Matrix algebra

Definition

The algebra made of square matrices is typically called the matrix algebra. More precisely, if F is a field (such as the real numbers or the complex numbers) and n is a positive integer, then the matrix algebra over F of size n×n is denoted by M(n,F) or Mn(F). It consists of all n×n matrices with entries in F, and the operations of addition and scalar multiplication are defined in the usual way.

Characteristics

It is a central algebra.
...

Action of the projective linear group

Let's focus on F=C. The action of the projective linear group PGL(n,C) on the matrix algebra is given by conjugation. That is, for any A in the matrix algebra and any g in PGL(n,C), the action of g on A is defined as gAg1, where g1 denotes the inverse of g. This action is well-defined since scalar matrices are in the center of GL(n,C) and are therefore ignored in the quotient.

Automorphisms

The automorphisms of this algebra are all inner automorphism (see here), so they are all elements of PGL(n,C), because two elements of GL(n,C) give rise to the same inner automorphism if they are multiples of each other by a scalar matrix.

1-parameter group of automorphisms and derivations

Any given 1-parameter group of automorphisms {φt} is, then, a one-parameter subgroup of PGL(n,C), and hence is generated by an element of the Lie algebra pgln(C). So there must exist an AMn(C)gln(C) such that

φt(B)=exp(At)Bexp(At),

for an arbitrary matrix B.
The flow B(t)=φt(B) satisfies B(0)=B and

ddtB(t)|t=0=ABBA=[A,B]

So the 1-parameter group of automorphisms {φt} gives rise to a derivation Dφ of the algebra:

BDφ(B):=ddtB(t)|t=0

Conversely, all the derivations on this algebra are inner: for every derivation D it exists a matrix A of size n×n, such that D=[,A].

Then, given a derivation D we can construct a 1-parameter group of automorphisms φtD in the following way:

φtD(B)=exp(At)Bexp(At),

It turns out that

Compare with flow of observables in CM: the algebra is the algebra of smooth functions, the derivations are vector fields.