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 am trying to reconcile the two ways of forming SU(2) singlets out of a pair of doublets.

Method (1): If $v=\begin{pmatrix}v^1\\ v^2\end{pmatrix}$ and $w=\begin{pmatrix}w^1\\ w^2\end{pmatrix}$ are two SU(2) doublets, then I can form a singlet by taking the antisymmetric combination:

$$(v\otimes w)_\text{singlet}=\epsilon_{ij}v^iw^j=v^1w^2-v^2w^1$$

Method (2): Using the same objects from method 1, I can form a singlet by simply taking the conjugate-transpose of one of them, say $v^*=\begin{pmatrix}v_1^*& v_2^*\end{pmatrix}$, and contracting them straight-up:

$$(v\otimes w)_\text{singlet}=(v^*)_i w^i=v_1^*w^1+v_2^*w^2.$$

So are these two different ways to form singlets? Is there a way to understand this from a more general point of view, say SU(3)? Also, in elementary quantum mechanics, why is it that when we form spin-singlets out of wavefunctions using method (1), and not using method (2)?

share|cite|improve this question
As a note to myself and others, $(v^*)_i$ and $\epsilon_{ij}v^j$ are two different objects. But, both transform as in the same way; namely as $\bar{\bf{2}}$. – QuantumDot Feb 28 '13 at 2:55
up vote 3 down vote accepted

OP's two methods are isomorphic. In general one is only interested in classifying representations modulo isomorphism. The point is that for the Lie group $SU(2)$, the spinor representation ${\bf 2}$ and the complex conjugate spinor representation $\bar{\bf 2}$ are equivalent representations ${\bf 2}\cong \bar{\bf 2}$. The equivalence is precisely given by multiplying with the epsilon symbol.

In contrast, the representation ${\bf 3}$ and complex conjugate representation $\bar{\bf 3}$ of the Lie group $SU(3)$ are not equivalent.

share|cite|improve this answer
Just to clarify, I have a spinor $\psi$ in representation $\bf{2}$. Then is $\psi^*=\epsilon\psi$ an equality? or is this taken to mean that only the LHS and RHS transform in the same way (i.e. $\bar{\bf 2}$). – QuantumDot Feb 27 '13 at 20:27
Well, if the metric to raise and lower spinor indices is $\delta_{ij}$, then you can get from an object that transform as a spinor, to an object that transforms as a complex conjugate spinor, by multiplying with the epsilon symbol. This works for any sign convention of epsilon. – Qmechanic Feb 27 '13 at 20:36
Oh this is something new! I thought that the metric necessarily had to be $\epsilon$. I can chose it to be $\delta$? Anyway, what if my metric to raise and lower indicies is in fact $\epsilon_{ij}$; what then would be the appropriate thing to multiply by? Thanks! – QuantumDot Feb 27 '13 at 21:56
The epsilon symbol is not a metric. A (real) metric $\eta_{ij}$ is by definition symmetric. For the Lie group $U(2)$, the metric is positive definite. A $2\times 2$ matrix $U\in U(2)$ satisfies $U^{\dagger}\eta U= \eta$. So a change in the choice of metric changes the possible choice of group element $U$. Any sane person would chose $\eta_{ij}=\delta_{ij}$. – Qmechanic Feb 27 '13 at 22:02

I think confusion arises because the metric tensor is hidden in $SU(2)$, so I'll work in $SU(p,q)$ and then specialize to $SU(2)$. The inequivalent defining reps of the general linear group $GL(m,C)$ carried on the four vector spaces $V_{m},\tilde{V}_{m}, V^{*}_{m}, \tilde{V}^{*}_{m}$ transform vectors as follows. $$ v'^{a}=\,[D(g)]^{a}_{\ \ b}v^{b}\\ v'_{a}=\,[D(g^{-T})]_{a}^{\ \ b}v_{b}\\ v'_{\bar{a}}=[D(g^{*})]_{\bar{a}}^{\ \ \bar{b}}v_{\bar{b}}\\ v'^{\bar{a}}=[D(g^{-\dagger})]^{\bar{a}}_{\ \ \bar{b}}v^{\bar{b}} $$ $SU(p,q)$ with $p+q=m$ is defined as the subgroup under which the Hermitian metric tensor $I^{\bar{a}}_{\ \ b}$ transforms trivially. The metric tensor can be used to change ordinary indices into barred indices. $$ v^{\bar{a}}=I^{\bar{a}}_{\ \ b}v^{b} $$ The metric tensor is invariant under $SU(p,q)$ so it commutes with the group matrices and hence the rep $D(g)$ is equivalent to the rep $D(g^{-\dagger})$. A physical quantity represented by states $v^{a}$ is the same kind of physical quantity as represented by states $v^{\bar{a}}$.

Now specialize to $m=2$ so the group is $SU(2)$ or $SU(1,1)$. The Levi-Civita tensor $\epsilon_{ab}$ can be used to change contravariant indices into covariant indices. $$ v_{a}=\epsilon_{ab}v^{b} $$ The Levi-Civita tensor is also invariant under the group and hence the rep $D(g^{-T})$ is equivalent to the rep $D(g)$.

The singlet $\epsilon_{ab}v^{a}w^{b}=v^{a}w_{a}$. These are two equivalent 1-d irreps carried on $V_{2}\otimes V_{2}$ and $V_{2}\otimes\tilde{V}_{2}$. They represent the same kind of physical quantity: a spin 0 state. So, if one has two spin half states $\psi^{a}$ and $\phi^{a}$, the spin zero state is $\epsilon_{ab}\psi^{a}\phi^{b}$ or, equivalently $\psi^{a}\phi_{a}$. In the second form one needs a state $\phi_{a}$ which is the same kind of physical quantity (a spin half) as $\phi_{\bar{a}}$. A $\phi_{\bar{a}}$ transforms as $D(g^{*})$ so if one has a $\phi^{a}$ then the natural way to get something transforming as $D(g^{*})$ is to use $(\phi^{a})^{*}$. In this way, a spin zero is also $\psi^{a}(\phi^{a})^{*}$.

share|cite|improve this answer
This answer together with @Qmechanic 's result has clarified the confusion! This answer, in particular is invaluable. Would you know where I could find info on the representation theory of $GL(m,C)$ as you mention in this answer? – QuantumDot Feb 28 '13 at 0:56
@QuantumDot : I learnt this from Wu-Ki Tung's book "Group Theory in Physics". – Stephen Blake Feb 28 '13 at 1:15

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.