3
$\begingroup$

We have the following beautiful result for Pauli $su(2)$ matrices

$$(\vec{\sigma}\cdot\vec{a})(\vec{\sigma}\cdot\vec{b}) = \mathbb{I} ~\vec{a}\cdot\vec{b} + i (\vec{a} \times \vec{b}) \cdot \vec{\sigma}.$$

Do we have a similar structure for Gell-Mann $su(3)$ matrices? Specifically, what would the following be

$$(\vec{\lambda}\cdot\vec{a})(\vec{\lambda}\cdot\vec{b}) = ~?$$

$\endgroup$

2 Answers 2

4
$\begingroup$

Yes, of course. The anticommutator for Gell-Mann matrices is somewhat more elaborate than for Pauli matrices, as there is also a d-coefficient, so splitting the $\lambda$-matrix bilinear into commutators and anticommutators yields $$ (\vec{\lambda}\cdot\vec{a})(\vec{\lambda}\cdot\vec{b}) = a^\mu \lambda^\mu ~b^\nu \lambda^\nu = a^\mu b^\nu \left (\tfrac{1}{2} [\lambda^\mu,\lambda^\nu] + \tfrac{1}{2} \{\lambda^\mu,\lambda^\nu \}\right )= \\ =a^\mu b^\nu ( if_{\mu \nu\kappa} \lambda^\kappa + d_{\mu\nu\kappa} \lambda^\kappa + \tfrac{2}{3} \delta_{\mu\nu} 1\!\!1) \\ =\tfrac{2}{3} 1\!\!1 a\cdot b +a^\mu b^\nu (if_{\mu \nu\kappa}+d_{\mu \nu\kappa})\lambda^\kappa, $$ the second term being analogous to the cross-product, except now it has both an antisymmetric and a symmetric piece.

Bonus point. Combining two octets will yield a reducible 64, $$8\otimes 8= 27\oplus\overline{10}\oplus10\oplus8\oplus8\oplus1 .$$ The symmetric singlet is explicit above (just as the SU(2) singlet for the Pauli matrices is), and the symmetric d term above reduces to one of the two 8 s, and not the 27.

The antisymmetric f term reduces to the other 8 and not the 10 and its conjugate.

The 8 hermitian matrices $m^\kappa _{\mu\nu}\equiv (if_{\mu\nu\kappa} +d_{\mu\nu\kappa} )$ are very sparse, much more so than their SU(2) angular momentum analogs. Their (imaginary) antisymmetric piece vanishes unless there are 1 or 3 indices from the set 2,5,7; and their (real) symmetric piece vanishes unless there is an even number of indices from the same set. For instance, $$ m^2= \begin{pmatrix} 0 & 0 & -i &0 &0 &0 &0 &0 \\ 0 & 0 & 0 &0 &0 &0 &0 &1/\sqrt{3} \\ i & 0 & 0 &0 &0 &0 &0 &0 \\ 0 & 0 & 0 &0 &0 &i/2 &-1/2 &0 \\ 0 & 0 & 0 &0 &0 &1/2 &i/2 &0 \\ 0 & 0 & 0 &-i/2 &1/2 &0 &0 &0 \\ 0 & 0 & 0 &-1/2 &-i/2 &0 &0 &0 \\ 0 & 1/\sqrt{3} & 0 &0 &0 &0 &0 &0 \end{pmatrix} , $$ and so on. Note this matrix is only 3/16 full!

$\endgroup$
2
  • $\begingroup$ Thanks, @Cosmas Zachos. Is that imaginary i just sitting with f_{\mu \nu k} or it should be outside the bracket, multiplying both f and d?. $\endgroup$
    – W. Voltera
    Commented Jul 4, 2018 at 2:37
  • $\begingroup$ No, just f. Remember d is symmetric under μν interchange but f is antisymmetric, so i is required from hermitian transposition. $\endgroup$ Commented Jul 4, 2018 at 7:52
1
$\begingroup$

(Unfortunately) there is no such generalization: the properties of the Pauli matrices that make such identities possible are closely tied to the $\mathbb{Z}_2\times \mathbb{Z_2}$ graded structure of the matrices (see 2. below).

However, as part of this negative answer I will point you to the following:

  1. Arvind, K. S. Mallesh and N. Mukunda, A generalized Pancharatnam geometric phase formula for three-level quantum systems, available from arxiv.
  2. Patera, J., and H. Zassenhaus. The Pauli matrices in n dimensions and finest gradings of simple Lie algebras of type $A_{n− 1}$, Journal of Mathematical Physics 29.3 (1988): 665-673 (pre-arxiv, behind paywall). See also: Patera, J. The four sets of additive quantum numbers of SU(3). Journal of mathematical physics 30.12 (1989): 2756-2762 (also behind paywall).

The first will give you geometric relations similar to the cross product of Pauli matrices, and also a $\star$ operation on the Gell-Mann matrices, but not what you want. The second will provide you with an alternate basis (of non-hermitian but unitary matrices) that nevertheless have some nice properties (such as $A^3\sim 1_{3\times 3}$, which generalize some of the properties of the Pauli’s.

(I wish someone can show my answer to be wrong as I’d love to know such a relation.)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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