MathOverflow is a question and answer site for professional mathematicians. 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

Thank you Cristi Stoica for your answer to the previous post of this question. Your hint is to the point I think. We should look at the requirements to construct the corresponding root system.

My apology to Yemon Choi, Will Jagi , Theo Johnson-Freyd and all other readers. My question was formulated extremely short without any context. It is the first time I am asking the MathOverflow community for help. And it is hard to judge how far to go in the description of the theory behind the question and the context of the question. So, let me try again. Cristi gave a good hint I think but the question is not answered yet.

The special unitary group SU(3) is a inherent component of the standard model of particle physics. It models the gauge field of the color charge property that is related the strong particle interaction. Generally for SU(n), matrices are used as a representation of the generators of its Lie algebra. Those matrices (L_i) are complex, traceless and antihermitian and their Lie bracket is the commutator.
The structure constants f_abc are defined by [L_a,L_b]=2i*f_abc*L_c The matrices in this set that are diagonal, form a basis for the Cartan subalgebra.

Gell-Mann proposed a set of eight 3D-matrices (L_i, i=1 to 8) to be used for SU(3) in the standard model. They are similar to the Pauli matrices in the SU(2)-case. Additionally Gell-Mann requires that all eight matrices are trace orthogonal, or that tr(L_a L_b)=2*delta(a,b) for a and b =1 to 8. This means that the 9-dimensional vectors, corresponding to the matrices, are orthogonal. The basis for the Cartan-algebra in this case is the pair L_3 and L_8. All entries in L_3 and L_8 are zero except L_3(1,1)=-L_3(2,2)=sqrt(3)*L_8(1,1)=sqrt(3)*L_8(2,2)=-.5*sqrt(3)*L_8(3,3)=1. The completely antisymmetric structure constants are: .5* f_123=f_147=f_165=f_246=f_257=f_345=f_376=.5, f_458=f_678=sqrt(3)*.5

In G-M’s particular choice of the matrices you see the appearance of the sqrt(3). I have the feeling that the expressions could be simpler (without sqrt(3)) if L_8 is replaced by (L_3+L_8), so that the entries of L_8 are all zero except L_8(2,2)= -L_8(3,3)=1. Of course the structure constants would change: F_123=f_678=f_458=f_345=1,f_147=f_165=f_246=f_257=f_376=f_128=.5 However we would get tr(L_3 L_8) =.5 instead of 0. So the question is: Why the Gell-Mann matrices in the SU(3)-model need to be trace orthogonal?

share|cite|improve this question
This post is historically inaccurate. Gell-Mann proposed these matrices not for use in the $SU(3)$ gauge theory of the Standard Model but rather for the approximate $SU(3)$ flavor symmetry of the quark model. And as mentioned in the answer below, there is nothing special about this particular set of matrices. It is just one basis and any other basis would work equally well. Why is this question appropriate to MO? – Jeff Harvey Feb 26 '12 at 15:07
up vote 2 down vote accepted

I'm no expert, and I haven't asked Professors Gell-Mann or Ne'eman, but with their choice of matrices, L_3 measures the familiar Heisenberg iso-spin quantum number, while L_8 measures the then-novel hypercharge. Mixing the operators would mix the quantum numbers.

share|cite|improve this answer
Thanks Art. This is the answer I needed I guess. Indeed, the orthogonality must guarantee the independence of the quantum numbers. – HAJV Feb 26 '12 at 7:26

It's just a choice of a basis. Compare it to an orthogonal vector basis. And please... try to write math in LaTeX :) (see the "How to write math" box on the right and below).

share|cite|improve this answer
Of course, it is a choice. Any idea why G-M preferred or required the one he chose? – HAJV Feb 24 '12 at 21:18
It's simply a choice which simplifies calculations. Choosing an orthonormal basis for a vector space simplifies any calculation involving an inner product, of which in computing Feynman diagrams there are a-plenty. (Is this really a question for MO?) – José Figueroa-O'Farrill Feb 26 '12 at 1:57

Your Answer


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

Not the answer you're looking for? Browse other questions tagged or ask your own question.