Natural transformation in category theory

72 Views Asked by At

I am learning basics from Category theory and in Steve Awodey's book "Category Theory" I've found the following:

If you think of a functor $F : C \to D$ as “picture” of $C$ in $D$, then you can think of a natural transformation $\varphi_C : FC → GC$ as a “cylinder” with such a picture at each end.

Can someone help me understand where did this "cylinder" come from? How to visualize it?

1

There are 1 best solutions below

0
On

As always, to get an understanding of abstract definitions one should consider examples. The easier, the better.

Take $C=\bullet$ the category with one object and no nontrivial morphisms and $D=\mathsf{Set}$. A functor $F:C\rightarrow D$ does nothing more than picking a set $F(\bullet)$. A natural transformation $\alpha:F\Rightarrow G$ just consists of a function $\alpha:F(\bullet)\rightarrow G(\bullet)$. In some sense a line is a cylinder, isn’t it?

If you are not convinced, take the next best example, where $C=(0\rightarrow 1)$ the free walking arrow (no nontrivial morphisms besides the depicted one). A functor picks two sets $F(0)$ and $F(1)$ together with a function $f:F(0)\rightarrow F(1)$. A natural transformation amounts to two functions $\alpha_0:F(0)\rightarrow G(0)$ and $\alpha_1:F(1)\rightarrow G(1)$ making the square $$\begin{array}{ccc} F(0)&\overset{\alpha_0}\longrightarrow & G(0)\\ \downarrow&&\downarrow\\ F(1)&\overset{\alpha_1}\longrightarrow & G(1) \end{array}$$ commute. In some sense, a square can be regarded as a cylinder, doesn’t it?

If you are still not convinced, take the next best example, where $$C=\begin{array}{ccc} 0&\rightarrow&1\\&\searrow&\downarrow\\&&2\end{array}$$ I leave it as an exercise to spell out what a functor and a natural transformation look like. Can you see how this is a cylinder?

Finally, as some kind of foreshadowing, let me explain, how a natural transformation can be formally considered as a cylinder: A natural transformation can be described as a functor $(0\rightarrow 1)\rightarrow\operatorname{Fun}(C,D)$. Using cartesian closedness of $\mathsf{Cat}$ this is the same as a functor $(0\rightarrow 1)\times C \longrightarrow D$. If you think of $(0\rightarrow 1)$ as a line, $(0\rightarrow 1)\times C$ is a cylinder with base-shape $C$. The functor takes that to its image inside $D$.