Decomposing arbitrary representations

  • Context: Undergrad 
  • Thread starter Thread starter StaticZero
  • Start date Start date
  • Tags Tags
    Representations
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
2 replies · 2K views
StaticZero
Messages
2
Reaction score
0
TL;DR
How to split a non-simply reducible representation?
Suppose we have a group ##G = D_3## and its reducible representation ##T = E \oplus E## where ##E## is the two-dimensional irrep. In its own basis, this representation can be written like this (I provide the operator only for ##C_3^+## group element which is counter-clockwise rotation by ##2\pi/3## around ##z##)
$$
T(C_3^+) = \begin{pmatrix}
-\frac12 & -\frac{\sqrt3}2 & 0 & 0 \\
\frac{\sqrt3}{2}&-\frac12 & 0 & 0 \\
0 & 0 & -\frac12 & -\frac{\sqrt3}{2} \\
0 & 0 & \frac{\sqrt3}{2}&-\frac12
\end{pmatrix}
$$

Suppose now that we have done a basis change and forgot what was that old basis which delivered a block-diagonal form for the representation ##T##, so in our new basis all matrices of ##T## are of generic form.

How could we find a basis in which the given representation returns to block-diagonal form?

In the case of simply reducible representation (=that consists of only copies of irreps) it can be easily done by observing the character table and constructing the projector
$$
P^{(\alpha)} = \frac{s_\alpha}{|G|} \sum_g \chi^{(\alpha)}_{g'} T_g
$$
where we sum with respect to group elements and ##s_\alpha## is the dimension of the irrep ##\alpha## and ##g' \equiv g^{-1}##.

Unfortunately, this operator does not help in case of multiple copies of irreps as in the example being investigated where ##T = E \oplus E## because it can 'filter out' only those irreps that have distinct character.

Two other projectors which I can recall for
$$
P^{(\alpha)}_i = \frac{s_\alpha}{|G|} \sum_g T^{(\alpha)}_{g', ii} T_g
$$
$$
P^{(\alpha)}_{ik} = \frac{s_\alpha}{|G|} \sum_g T^{(\alpha)}_{g', ik} T_g
$$
are neither helpful, too, by the similar reasoning.

Could you please explain whether these multiple copies of the same irreps can be 'filtered out'?
 
on Phys.org
Is this question ill-stated, am I missing something trivial or are there other problems with it?
 
If you're just interested in the case ##G=D_3##, here's a possibility: let ##r## and ##s## be the usual generators of ##D_3##, viewed as linear operators on your ##4##-dimensional representation. Then ##r-s## has a nontrivial kernel (since in the ##2##-dimensional standard representation of ##D_3,## there is vertex on an equilateral triangle where these group elements act identically). Now if ##0\neq x\in\text{ker}(r-s),## note that ##\text{Span}(x,rx)## is a ##2##-dimensional invariant subspace. Apply the usual procedure to find an invariant complement.
 
Last edited: