Alg_M \to \mathcal{X}$$

preserves and reflects limits, and in particular products.

$\endgroup$
4
  • $\begingroup$ @Todd: Do you cite that M is closed under composition for (E,M)-structured categories when you want to explain someone that the composition of two injective maps is injective? ;) (Do not take this comment too seriously) $\endgroup$ Nov 27, 2011 at 21:07
  • 1
    $\begingroup$ Okay, I won't. :-) It's just another point of view, and may carry some resonance for some readers. If it doesn't, then fine. $\endgroup$
    – Todd Trimble
    Nov 27, 2011 at 22:05
  • $\begingroup$ A bit more serious: Your proof only works if $X$ has coproducts. But the fact that $A/X \to X$ creates and preserves limits holds in general. If $X$ has coproducts, I would just use that $A \coprod -$ is left adjoint to $A/X \to X$. $\endgroup$ Nov 28, 2011 at 8:24
  • 1
    $\begingroup$ Yes, I know. But sometimes a little light can be shed with a mild extra hypothesis. Knowing that co-slices are categories of algebras, and that slices are categories of coalgebras, can come in handy on occasion, even if you have to assume a little extra hypothesis. That's really the reason for giving my answer -- to share a factoid that could come in handy for someone one day. I hope you don't mind! $\endgroup$
    – Todd Trimble
    Nov 28, 2011 at 12:17

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.