5
$\begingroup$

If the product of all objects in a finite category exists, is it an initial object? I presume so, but I'm still learning this subject and I can't make a proof go through. Advice welcome. (Not a homework problem, except insofar as I set it for myself.)

By definition, if $x_1, \dots, x_n$ are the objects of the category, their product $\Pi$ is an object equipped with "projections" $p_i : \Pi \to x_i$, such that for every object $x$ of the category and collection of morphisms $f_i : x \to x_i$, there exists a unique morphism $f : x \to \Pi$ such that $p_i \circ f = f_i$.

$\endgroup$
  • $\begingroup$ Interesting question. Its certainly true in a poset. $\endgroup$ – goblin Apr 16 '15 at 14:08
  • $\begingroup$ are you clear what you mean by product here? i am not very well-up in category theory, so perhaps you can indicate for me what is wrong with the following line of thinking: if the category includes all products of (let us concede, distinct) objects, then unless such a product is trivial the category cannot be finite if it contains more than one object! $\endgroup$ – David Holden Apr 16 '15 at 14:08
  • $\begingroup$ @armchairprogrammer Perhaps the result you are thinking of is that if the limit of the identity diagram exists, then it is an initial object. $\endgroup$ – Zhen Lin Apr 16 '15 at 14:10
7
$\begingroup$

Consider the category $\mathsf{C}$ with a single object $x$ and two morphisms: $\operatorname{id}_x$ and $f : x \to x$ such that $f \circ f = f$. Then $x$ is of course the product of all the objects in $\mathsf{C}$. But it's not an initial object, because there are two morphisms $x \to x$.

$\endgroup$
  • $\begingroup$ Right. More generally, any finite non-trivial monoid (viewed as a category) provides a counterexample. $\endgroup$ – goblin Apr 16 '15 at 14:10
  • $\begingroup$ What are the left and right projections out of $x\times x$ here? Once they are fixed, I feel I can always find a diagram so that it doesn't work out. $\endgroup$ – Nikolaj-K Apr 16 '15 at 14:14
  • $\begingroup$ @NikolajK I didn't say $x$ was the product of $x$ with itself (it's not, I think), but $x$ is the product of the family of objects $\{x\}$, which has one element (and the projection $x \to x$ is the identity). If the objects of $\mathsf{C}$ were $x_1, \dots, x_n$, you would be looking at $x_1 \times \dots \times x_n$; here $n=1$. $\endgroup$ – Najib Idrissi Apr 16 '15 at 14:16
  • $\begingroup$ Perfect. That has clarified for me the problem I am actually trying to solve (and which I failed correctly to ask -- rookie mistake). (I am trying to show that a finite category in which all products exists is such that the product of all objects is initial. But I won't ask this officially as I think I can now have a better go at it after your solution.) $\endgroup$ – armchairprogrammer Apr 16 '15 at 14:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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