Etydhv

I'm trying to understand what a Monoid is from category theory perspective, but I'm a bit confused with notation used to describe it. Here is Wikipedia:

In category theory, a monoid (or monoid object) $(M, mu, eta)$ in a monoidal category $(mathcalC, otimes, I)$ is an object $M$ together with two morphisms

• 
  $mu: M otimes M to M$ called multiplication,


• 
  $eta: I to M$ called unit, [...]

My confusion is about morphism notation. Why is the binary operation $otimes$ a part of the morphism notation? My understanding of morphism is that it's a kind of function that can map one type to another (domain to codomain) like $M to M$... Why is the operation $otimes$ a part of the domain in the definition?

The second confusion is about $I$. Why $I$ is a domain... there is no $I$ object in a Monoid at all. It's just the neutral element of object $M$.

I understand that a Monoid is a category with one object, identity morphism and binary operation defined on this object, but the notation makes me think that I don't understand something.

Is $M otimes M$ somehow related to cartesian product, so the domain of the morphism is defined as $M times M$ ?

An ordinary monoid is a set $M$ equipped with a map $mucolon Mtimes Mto M$ and an element $ein M$ such that certain axioms are satisfied.

Now the goal of the definition you quoted (monoid object in a monoidal category) is to generalize the notion of ordinary monoid to the most general categorical context possible.

The first step is to observe that in a general category it doesn't make any sense to talk about an "element" of an object. So let's replace the element $ein M$ with a map $ecolon 1to M$, where $1 = *$ is a set with a single element. The point is that (in the category of sets) specifying a map $1to M$ is exactly the same as specifying an element of $M$: given an element $ein M$, we can define a map $1to M$ by $*mapsto e$, and given a map $ecolon 1to M$, we can define an element of $M$ by $e(*)$.

Ok, now we want to generalize away from the category of sets and talk about monoids living in other categories. So let $mathcalC$ be a category. A monoid in $mathcalC$ should be an object $M$ in $mathcalC$ together with maps $mu$ and $e$ satisfying certain axioms. To make sense of the maps $mu$ and $e$, we need:

• A specified way of putting two copies of $M$ together to make the domain of $mu$, analogous to $Mtimes M$ in the case of sets. Let's denote this by $Motimes M$.


• A special object to be the domain of $e$, analogous to $1$ in the case of sets. Let's denote this by $I$.

The operation $otimes$ on objects and the special object $I$ are the basic data of a monoidal category.

If $mathcalC$ has finite products, you're welcome to define $otimes$ to be the product $times$ and $I$ to be the terminal object $1$. This is called the Cartesian monoidal structure. But the point is that we don't need $Motimes M$ to be the product and $I$ to be the terminal object in order to make sense of the definition of monoid.

We do need to assume a little bit more about $otimes$ and $I$ in order to make sense of the monoid axioms. In particular, we need $otimes$ to be associative (up to isomorphism), $$Motimes(Motimes M) cong (Motimes M)otimes M,$$ and we need $I$ to act as an identity for $otimes$ up to isomorphism, i.e. $$Mcong Iotimes Mcong Motimes I.$$
Making all of this appropriately functorial and natural leads directly to the full definition of monoidal category.