4
$\begingroup$

Let $(\mathcal{C}, \otimes , I)$ and $(\mathcal{C}, \otimes', I')$ be tensor categories. A tensor functor $F: (\mathcal{C}, \otimes , I)\to (\mathcal{C}', \otimes' , I')$ consists of a functor $F: \mathcal{C}\to \mathcal{C'}$ together with natural isomorphisms $J_{X,Y}: F(X)\otimes' F(Y) \to F(X\otimes Y)$ and an isomorphism $\varphi: F(I)\to I'$ such that three compatibility diagrams with respect to the associators commute.

Next, consider the following two definitions:

(1) A tensor functor $F$ is called an equivalence of tensor categories if it is an equivalence of ordinary categories.

(2) A tensor functor $F: (\mathcal{C}, \otimes , I)\to (\mathcal{C}', \otimes' , I')$ is called an equivalence of tensor categories if there exists a tensor functor $F': (\mathcal{C}', \otimes' , I')\to (\mathcal{C}, \otimes , I)$ together with natural tensor isomorphisms $\eta: \operatorname{id}_{\mathcal{C'}}\to FF'$ and $\theta: F'F\to \operatorname{id}_{\mathcal{C}}$.

Definition (1) is the definition in Etingof's book "Tensor categories" and definition (2) is in Kassel's book "Quantum groups". Clearly definition (2) implies definition (1). Is the converse true?

$\endgroup$
1
  • 1
    $\begingroup$ (2) is the "correct" defintion. Equivalences are defined in any 2-category, and we just apply it to the special example of the (strict) 2-category of tensor categories. (1) is not the "correct" definition, it is a characterization, or actually a theorem (using (2) as a definition) that the forgetful 2-functor from tensor categories to categories is 2-conservative - a categorification of the statement that the forgetful functor from monoids to sets is conservative. $\endgroup$ Commented Jun 3, 2021 at 22:27

1 Answer 1

7
$\begingroup$

Yes. Given (1) with quasi-inverse $F':\mathcal{C}'\to\mathcal{C}$, there is a unique way to make $F'$ a tensor functor such that $\eta$ and $\theta$ are tensor isomorphisms. Specifically, the tensor structure is

$$ F' X \otimes F' Y \cong F' F(F' X \otimes F' Y) \cong F'(F F' X \otimes' F F' Y) \cong F'(X\otimes' Y) $$

and similarly for the unit.

More generally, for any 2-monad $T$ on a 2-category $\mathcal{K}$, a pseudomorphism $f:A\to B$ of $T$-algebras is an internal equivalence in the 2-category of $T$-algebras if and only if its underlying morphism in $\mathcal{K}$ is an equivalence. This is a categorification of the fact that any monadic functor is conservative.

$\endgroup$
2
  • 1
    $\begingroup$ Actually a lot of things are to be checked here, namely the coherence diagrams (when no 2-monad-machinery is available). Also, we need to turn the equivalence into an adjoint equivalence first. See for instance my book, Satz 8.4.6, which contains the large diagram showing coherence of $F'$ with the associator. $\endgroup$ Commented Jun 3, 2021 at 22:19
  • $\begingroup$ Yes, of course. One has to check the coherence diagrams when using 2-monads too, it's just that one can do it once and for all for all 2-monads. $\endgroup$ Commented Jun 4, 2021 at 2:36

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .