I am reading Category Theory for Programmers and I am having some trouble in Part 2, Chapter 2: Limits and Colimits.

The author writes:

Now that we have two functors, we can talk about natural transformations between them. So without further ado, here’s the conclusion: A functor $D$ from $I$ to $C$ has a limit $\lim D$ if and only if there is a natural isomorphism between the two functors I have just defined: $$C(c, \lim D) ≃ Nat(Δc, D)$$

How can you express the existence of this natural isomophism as a commutative diagram?

  • $\begingroup$ Can you elaborate on what you mean by "as a diagram"? $\endgroup$ – Hurkyl Sep 22 '17 at 18:40
  • $\begingroup$ @Hurkyl A commutative diagram with objects $I$, $C$ and possibly a topos like $\mathbf{Set}$ for the Hom-objects, and functors as arrows $D$, $\Delta_c$, $C(-,\lim D)$, $Nat(\Delta_-, D)$ such that there's some dotted arrow representing this natural isomorphism which makes the diagram commute if it exists, thus proving the existence of the limit? $\endgroup$ – Karl Damgaard Asmussen Sep 24 '17 at 8:55

The formula you write above is actually a description of the relevant diagram!

For an object $c\in \mathcal C$, and $D\in\mathcal{C^D}$, what is a natural transformation $\gamma\in\mathrm{Nat}(\Delta c,D)$ like? Like usual, it's going to need to satisfy the commutativity condition $\gamma_{d'}\circ\Delta c(f)=D(f)\circ\gamma_d$ for all $f:d\to d'$ in $\mathcal D$; however, no matter what $f$ is, by definition $\Delta c(f)$ is just $id_c$, so we can simplify that condition to $\gamma_{d'}=D(f)\circ \gamma_d$. But that makes $\gamma$ precisely a cone over $D$. In particular, the limiting cone is just the cone $\alpha:\Delta\circ\lim D\to D$ corresponding to $id_{\lim D}$ by the above isomorphism.

Moreover, given such a $\gamma$, the isomorphism above means it corresponds to a unique morphism $\mathcal C(c,\lim D)$ that is the arrow $f_\gamma$ typically drawn as "the dotted line" on paper. To see why this has the desired property, remember that the isomorphism $\phi:\mathcal C(-,lim D)\to\mathrm{Nat}(\Delta -,D)$ is natural, so that $$\mathrm{Nat}(\Delta(f_\gamma),D)\circ\phi_{\lim D}=\phi_c\circ\mathcal C(f_\gamma,\lim D).$$ If you chase what happens to $id_{\lim D}$ in this square, you'll see that this says exactly that composing the limiting cone with $f_\gamma$ gives you the cone $\gamma$.

  • $\begingroup$ Do you think you could maybe draw it up in this very nice editor and paste the generated tikzcd code? (More info here.) $\endgroup$ – Karl Damgaard Asmussen Sep 25 '17 at 10:58
  • $\begingroup$ I'm not sure how I'm going to draw an arbitrary limit diagram, which might be infinite... $\endgroup$ – Malice Vidrine Sep 25 '17 at 12:53
  • $\begingroup$ All right, I'll try to follow your instructions. Thanks a bunch! $\endgroup$ – Karl Damgaard Asmussen Sep 25 '17 at 13:56
  • $\begingroup$ If it makes it easier, just try to use a simple category for $\mathcal D$, like $\mathbb N$ with its usual ordering, or the diagram $\bullet\rightarrow\bullet\leftarrow\bullet$ (which is the shape whose limits are pullbacks). $\endgroup$ – Malice Vidrine Sep 25 '17 at 17:36
  • $\begingroup$ I think my core confusion lies in formulating what the two functors actually do: how their mappings work, and consequently what their mutual naturality condition looks like. I've drawn a VERY big diagram trying to explain it to myself. $\endgroup$ – Karl Damgaard Asmussen Sep 26 '17 at 11:40

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.