Let $\mathcal{C}$ a symmetric monoidal category. By using symmetry, it is very easy to show that the contravariant internal hom functor $[-,A]\colon\mathcal{C}^{op}\longrightarrow\mathcal{C}$ is adjoint on the right to itself.

My question is: when is this functor additionally $[-,A]$ a left-adjoint? Do such objects $A$ have a name? In the category of $\textbf{Set}$ it is easily seen that $[-,A]$ preserves the terminal object $\textbf{Set}^{op}$ iff $A$ is the empty set.

New contributor
Cat John is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
  • 3
    Are you sure your last claim is correct? Unless I’m missing something, $[-,0]$ isn’t a left adjoint. If it were, then mapping out of the adjoint into 2 and noting that $0$ and $1$ have unique “square roots” in $\mathrm{Set}$ shows that the adjoint would be the functor $[-,0]$ itself. But $\mathrm{Set}([X,0],Y) \simeq \mathrm{Set}([Y,0],X)$ is not true in general: consider $X=0$, $Y = 2$. – Peter LeFanu Lumsdaine Nov 9 at 11:00
  • Sorry! I really meant $[-,A]$ preserves the empty set (limit in $\textbf{Set}^{op}$) iff $A=\varnothing$. – Cat John Nov 9 at 12:27

Suppose that everything is dualizable (as in the category of finite-dimensional vector spaces over a field, for example), and write $TX=[X,A]=X^*\otimes A$. Then \begin{align*} \mathcal{C}(TX,Y) &= \mathcal{C}(1,X\otimes Y\otimes A^*) = \mathcal{C}(Y^*\otimes A,X) = \mathcal{C}^{\text{op}}(X,TY) \\ \mathcal{C}(W,TX) &= \mathcal{C}(1,W^*\otimes X^*\otimes A) = \mathcal{C}(X,W^*\otimes A) = \mathcal{C}^{\text{op}}(TW,X) \end{align*} so $T$ is self-adjoint on both sides.

If $\mathcal{C}$ is Cartesian, then $A$ must be initial.

Indeed, let $G : \mathcal{C} \to \mathcal{C}^{\mathrm{op}}$ be the right adjoint to $[-,A]$. Then we have a natural bijection $\mathrm{Hom}_\mathcal{C}([X,A],Y) \simeq \mathrm{Hom}_{\mathcal{C}^{\mathrm{op}}}(X,G(Y))$. Let $X = 1$. Then $\mathrm{Hom}_\mathcal{C}([1,A],Y) \simeq \mathrm{Hom}_\mathcal{C}(A,Y)$. On the other hand, $\mathrm{Hom}_{\mathcal{C}^\mathrm{op}}(1,G(Y)) \simeq {*}$ since $1$ is initial in $\mathcal{C}^\mathrm{op}$. So, $A$ is initial.

As Peter noted in the comments, $[-,0]$ is not a left adjoint in $\mathrm{Set}$. To give an example of a Cartesian closed category in which this is true, this is true in any Boolean algebra considered as a category. In this case $[-,0]$ is simply the negation. So, $[[X,0],0] = X$ and $[-,0]$ is its own right adjoint.

  • 1
    Wait... $G(1)$ is terminal in $\mathcal{C}^{op}$, so it must be initial in $\mathcal{C}$ and you get no further condition on $G(1)$. Am I missing something? – Denis Nardin Nov 9 at 10:13
  • 2
    Oops. My mistake. I'll edit the answer – Valery Isaev Nov 9 at 10:16

Your Answer

Cat John is a new contributor. Be nice, and check out our Code of Conduct.

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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