I have attempted to find a definition of a monoidal category which incorporates $n$-fold tensor products instead of just binary tensor products.

Definition. A "multi-monoidal category" consists of

  • a category $\mathcal{C}$,
  • for every $n \geq 0$ a functor $T_n : \mathcal{C}^n \to \mathcal{C}$, denoted by $(A_1,\dotsc,A_n) \mapsto A_1 \otimes \dotsc \otimes A_n$,
  • an isomorphism $\eta : T_1 \cong \mathrm{id}_{\mathcal{C}}$,
  • for all $n_1,\dotsc,n_k \geq 0$ an isomorphism $$\mu_{n_1,\dotsc,n_k} : T_k \circ (T_{n_1} \times \dotsc \times T_{n_k}) \cong T_{n_1+\dotsc+n_k}.$$

The following coherence conditions should hold:

  • Coherence of $\eta$ with $\mu$: We have $\mu_{n_1} = \eta \circ T_{n_1} : T_1 \circ T_{n_1} \to T_{n_1}$.

  • Coherence of $\mu$ with $\mu$: The square

$$\begin{array}{cc} T_k \circ \bigl(T_{n_1} \circ (T_{m_{11}} \times \dotsc \times T_{m_{1n_1}}) \times \dotsc \times T_{n_k} \circ (T_{m_{k1}} \times \dotsc \times T_{m_{k n_k}})\bigr) & \rightarrow & T_k \circ (T_{m_{11}+\dotsc+m_{1 n_1}} \times \dotsc \times T_{m_{k1}+\dotsc+m_{k n_k}}) \\ \downarrow && \downarrow \\ T_{n_1+\dotsc+n_k} \circ (T_{m_{11}} \times \dotsc \times T_{m_{k n_k}}) & \rightarrow & T_{m_{11}+\dotsc+m_{k n_k}} \end{array}$$ commutes.

Notice that when $\mathcal{C}$ is discrete, this is the monadic definition of a monoid (as compared to the usual definition).

Questions. (1) Did I forget some coherence condition? (2) Is this concept already known? Does it have a name? It really looks like the most natural thing in the world, especially when you think "operadic" or "monadic". (3) Most important for me: Is this concept equivalent to the definition of a monoidal category? If yes, what is a reference for this? The idea for the equivalence is straight forward (the $n$-fold tensor product is an iteration of binary tensor products etc.), but I believe that it will probably require much work to check the coherence conditions in both directions.

  • 4
    $\begingroup$ Almost surely what you've written down is an action of the operad Assoc on a category. The equivalence with the standard definition is probably "implicit in results of Mac Lane", or some such muttering. I'm reminded of a definition of "symmetric monoidal category" that I think I saw in a paper of Deligne's (although I don't remember where exactly) that asks for a functor $\mathcal C^S \to \mathcal C$ for each finite set $S$, with natural maps that include the permutations of $S$. So if you think the coherences are correct (I didn't check carefully), then go ahead and use it! $\endgroup$ Jan 8, 2015 at 5:27
  • 1
    $\begingroup$ I believe that what you're after is Max Kelly's notion of a club; there is a club whose pseudo-algebras are precisely monoidal categories, and if you unravel the definition of being a pseudo-algebra for this club you should get what you wrote above. $\endgroup$ Jan 8, 2015 at 7:25
  • $\begingroup$ Also relevant: mathoverflow.net/questions/8252/… $\endgroup$ Jan 8, 2015 at 9:41
  • $\begingroup$ You should look at the beginning of Lurie's Higher Algebra section 2 (and then the rest of it) $\endgroup$
    – Adam Gal
    Jan 8, 2015 at 10:28
  • $\begingroup$ Have you considered an approach similar to Durov? "Algebraic (Strong-/Pseudo-) Monads on Cat" or something like this. $\endgroup$ Jan 17, 2015 at 8:32

1 Answer 1


Look at Section 3 of Leinster's Higher Operads, Higher Categories, where the term used is "unbiased monoidal category."


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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