Given a 4-vector $p^\mu$ the Lorentz group acts on it in the vector representation: $$ \tag{1} p^\mu \longrightarrow (J_V[\Lambda])^\mu_{\,\,\nu} p^\nu\equiv \Lambda^\mu_{\,\,\nu} p^\nu. $$ However, I can always represent a 4-vector $p^\mu$ using left and right handed spinor indices, writing $$ \tag{2} p_{\alpha \dot{\alpha}} \equiv \sigma^\mu_{\alpha \dot{\alpha}} p_\mu.$$ So the question is: in what representation does the Lorentz group act on $p_{\alpha \dot{\alpha}}$?

There are a lot of questions about this and related topics around physics.se, with a lot of excellent answers, so let me clear up more specifically what I am asking for.

I already know that the answer to this question is that the transformation law is $$ \tag{3} p_{\alpha \dot{\alpha}} \rightarrow (A p A^\dagger)_{\alpha\dot{\alpha}}$$ with $A \in SL(2,\mathbb{C})$ (how is mentioned for example in this answer by Andrew McAddams). I also understand that $$ \tag{4} \mathfrak{so}(1,3) \cong \mathfrak{sl}(2,\mathbb{C}),$$ (which is explained for example here by Edward Hughes, here by joshphysics, here by Qmechanic).

So what is missing? Not much really. Two things:

  1. How do I obtain (3) and what is the specific form of $A$, i.e. its relation with the vector representation $\Lambda^\mu_{\,\,\nu}$? Defining the following $$ (\tilde p) \equiv p^\mu, \qquad \Lambda \equiv \Lambda^\mu_{\,\,\nu},$$ $$ \sigma \equiv \sigma_{\alpha \dot \alpha}, \qquad \hat p \equiv p_{\alpha \dot \alpha},$$ we can rewrite (1) and (2) in matrix form as $$ \tag{5} \hat p \equiv \sigma \tilde p \rightarrow \sigma \Lambda \tilde p = ( \sigma \Lambda \sigma^{-1}) \hat p,$$ however, this disagrees with (3) which I know to be right, so what is wrong with my reasoning?

  2. Why does the transformation law (3) has a form $$\tag{6} A \rightarrow U^{-1} A U,$$ while the usual vector transformation (1) has a form $V \rightarrow \Lambda V$? I suspect this comes from a similar reason to that explained here by Prahar, but I would appreciate a confirmation about this.


1 Answer 1


Your equation (3) comes from the following steps. First, a dotted index transforms in the complex conjugate representation of an undotted index. For a tensor product, each index transforms according to its own representation. Thus $$p_{a\dot a} \mapsto A_{ab} \bar A_{\dot a \dot b} p_{b\dot b} = A_{ab} p_{b\dot b} A^\dagger_{\dot b \dot a}$$ where on the left side of the equal sign we have elementwise complex conjugation. Putting the conjugate matrix on the right we have to take a transpose to get order of indices right.

In reasoning about (4) and (5) you are neglecting the transformation of $\sigma^\mu_{a\dot a}$. The correct description of the relation $p_{a\dot a} = \sigma^\mu_{a\dot a} p_\mu$ is that the 4-vector representation is equivalent to the $(\frac 1 2,0)\otimes(0,\frac 1 2)$ representation, by means of the linear transformation $$\sigma^\mu_{a \dot a} : V\to (\frac12,0)\otimes(0,\frac12)$$ meaning that $\sigma^\mu_{a\dot a}$ belongs to the space $(\frac 1 2,0)\otimes (0,\frac12) \otimes V^*$, on which the (double cover of the) Lorentz group acts. In fact, it acts like $$\sigma^\mu_{a\dot a} \mapsto A_{a\dot b} \sigma^\nu_{b\dot b} A^\dagger_{\dot b \dot a} (\Lambda^{-1})_\nu^\mu $$ so that $A^\mu_{a \dot a} p_\mu$ indeed has the correct transformation law.

  • 1
    $\begingroup$ Thank you very much, that definitely sorted it out. Just one thing: could you also provide some reference where I can find more on the subject (particulary where I can find an exposition of the transformation rules of $\sigma^\mu_{\,\,\alpha\dot \alpha}$ that you quoted)? $\endgroup$
    – glS
    Jan 12, 2015 at 15:38
  • $\begingroup$ That $\sigma^\mu_{a \dot a}$ transforms in that manner is really implicit in the indices it has, so I don't know if it's written out anywhere. My best guess is Penrose & Rindler, but I haven't checked. $\endgroup$ Jan 12, 2015 at 16:06

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.