In the literature on supersymmetry, the following spinor summation convention is often used (eg. Wess & Bagger's book Supersymmetry and Supergravity) $$ \psi\chi = \psi^{\alpha}\chi_{\alpha} = -\psi_{\alpha}\chi^{\alpha} = \chi^{\alpha}\psi_{\alpha} = \chi\psi $$ where the spinors are treated as Grassmann numbers and anticommute with each other. However, in some articles like arXiv:hep-th/0312171, the angle spinor bracket is defined as $$ \langle\lambda_1\lambda_2\rangle = \lambda_1^{\alpha}\lambda_{2\alpha} = \varepsilon_{\alpha\beta} \lambda_1^{\alpha}\lambda_2^{\beta} = -\varepsilon_{\beta\alpha} \lambda_2^{\beta}\lambda_1^{\alpha} = -\lambda_2^{\beta}\lambda_{1\beta} = -\langle\lambda_2\lambda_1\rangle $$ Here the spinors are treated as c-number objects and commute with each other. Why can we use such two different conventions? Are they contradictory?

  • 3
    $\begingroup$ The spinors in Wess & Bagger corresponds to fermionic fields and hence have anti-commuting components. The spinors in Witten's papers, on the other hand, are part of the twistor formalism and encode the momentum in a convenient way. They do not represent fields and are not fermionic, so their components still commute. $\endgroup$ – Olof Dec 8 '14 at 11:52

Lorentz spinors appear as irreducible representations of the group SL(2,C). Elements of the group are 2x2 matrices with complex entries and unity determinant. A Lorentz spinor is a two component vector $\psi^{A},\chi^{A}\in V_{2}$ with $A=1,2$. The Levi-Civita tensor $\epsilon_{AB}$ is an invariant tensor under SL(2,C). This means that if we have an irrep $\psi^{A}$ of SL(2,C) we can transform it with the Levi-Civita tensor and this will give an equivalent irrep. So, the covariant vectors $\psi_{A}\in \tilde{V}_{2}$ made as $\psi_{A}=\psi^{B}\epsilon_{BA}$ are equivalent to the contravariant vectors $\psi^{A}$; it doesn't matter whether we use $\psi^{A}$ or $\psi_{A}$ because both quantities represent the same physical thing.

The situation is the same as in Minkowski spacetime when we use $x^{\mu}$ or $x_{\mu}=\eta_{\mu\lambda}x^{\lambda}$. Covariant and contravariant vectors in Minkowski spacetime are equivalent irreps of the Lorentz group $O(1,3)$ because the metric $\eta_{\mu\lambda}$ is an invariant tensor.

The only problem with using the Levi-Civita tensor to lower a spinor index is that it is antisymmetric so that it matters which index is summed. I've decided to lower spinor indices as $\psi_{A}=\psi^{B}\epsilon_{BA}$ and so, for consistency, I've got to stick to this convention and not be tempted to use $\psi_{A}=\epsilon_{AB}\psi^{B}$.

Having picked a lowering convention, I'm forced to raise a spinor index with $\psi^{A}=\epsilon^{AB}\psi_{B}$ because then both operations are consistent. \begin{equation} \psi_{A}=\psi^{B}\epsilon_{BA}=\epsilon^{BC}\psi_{C}\epsilon_{BA}=\delta^{C}_{A}\psi_{C}=\psi_{A} \end{equation}

Now we can make a SL(2,C) scalar $\psi_{A}\chi^{A}=\chi^{A}\psi_{A}$. The order of the vectors does not matter because the components $\psi^{1},\psi^{2}$ are just complex numbers. The following bit of index gymnastics recovers the property in the second equation in soliton's question. \begin{equation} \psi_{A}\chi^{A}=\psi^{B}\epsilon_{BA}\chi^{A}=-\psi^{B}\chi^{A}\epsilon_{AB}=-\psi^{A}\chi_{A} \end{equation}

Everything so far has been classical. When we go over to quantum theory, the spinors are promoted to operators $\psi^{A}\rightarrow \hat{\psi}^{A}$. These operators represent fermions: as operators, they have to obey anticommutation relations, in this case $[\hat{\psi}^{A},\hat{\chi}^{B}]_{+}=0$. So, repeating the last calculation with operators, \begin{equation} \hat{\psi}_{A}\hat{\chi}^{A}=\hat{\psi}^{B}\epsilon_{BA}\hat{\chi}^{A}=-\hat{\psi}^{B}\hat{\chi}^{A}\epsilon_{AB}=-\hat{\psi}^{A}\hat{\chi}_{A}=+\hat{\chi}_{A}\hat{\psi}^{A} \end{equation} recovers the first equation in soliton's question.

I have to own up to the fact that I've not yet studied supersymmetry, but I think this is what must be going on based on general principles.

  • $\begingroup$ If I just study the representations of $SL(2,\mathbb{C})$, without involving any quantum theory, do I still need to require $\psi$ to be anticommuting? $\endgroup$ – Nahc Jan 22 '17 at 21:28

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.