Skip to main content
4 of 6
Changed the answer to the opposite conclusion, which has the virtue of being accurate.
Tim Campion
  • 64k
  • 13
  • 143
  • 384

Note: I originally posted an answer claiming the opposite, and then deleted it because it was wrong. I have since reworked it and it should be accurate now.

This doesn't directly answer your question, but you may be interested to know that the conjecture is true in the infinite case. In fact, for every infinite $\kappa$ there are $2^\kappa$ categories with $<\kappa$ morphisms, up to equivalence or isomorphism, $2^\kappa$ of which are monoids.

On the one hand, $2^\kappa$ is an obvious upper bound on the number of categories with $<\kappa$ morphisms: since composition is a partial binary operation on the set of morphisms, there are at most $\sum_{\lambda<\kappa} 2^{\lambda\times \lambda \times \lambda} \leq \kappa 2^\kappa = 2^\kappa$ category structures.

On the other hand, it's easy to find $\kappa$-many pairwise-inequivalent connected categories with $<\kappa$ morphisms: for instance, it suffices to consider the oridnals. By taking disjoint unions of these, we get $2^\kappa$ pairwise-inequivalent categories.

We can tweak this construction to get $2^\kappa$ pairwise-inequivalent monoids. For each $S\subseteq \kappa$, we can adjoin a bottom element to $\coprod_{\lambda \in S} \lambda$ to get a semilattice $(\coprod_{\lambda \in S} \lambda)_{\bot}$, and then consider the corresponding monoid under $\vee$. We obtain $2^\kappa$ monoids this way by taking different $S$, and they are still pairwise nonisomorphic as monoids, because a monoid isomoprhism entails a semilattice isomorphism, which entails an isomorphism of the original disjoint-union categories, which are clearly pairwise inequivalent.

Tim Campion
  • 64k
  • 13
  • 143
  • 384