Let $K$ be a field, regarded as a monoidal category in the following way:

  1. Objects are elements $x\in K$;
  2. Morphisms between $x\to y$ are only identities;
  3. The (strictly symmetric) monoidal structure $x\otimes y$ is given by the product $x\cdot y $ (it is not closed because of $0$)

Is there a special name for categories enriched over $K$?

Whatever these are, they have to fulfill quite strange properties: let $V\in K\text{-Cat}$, then $\hom(v,w)\otimes \hom(w,z) = \hom(v,w)$, which taking into account the fact that each $\hom(v,w)$ is a scalar in $K$, implies that $\hom(v,w)\hom(w,z) = \hom(v,z)$, i.e. $\hom(v,v)=1$ and $\hom(v,w) = \hom(w,v)^{-1}$.

I started wondering if a $K$-vector space could be regarded as a $K$-category, but this seems to be false. I thought that $K$ was a $K$-category, but it seems that if it is, then $\hom(x,y)=x^{-1}y$, impossible if $x=0$.


You make no use of the additive structure, so let's instead talk about categories enriched over monoids $M$. As you observe, in such an enrichment only the invertible elements of $M$ arise, so let's instead talk about categories enriched over groups $G$. Then my claim is that

a category enriched over $G$ is the same thing as a set on which $G$ acts freely and transitively, or in other words a $G$-torsor.

The point is that an equivalent definition of a $G$-torsor is that it is a $G$-set $S$ in which, for every $x, y \in S$, there is a unique $g \in G$ such that $gx = y$. Given this unique $g$ there is a unique enrichment of $S$ over $G$ which sets $\text{Hom}(x, y) = g$. Conversely, given a $G$-enriched category, if $\text{Hom}(x, y) = g$ then define $gx = y$. This definition is independent of the choice of $y$ precisely because composition is compatible with the group operation in $G$.

So, unfortunately, categories enriched over monoids are not too interesting. It's much more interesting to enrich categories over monoids equipped with poset structures; these give you generalizations of (Lawvere) metric spaces.

In general, $V$-enriched categories may be thought of as at least analogous to $V$-module categories. You can try to pass from one to the other if certain adjunctions exist.

  • 1
    $\begingroup$ What should a $K$-torsor be? I mean, yes, it is a $K$-category. But are there niftier characterizations? $\endgroup$ – Fosco Loregian Feb 7 '15 at 22:45
  • $\begingroup$ @tetrapharmakon: one definition is invertible $K$-module. Then it's true that the groupoid of invertible $K$-modules is equivalent to the groupoid of $K^{\times}$-torsors. $\endgroup$ – Qiaochu Yuan Feb 7 '15 at 22:47

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.