Representation Theory
Some definitions to build enough structure to talk about symmetries in physics by thinking of symmetry groups like shapes that partition space and time.
**Suggested Reading Order: Geometry > Forms > Lie Crap > ** You are Here
Representation Definitions
Representations are maps that take you from a Lie Group, or a Lie algebra, to a place that contains transformations of a vector space. In other words it is a way to interpret the elements of a Lie group as the actual actions that would turn adn reshape the vector space.
Defintion: Let
be a Lie group and
a vector space over the field
. Then a representation of
on
is a group homomorphism
given by
We denote by
the linear automorphism group of the vector space
. In other words the group, under composition, of maps from
that are linear isomorphisms. Note that
is a Lie Group. One can prove this by viewing
as a matrix group and then realizing that matrix groups are isomorphic to
. More to the point we can say that
Where this is the general linear goup over the equivalent field.
Notation: When the representation
is clear from context we often write for
and
Defintion: A representation
is faithful iff it is injective.
Remark: Notice that since a Lie Group representation
is a homomorphism we have that for any
Definition: Given two representations
and
of Liw group
to the vector spaces
and
respectively, a morphism
is a linear map such that
in simpler notation
If the map is also an isomorphism of vector spaces it is called an isomorphism or an equivalence of representations for some reason.
Now we copy paste for a Lie Algebra. Because we also want to do this for the algebras.
Definition: Let
be the Lie Algebra of Lie Group
and
some vector space over field
. A representation of the Lie Algebra
is a Lie algebra homomorphism
Where
denotes the set of linear endomorphisms of
, these are linear maps from
. Notice that they don’t have to be invertible (i.e. isomorphisms). With the representation of the group we are mapping to the set of Automorphisms
. Notice that the automorphisms are also a subgroup under composition.
Defintion: A lie algebra representation
is faithful iff it is injective.
Remark: Since the representation
is a lie algebra homomorphism we have that
Definition: Given two representations
and
of Lie Algebra
to the vector spaces
and
respectively, a morphism
is a linear map such that
in simpler notation
If the map is also an isomorphism of vector spaces it is called an isomorphism or an equivalence of Lie Algebra representations.
Proposition: Let
be a Lie Group representation then the pushforward of
is a Lie Algebra representation
.
In particular take a look at this diagram and now we can move around it.
This implies that for any vector field
we can write its representation as
We will see that this will later correspond to moving along our symmetries with a some operator.
Definition: Under any representation
of a Lie Group
(or its Lie algebra) on a vector space
over
, an invariant subspace
is a subspace of
such that
(
).
Definition: A representation
of a Lie Group
(or its Lie algebra) on a vector space
over
is irreducible iff threre is no proper invariant subspace
. Otherwise it is reducible
This is really important. In particle physics we define a particle as an irreducible representation of the symmetry group of the theory. Later we will define another way of finding irreducible representations, but for now this should be ok. What it says is that you can take a vector in V where there is some element in
that can map it out of whatever subspace you choose.
Representation Flavours
A quick list of ways to describe representations using your favourite set and space operations.
Definition: Let
be vector spaces with representations
of some Lie group
. Then we can define the follwoing representations of
.
-
Direct Sum:
on
given by -
Tensor Product:
on
Given by -
Dual:
on
given by -
Exterior Power:
on
given by -
Complex Conjugate:
on
given by -
Homomorphism Space:
on
given by -
Outer Product:
on
for some reprentation
of Lie Group
given
by
You can define the same representations for ones that are from Lie Algebras, but I am not going to write them down. In any case, they all act as one would expect, so we just want to use them as tools.
Representing in Inner Product spaces
A vector space
over
with an inner product
is a very useful tool in physics as such spaces can help us extract scalar quantities, such as angles or distances, out of the vector space.
Definition: An inner product
over
can be of many categories, two of which are:
- Euclidean: Bilinear, symmetric, and positive definite.
- Hermitian: Complex linear in the second argument while complex antilinear on the first, conjucate symmetric, and postive definite.
The second is the notion of distance that lets us construct unitary operators in the Hilbert spaces encountered in Quantum Mechanics.
Definition: A representation
of a Lie group
to a vector space
with a Euclidean (Hermitian) inner product is called orthogonal (unitary) iff
for all
and
. This means that the image of
stays within the orthogonal subgroup
(unitary subgroup
) as determined by the inner product
. This inner product is called G-invariant.
Similarly we copy paste for representations of Lie algebras
Definition: A representation
of a the Lie algebra of
to a vector space
with a Euclidean (Hermitian) inner product is called skew-symmetric (skew-hermitian) iff
for all
and
. This means that the image of
stays within the orthogonal lie subalgebra
(unitary Lie subalgebra
) as determined by the inner product
.
