If $M$ is a left-module over a ring $R$, then $M$ carries an action of $Z(R)$, the center of $R$, since it is a subring of $R$, by restriction of scalars. Since $Z(R)$ is commutative we may view this as a left or right action. Similarly, if $M$ is a right $R$-module and $N$ is a left $R$-module, while one usually defines $M\otimes_R N$ as just an abelian with no action, it actually carries an action of $Z(R)$.

Categorifying this picture horizontally, let $C$ be any category enriched over $\mathcal{V}$, let $M\colon C^\text{op}\to \mathcal{V}$ and $N\colon C\to\mathcal{V}$ be contravariant and covariant $\mathcal{V}$-enriched functors, which we view as right- and left-modules, respectively. Then we define the center to be $Z(C)=\hom(1_C,1_C),$ the endonatural transformations of the identity functor on $C$. We can define $\mathcal{V}$-natural transformations as elements $I\to\hom(x,x)$, for $x\in C$ and where $I$ is the unit of the monoidal structure on $\mathcal{V}$. In this case, $Z(C)$ is a set and it is easy to see how elements act on modules. As Martin points out in the comments, for $\alpha$ any natural transformation in $Z(C)$, $M\alpha$ is a natural transformation $M\to M$.

However it is more natural for $Z(C)$ to retain the structure of an object of $\mathcal{V}$, (the center of a ring is a ring, for example). Hence we should define $Z(C)$ to be the end $\int_{x\in C}\hom_C(x,x)$. This is an object of $\mathcal{V}$, hence is not a set, and we may not speak of its elements. We might consider the set of generalized elements, morphisms $I\to\hom_C(x,x)$, which are the $\mathcal{V}$-natural transformations. But these do not constitute an object of $\mathcal{V}$, and hence do not give an action on $M$ (which, remember, is defined as a $\mathcal{V}$-enriched functor).

So how do we construct this action as an enriched functor, from the functor $M$? How should we think about this generally? Like any endomorphism hom-object, $Z(C)$ must be a monoid object in $\mathcal{V}$. Does a functor out of a category somehow induce a functor on the delooping of this monoid of endomorphisms of the identity functor on the category?

  • $\begingroup$ Surely the first thing to do is to determine how $Z (\mathcal{C})$ acts on $M$? $\endgroup$ – Zhen Lin Mar 3 '14 at 10:08
  • $\begingroup$ @ZhenLin: Yes, that is surely the thing to do. But it's not clear in the enriched case, is it? $\endgroup$ – ziggurism Mar 6 '14 at 1:12

I think I was confused by the module language. I was looking for a functor from $Z(C)$ to $\mathcal{V}$. But an action of a category $C$ on a module $M$ is just morphisms $C(a,b)\otimes M(a)\to M(b)$. So an action isn't a functor, although a functor is a thing which always has an action. And then we can loosen the notion of action to be any collection $\{C_a\}_{a\in C}$ of $\mathcal{V}$-objects indexed by objects of $C$ with morphisms $C_a\otimes M(a)\to M(a)$, and the center $Z(C)$ has an action of this form by virtue of the components of the end $\int_C\hom(1,1)\to\hom(a,a)\to\hom(Ma,Ma).$ Basically the center acts because it maps to the hom-objects of the category, so we have something like "restriction of scalars" of the module $M$. And this of course agrees with the obvious action by multiplication in the set-enriched case.


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.