I'm currently working through Tom Leinster's Basic Category Theory, and I have searched the internet fruitlessly for examples of functors between functor categories. I haven't yet come up with any intuition on how one might construct or characterize a functor $F : [\mathcal{A}, \mathcal{B}] \to [\mathcal{A}', \mathcal{B}']$, especially with the functor categories $[\mathcal{A}, \mathcal{B}]$ and $[\mathcal{A}', \mathcal{B}']$ as objects of CAT. I feel very bogged down in the details of how the functor $F$ sends functors to functors and natural transformations to natural transformations; I understand that depending on the scenario there might be a natural choice for $F$ mapping $f \in [\mathcal{A}, \mathcal{B}]$ to another functor, and similarly for the natural transformations, but that choice would have to be consistent with the functor and natural transformation axioms for each morphism and object in $\mathcal{A}$. I suppose my confusion is in the multiple layers of information that such a functor $F$ must encode, and how to keep this information consistent and organized when working with functors such as these.

For context, I'm trying to show that $[\mathcal{A^{\text{op}}}, \mathcal{B^{\text{op}}}] \cong [\mathcal{A}, \mathcal{B}]^{\text{op}}$ as objects of CAT, so I need to find two functors $F,G$ so that $FG = 1_{[\mathcal{A}, \mathcal{B}]^{\text{op}}}$ and $GF = 1_{[\mathcal{A^{\text{op}}}, \mathcal{B^{\text{op}}}] }$. I also have terrible intuition for opposite categories, and I cannot see a natural choice for $F$ or $G$. Every attempt I've made ends up with a construction of $F$ that is contravariant, which I don't think can give me an isomorphism.

Any intuition you can impart about my first question, and any hints as to the second would be appreciated.


  • 1
    $\begingroup$ A contravariant functor can be an isomorphism! $\endgroup$
    – Mathmo123
    Dec 28 '14 at 18:28
  • 1
    $\begingroup$ @Mathmo123 Wouldn't it be an anti-isomorphism? $\endgroup$ Dec 28 '14 at 20:14
  • 1
    $\begingroup$ For a trivial example consider the functor $ G-\mathbf {Set}\longrightarrow G'-\mathbf {Set} $ induced by a group homomorphism $ G'\longrightarrow G $. This can be generalized by replacing $ G, G',\mathbf {Set} $ by arbitrary categories. $\endgroup$
    – user158047
    Dec 28 '14 at 20:54
  • $\begingroup$ @JakobWerner That's very nice and simple. It doesn't appear to apply directly to the problem I'm considering, but that is a neat arrangement. Thanks. $\endgroup$ Dec 28 '14 at 21:48

You want to define a contravariant functor $[\mathcal{A}^{op}, \mathcal{B}^{op}] \to [\mathcal{A},\mathcal{B}]$.

Given a functor $T : \mathcal{A}^{op} \to \mathcal{B}^{op}$, define $T_\ast : \mathcal{A} \to \mathcal{B}$ on objects by $T_\ast(X) = X$. Given a morphism $f:X\to Y$ of $\mathcal{A}$, view it as a morphism $f^{op} : Y \to X$ in $\mathcal{A}^{op}$; then $T(f^{op}) : T(Y) \to T(X)$ is a morphism in $\mathcal{B}^{op}$ which corresponds to some morphism $T(f^{op})^{op} : T(X) \to T(Y)$ in $\mathcal{B}$.

So we define $T_\ast$ on morphisms by setting $T_\ast(f) = T(f^{op})^{op}$.

You now need to check that $T_\ast$ is indeed a functor $\mathcal{A} \to \mathcal{B}$.

Assuming this is done, you have a rule that associates to each object $T \in [\mathcal{A}^{op}, \mathcal{B}^{op}]$ an object $T_\ast \in [\mathcal{A},\mathcal{B}]$.

Let's now define $()_\ast$ on morphisms in $[\mathcal{A}^{op},\mathcal{B}^{op}]$, i.e. natural transformations.

Given $\eta : T \to S$ a natural transformation between $T, S : \mathcal{A}^{op} \to \mathcal{B}^{op}$, we want to define a natural transformation $\eta_\ast : S_\ast \to T_\ast$ between $S_\ast, T_\ast : \mathcal{A} \to \mathcal{B}$.

Let $X$ be an object in $\mathcal{A}$. Then $\eta_X : T(X) \to S(X)$ is an arrow in $\mathcal{B}^{op}$, so $\eta_X^{op} : S(X) \to T(X)$ is an arrow in $\mathcal{B}$.

So we define $\eta_{\ast,X} : S_\ast(X) \to T_\ast(X)$ by the rule $\eta_{\ast,X} = \eta_X^{op}$.

You should now check that $\eta_\ast$ really is a natural transformation $S_\ast \to T_\ast$, and that $(\xi \circ \eta)_\ast = \eta_\ast \circ \xi_\ast$ in $[\mathcal{A},\mathcal{B}]$ whenever $\xi : S \to U$ is some other morphism in $[\mathcal{A}^{op}, \mathcal{B}^{op}]$.

Then you have a contravariant functor $()_{\ast} : [\mathcal{A}^{op}, \mathcal{B}^{op}] \to [\mathcal{A}, \mathcal{B}]$.

Applying this construction to the pair of categories $(\mathcal{A}^{op}, \mathcal{B}^{op})$ instead of $(\mathcal{A}, \mathcal{B})$ gives you a contravariant functor $()^{\ast} : [\mathcal{A}, \mathcal{B}] \to [\mathcal{A}^{op}, \mathcal{B}^{op}]$.

You should now check that $()_\ast$ and $()^{\ast}$ are mutually inverse equivalences of categories.

  • $\begingroup$ Thanks. I'll work through the details. $\endgroup$ Dec 28 '14 at 23:02

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.