1
$\begingroup$

Consider $A\xrightarrow{ \ f \ }B$, be a mapping in the category of sets. I would like to get a glimpse of how one argues in category theory with regards to finding the limit of the above diagram.

I understand that a limit is a left cone $C$ and two maps $C\xrightarrow{ \ c_1 \ }A$ and $C\xrightarrow{ \ c_2 \ }B$ (satisfying $c_2 = f \circ c_1$) which is terminal, i.e., for any other cone $\tilde{C}$, there exists unique morphism from $\mu: \tilde{C}\to C$, which makes everything commute.

My first question is the following:

How does one go about constructing $C$, $c_1 , c_2$ in the first place? Everything feels terribly abstract. My gut feeling is that we should construct $C$, only from the given data, namely $A\xrightarrow{ \ f \ }B$, and the only "natural thing" for me would be $C=A$, $c_1=\mathbb{1}_A , c_2=f$, as this is effectively the only givens that I have.

Is the whole point of category theory to make a complicated construction once and then by the universal property argue that this is the only construction available (up to isomorphisms in the category)?

My second question is with regards to a similar construction of the pullback:

If I am given $f:A\to W$ and $g:B\to W$, then a limit of this diagram would be some set $C$ and a pair of maps $c_1 , c_2$ that makes everything commute and is furthermore a terminal cone. Could I see please a clear construction and explanation of why $C$ should be $A\times_W B$? Does this construction require of the category to have equalizers?

$\endgroup$
1
$\begingroup$

How does one go about constructing $C$, $c_1 , c_2$ in the first place?

Your answer for the limit of $A\xrightarrow{ \ f \ }B$ is correct, as is the intuition you used to arrive at it. If however you would like a more methodical answer, there is a general construction for concrete limits over directed systems. It tells you that the limit is $\{(a,b)\in A\times B\mid f(a)=b\}.$ This set is called the graph of $f$, and it is isomorphic to $A$.

Could I see please a clear construction and explanation of why $C$ should be $A\times_W B$?

Well $A\times_W B$ makes the diagram commute, and it imposes only those equations required for making it commute.

Does this construction require of the category to have equalizers?

If a category has all pullbacks and a terminal object, then it has all equalizers as well (since an equalizer may be written as a pullback). But it is possible for a category to have one pullback, and fail to have some other equalizer.

$\endgroup$
  • $\begingroup$ It's not quite true that an equalizer can be written as a pullback. The usual construction requires also having products which having just pullbacks does not imply. However, if you have a terminal object, then having pullbacks implies having products and thus equalizers. Of course, Set has all these things. $\endgroup$ – Derek Elkins Nov 4 '17 at 14:12
  • $\begingroup$ @DerekElkins thank you for the correction. $\endgroup$ – ziggurism Nov 4 '17 at 14:15
1
$\begingroup$

In the category of sets, there's an easy construction of the limits $L$ of diagrams $D$. The elements of $L$ are in obvious bijection with functions $1\to L$, where $1$ is your favorite one-element set. By definition of limit, these functions are in natural bijection with the cones from $1$ to $D$. Such a cone is a family of functions, $1\to X$ for all the objects $X$ in $D$, subject to commutativity conditions (as in the definition of "cone"). But a function $1\to X$ amounts to an element of $X$, so you have a family of "chosen" elements, one in each set $X$ of the diagram $D$. The commutativity requirements for a cone then say that, if $f:X\to Y$ is any one of the morphisms in $D$, then $f$ has to map the chosen element of $X$ to the chosen element of $Y$. Putting all of this together, you get that an element of $L$ amounts to a choice of particular elements from all the sets in $D$, such that the chosen elements map to each other under all the morphisms in $D$. (Note that, when $D$ is just $f:A\to B$, this description reduces to the $(a,f(a))$ description in ziggurism's answer, and that the general description also reduces to the explicit description of pullbacks.)

$\endgroup$

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.