A
monoid consists of a
set A plus a
function δ:A→End(A) from
A to the
set of
functions A→A. This data must abide by:
Identity: exists
e∈A such that
δ(e)=idA
Associativity:
x δ(a) δ(b)=x δ(a δ(b))
(Where
xf denotes
f(x), so
x δ(a) is
δ(a)(x))
The correspondence of this form to the “
algebraic structure” definition of monoids is
δ(a):=(a×−); ie,
δ takes an element to its left-
action.
Example:
A=R where
δ(a) is the rightwards translation of all of
R by
a. (this is the
(R,+) monoid)
Example:
A=C where
δ(z) is the joined rotation+scaling of all of
C which takes
1 to
z. (this is the
(C,⋅) monoid)
How to understand this?
Firstly note that the Identity axiom implies that
δ is
injective, because we can recover
a from
δ(a) as
a=δ(a)(e).
Injectivity of
δ is actually what we care about (for this discussion).
Injectivity of
δ means that
A≅im(δ); ie, we are justified in
identifying an element
a∈A with its
action δ(a).
In other words, we can conflate
2+1i with the rotation+scaling taking
1 to
2+1i.
Next note that the
Associativity axiom can be rephrased as
δ(y δ(z))=δ(y) ; δ(z)
since we are allowing ourselves to identify elements of
A with their image under
δ, we can think of this as saying
y δ(z)=δ(y) ; δ(z)
where "
=" denotes equality up to
δ.
This tells us that
application of a
δ-
action is the same thing as
composition of
δ-
actions.
Overall, a monoid is a
set A where the elements of
A may be considered themselves as
A-
actions, and where performing
A-
actions is the same thing as composing them.
The elements of
C may be considered themselves as
C-
actions (via rotation+scaling), and multiplication in
C is the same thing as composition of these rotation+scalings.