The quantities $\delta_{ij}a_ib_j$ and $\epsilon_{ijk}a_ib_jc_k$ are invariant under the transformation of the $j=1$ (fundamental) representation of $SO(3)$. What would be the analogous expressions for the $j=1/2$ $SU(2)$ representation?

Attempt: The canonical definition of these groups are $$SO(3) = \left\{O \in GL(3, \mathbb{R}) : O^TO = 1, \det O = 1 \right\}\,\,\,\,\\SU(2) = \left\{U \in GL(2,\mathbb{C}) : U^{\dagger}U = 1, \det U = 1\right\}$$

Since the invariant in the rep $j=1$ in $SO(3)$, $\epsilon_{ijk}a_ib_jc_k$, only makes sense for $3D$ vectors, I am not really sure how to generalize this to $2D$. Can I say the invariant is the same in both cases, with the realization that I have simply a zero entry for the third component. (i.e just write $(a, b)$ in $2D$ is equivalent to $(a,b,0)$ in $3D$.



The crucial difference between $\mathrm{SO(}N\mathrm{)}$ and $\mathrm{SU(}N\mathrm{)}$ is that for $\mathrm{SU(}N\mathrm{)}$ one can raise and lower indices, while for $\mathrm{SO(}N\mathrm{)}$ this is irrelevant. See also these notes for more information.

For $\mathrm{SU(}N\mathrm{)}$ we have: $$ U^\dagger U = 1 $$ which in component form is: $$ U^j{}_i U_j{}^k = \delta^k{}_i \;\;\; \text{and} \;\;\; U_i{}^j U^k{}_j = \delta^k{}_i $$ Thus: $$ \varphi_i \varphi^i \to \varphi'_i \varphi'^i = U_i{}^j U^i{}_k \varphi_j \varphi^k = \delta^j{}_k \varphi_j \varphi^k=\varphi_i \varphi^i $$ and: $$ \delta^{i}{}_{j} \to \delta'^{i}{}_{j} = U^i{}_k U_j{}^l \delta^k{}_l = U^i{}_k U_j{}^k = \delta^i{}_j $$ Furthermore: \begin{equation} \begin{aligned} \varepsilon^{i_1 i_2 \dots i_N} \to \varepsilon'^{i_1 i_2 \dots i_N} & = U^{i_1}{}_{j_1} U^{i_2}{}_{j_2} \cdots U^{i_N}{}_{j_N} \varepsilon^{j_1 j_2 \dots j_N} \\& = \mathrm{det}(U)\varepsilon^{i_1 i_2 \dots i_N} \\& = \varepsilon^{i_1 i_2 \dots i_N} \end{aligned} \end{equation} and using a similar argument one can show that: \begin{equation} \varepsilon_{i_1 i_2 \dots i_N} \to \varepsilon'_{i_1 i_2 \dots i_N} = \varepsilon_{i_1 i_2 \dots i_N} \end{equation}

  • $\begingroup$ Do you mean to say that the invariant quantities in the $j=1/2$ rep are the same? (i.e the Kronecker delta and the Levi-Civita tensor are both invariant under any transformation), so the corresponding invariants for the $SU(2)$ rep is exactly $\delta_{ij}a_ib_j$ and $\epsilon_{ijk}a_ib_jc_k$? $\endgroup$ – CAF Mar 27 '14 at 8:53
  • $\begingroup$ In the line $...U^i_{\,k} U_j^{\,k} = \delta^i_{\, j}$, where are you using the fact that it is the condition $U^{\dagger}U = 1$ that holds rather than simply $U^TU=1$? $\endgroup$ – CAF Mar 27 '14 at 9:11
  • $\begingroup$ @CAF Look at the notes, page 14, I have attached (although it the conventions I've used in my message above are slightly different, you will get the idea). $\endgroup$ – Hunter Mar 27 '14 at 14:55
  • $\begingroup$ I did look at the notes, but it is not clear how the condition $U^{\dagger}U = 1$ is used. Seems to me it would work without this condition (i.e it would work with simply having $U^TU=1$) Also, the dimension of SU(2) is three, and given that the Kronecker and Levi-Civita tensors transform analogously under $U \in SU(2)$, are the invariants the same? Thanks for your help! $\endgroup$ – CAF Mar 27 '14 at 15:15
  • $\begingroup$ I am not sure about this last statement because an explicit representation of the generators in $SU(2)$ are scaled Pauli matrices which are $2x2$. So the matrix product $U a_i$ say, would only make sense if $a_i$ was 2D. And, in that case, I am not sure how to generalise the $SO(3)$ invariant $\epsilon_{ijk}a_ib_jc_k.$ Thanks for your help! $\endgroup$ – CAF Mar 27 '14 at 15:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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