Here, 𝔤𝔩(n,𝐂) is a complex vector space, but 𝔤 is a vector space over the real numbers. When I say that R*:𝔤→𝔤𝔩(n,𝐂) is linear, the only thing that makes sense is for it to be real linear, i.e. R*(λX)=λR*(X)∀λ∈𝐑.
Representations
Representations
In the next part of the course, we're going to focus on representations of Lie groups. Remember that a (complex) representation of a group is a homomorphism R:G→GL(n,𝐂) , that is an assignment of a matrix R(g) to each group element such that R(g1g2)=R(g1)R(g2) and R(1)=I . The image of this representation is a group of matrices which is a quotient of G . In this course, we'll focus on smooth representations of matrix groups.
Why are we focusing on representations? There are many fantastic applications:
-
There are internal applications to the theory of Lie groups: by studying at the adjoint representation, you can completely classify the Lie algebras of compact Lie groups.
-
There are applications to other areas of mathematics, like invariant theory.
-
There are applications to particle physics. For example, by comparing tables of hadrons found in particle accelerators with weight diagrams of representations of the Lie group SU(3) , Gell-Mann (and independently Ne'eman) was able to guess at the underlying internal structure of these particles and so develop the quark model of matter. This led him to predict a particle with strangeness equal to minus 3, which was discovered in 1964. We will discuss this application in detail.
Recap
I just want to recap the punchline of the first half of the course, as this will be the basis for everything that comes after.
Given a smooth representation R:G→GL(n,𝐂) we get a Lie algebra representation R*:𝔤→𝔤𝔩(n,𝐂) , that is a linear map such that R*[X,Y]=[R*X,R*Y] .
The key property of R* was the equation R(expX)=exp(R*X).
If G is a path-connected group (i.e. any two matrices in G are connected by a smooth path of matrices in G ) then R is determined by R* .
This is because, in this case, G is generated as a group by exp(𝔤)⊂G . The proof of this lemma is an exercise.
Given R*:𝔤→𝔤𝔩(n,𝐂) , does R(expX)=exp(R*X) give a well-defined representation R:G→GL(n,𝐂) ? Lie's theorem told us that this is true if G is simply-connected. If G is not simply-connected, we need to think. The first example we'll consider is U(1) , which is not simply-connected, but all the other examples we will consider are simply-connected.
Plan
Our plan for the rest of the course is:
-
study the representations of U(1)
-
study the representations of SU(2)
-
study the representations of SU(3)
-
study the general theory.
Pre-class exercise
In an earlier video, we constructed a map R:SU(2)→SO(3) . Check that R(M1)=R(M2) if and only if M1=±M2 . Given a representation S:SO(3)→GL(n,𝐂) , we get a representation R∘S:SU(2)→SO(3) . Show that a representation T:SU(2)→SO(3) has this form if and only if T(-M)=T(M) for all M∈SU(2) .
Suppose we have a Lie algebra 𝔤⊂𝔤𝔩(n,𝐑) consisting of real matrices. Consider the subspace 𝔤⊗𝐂⊂𝔤𝔩(n,𝐂) consisting of matrices of the form M+iN with M and N in 𝔤 . Show that:
-
this is a Lie subalgebra, i.e. that it is preserved by Lie bracket
-
if f:𝔤→𝔤𝔩(m,𝐂) is a real-linear Lie algebra homomorphism then f𝐂:𝔤⊗𝐂→𝔤𝔩(m,𝐂) defined by f𝐂(M+iN)=f(M)+if(N) for M and N in 𝔤 is also a Lie algebra homomorphism.