And as one might expect, we state this with proof as an exercise.
Proposition: Let
be a representation of Lie group
and
a
-invariant Euclidean (Hermitian) inner product on
, i.e.
is orthogonal (unitary) on
. Then the induced Lie algebra representation
is skew-symmetric (skew-hermitian).
Now for some physics terms (finally).
Definition: Let
be a skew-hermitian (we also sometimes refer to it as unitary for some reason making the definitions aobove infinitely more confusing) representation of a real Lie algebra
on some complex vector space
. Then
is a skew hermitian endomorphism, implying that
is a hermitian endomorphism (notice that this endomorphism is actually invertible since
is skew hermitian). We call the eigenvalues of
the charges of
in the representation
.
The negative sign is obviously a convention originating form the fact that positive charge is positive.
Definition: A
Decomposition of Representations
Wwe have finally heard the word charge! We will apply this to the context of physics in the future, but for now we need to describe particles. As a result, we need to have a mechanism of describing irriducible representations.
Theorem: Given a representation
of a compact Lie group G to a vector space
over the field
with a
-invariant inner product
,
can be decomposed to the following direct sum representation
where
is an the restriction of representation of
on the subspace
of
. We also have that
Proof: If
is reducible
such that
. Consider the orthogonal complement
(which is also a subgroup with the property that
) . We will show that
.
Let
and
such that
. We know that
for some
. Then we have that
since
. Since
we have a contradiction.
As a result, it is possible to write the representation as
where
is irreducible, and
. Therefore, we can repeat this process until we have reached the direct product of irriducible representations.
Adjoints (ew but important)
In groups we have multiple fundamental operations. Multiplying from the left, multiplying from the right, and conjugation which is given by the following map
Even though it looks stupid, there is some merit to this map as it creates an automorphism of the lie algebra by taking its pushforward
.
So this map will take a vector at some tangent space
and it will give you a vector on the same tangent space! so it can also do this for the Lie algebras (the tangent space at the identity).
Theorem: The map
is a Lie group homomorphism, i.e. a representation of the Lie group
on the vector space
called the adjoint representation or adjoint action of the Lie Group G.
The proof is just calculation from the axioms of a representation.
The reason why we calll this the adjoint representation is because it has the properites that one would expect when reversing the composition. Basically:
Proposition: Given a Lie group representation
on a vector space
and some
we have that
Theorem: The map
is a Lie algebra homomorphism, i.e. a repersentation of the Lie Algebra
on itself as a vector space. This is called the Adjoint representation of the Lie algebra
.
Corollary: For any Lie algebra
the adjoint of the Lie algebra
is given by
This leads us to talk about metrics on Lie groups that are invariant under certain representations.
Definition: Let
be a metric on the Lie Group
. Then it is called:
-
Left Invariant: if
-
Right Invariant: if
-
Bivariant: if it is both left and right invariant. : )
Note: Given any metric
on
we can always define an inner product on
using the metric like so
We could make thi matric left invariant by taking the left translation map, or right invariant by using the right translation map. However, in general, to get a bivariant metric
hsd to be abelian.
Theorem: Let
be a left invariant metric on Lie group
. Then
is bi-invariant iff the inner product on
defined by the metric
is
-invariant, i.e.
Theorem: Let
be a compact Lie group. Then there exists aEuclidean scalar product on the Lie algebra
which is
-invariant. The adjoint representation is orthogonal with respect to this scalar product.
Corollary: Every compact Lie group can have a bi-invariant Riemannian metric.
The Killing Form
Not all
-invariant inner products are created equal. There is a special one that we use all the time to characterize symmetries. We call that the Killing form.
Theorem: Let
be a Lie Algebra over the field
. The Killing form
on
is defined by
This is a bilinear, symmetric form on
.
Proof: Consider
, then
This shows that the killing form is bilinear. To show that it is symmetric we have to prove some properties of the trance.
For an arbitrary linear endomorphism
and a basis
for
we have that
and that
Now consider a linear isomorphism
we can change the basis of
simply by performing
. We can show that the trace is invariant to this as:
Therfore, we know that the trace is symmetric.
Note: The Killing form for Complex Lie Algebras is not Hermitian.
Theorem: Let
be a Lie algebra automorphism of
. Then the Killing form
satisfies the following
A nice application of this theorem is the fact that the Killing form is invariant under the Lie algebra automorphism
.
Corollary: The killing form
defines a bi-invariant symmetric form on any Lie group
.