Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have just come across the Wigner-Eckart theorem and am not sure on how to apply it. How do I find the matrix elements of $\langle u|T_a|v\rangle$ in terms of tensor components and the Gell-Mann matrices, where $T_a$ are the SU(3) generators and $|u\rangle$ and $|v\rangle$ are tensors in the adjoint (8-dimensional) representation of SU(3)?

share|cite|improve this question
If $SU(3)$ is not the main issue, and you are just asking about how to apply Wigner-Eckart theorem, see also this question. – Qmechanic May 15 '12 at 11:43
@Qmechanic : Hey, I am a first year physics undergrad and have not done quantum mechanics yet. Actually my motivation for learning lie algebras is classical field theory. I am reading lie algebras in particle physics by Howard Georgi. Please could you tell me how to apply the theorem for the example mentioned above. I have tried hard but not succeeded. Georgi says the matrix elements would be proportional to transpose conjugate of u, T_a and the tensor v. From this, how do I find each of the matrix elements. Thanks in advance. – user7757 May 16 '12 at 5:45
@ramanujan I think that the Wigner-Eckart theorem is not the best method to find the matrix elements in the question. Please tell if you are interested in the matrix elements, I think that I can help you, using other methods – David Bar Moshe May 16 '12 at 6:53
@DavidBarMoshe : Sir, I would really appreciate that. I am interested in the matrix elements themselves, which I thought could only be solved by the Wigner-Eckart theorem. – user7757 May 16 '12 at 6:57
up vote 3 down vote accepted

There are many methods to compute the matrix elements of a simple Lie algebra generators in a given representation. For the problem at hand, I'll try to describe two methods quite in detail and sketch two other methods.

The individual computations actually involve elementary linear algebra and combinatorics, but they are quite lengthy. For example to obtain all matrix elements in the adjoint representation, one needs $ 8 \times 8 \times 8 $ computations


The adjoint representation (by definition) can be realized such that the matrix elements with respect to some orthonormal basis are the structure constants

$ \langle u_b | T_a | u_c\rangle = f^a_{bc}$


In $SU(3)$, the adjoint representation occurs in the tensor product of the fundamental representation and its dual:

$3\otimes\bar{3}\rightarrow 8 \oplus 1$

It is customary to use the quark and anti-quark names for the basis vectors.

the fundamental representation: $ u : (1,0), d : (-1, 1), s : (0, -1) $

its dual $ \bar{u} : (-1,0), \bar{d} : (1, -1), \bar{s} : (0, 1) $

The weights of the adjoint representation are given by: (please see From Slansky's page 32 equation (5.4))

$ v_1 : (1,1), v_2 : (-1,2), v_3 : (2,-1), v_4 : (0,0), v_5 : (0,0), v_6 : (1,-2), v_7 : (-2,1), v_8 : (-1,-1) $

While, for the nonzero weights, there is only one option to construct the tensor product is:

$ v_1 = u \otimes \bar{s} $

$ v_2 = d \otimes \bar{s} $

$ v_3 = u \otimes \bar{d} $

$ v_6 = s \otimes \bar{d} $

$ v_7 = d \otimes \bar{u} $

$ v_8 = s \otimes \bar{u} $

The zero weight subspace is spanned by $ u \otimes \bar{u}$, $ d \otimes \bar{d}$, $ s \otimes \bar{s}$, but we know from the tensor product decomposition that the weight vectors in this subspace must be orthogonal to the scalar $ u \otimes \bar{u} + d \otimes \bar{d} + s \otimes \bar{s}$, thus we can choose:

$ v_4 = \frac{u \otimes \bar{u} - d \otimes \bar{d}}{\sqrt 2} $

$ v_5 = \frac{u \otimes \bar{u} + d \otimes \bar{d} -2 s \otimes \bar{s}}{\sqrt 6} $

Now, The generators of the Lie algebra on the tensor product have the form:

$ T^a = T_{3}^a \otimes I + I \otimes T_{\bar{3}}^a$,

where, the action on the fundamental representation is through the Gell-mann matrices

$ T_{3}^a = \lambda^a $

and the action on the dual is by means of the negative transpose (not the Hermitian conjugate):

$ T_{\bar{3}}^a = -{\lambda^a}^t$

Now, we are in a position to compute matrix elements, for example:

$ \langle v_1 | T^3 | v_3\rangle = -\langle u \otimes \bar{s} | \lambda^3 \otimes I + I \otimes{ \lambda^3}^t | u \otimes \bar{d} \rangle = - \langle \bar{s} | { \lambda^3}^t |\bar{d} \rangle$

That is, one can compute the matrix elements in the adjoint representation in terms of the matrix elements in the fundamental representation and its dual.

It should be emphasized that the matrices obtained by this method will very likely be different from the ones obtained in the first method but they will be unitarily equivalent.


The basis of the adjoint representation space can be written in terms of 8 Young tableaux in which the first row is of length 2 and the second row of length 1. This method allows to write the matrix elements in terms of the matrix elements of a triple tensor product of the fundamental representation. However I won't elaborate an example here.


The matrix elements in the Cartan-Weyl basis can be computed in principle from the weight diagram.

share|cite|improve this answer
Thanks a lot. +1! – user7757 May 17 '12 at 6:02

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.