I am currently reading through Itzyskon and Zuber for my quantum field theory class, and I came across this regarding the unitary transformations of the Dirac bispinors in chapter 2. They show that the transformation from one frame to another: $\psi'(x') = S(\Lambda)\psi(x)$ has the property:

$$S(\Lambda) \gamma^{\mu} S(\Lambda^{-1}) = (\Lambda^{-1})^{\mu}_{\nu} \gamma^{\nu} $$

Which I understand. Then, if one were to make an infinitesimal transformation:

$$ \Lambda^{\mu}_{\nu} = \delta^{\mu}_{\nu} + \omega^{\mu}_{\nu} \hspace{1mm} , \hspace{1mm} (\Lambda^{-1})^{\mu}_{\nu} = \delta^{\mu}_{\nu} - \omega^{\mu}_{\nu} $$

Then we can say that the unitary transformations have the form:

$$S(\Lambda) = 1 - \frac{i}{4}\sigma_{\mu \nu} \omega^{\mu \nu} \hspace{1mm} , \hspace{1mm} S(\Lambda^{-1}) = 1 + \frac{i}{4}\sigma_{\mu \nu} \omega^{\mu \nu} $$

I am somewhat perplexed by the $\frac{i}{4}$ factors, but I figure that they're a result of a convention in anticipation of some result. My real problem comes in when attempting to use these with the first equation to derive the commutation relation $[\gamma^{\mu},\sigma_{\alpha \beta}]$. In the book they give this as the following:

$$ [\gamma^{\mu},\sigma_{\alpha \beta}] = 2i(\delta^{\mu}_{\alpha}\gamma_{\beta} - \delta^{\mu}_{\beta}\gamma_{\alpha}) $$

In an attempt to recover these relations, I did the following:

$$S(\Lambda) \gamma^{\mu} S(\Lambda^{-1}) = (\Lambda^{-1})^{\mu}_{\nu} \gamma^{\nu} \rightarrow $$

$$(1 - \frac{i}{4}\sigma_{\alpha \beta} \omega^{\alpha \beta}) \hspace{1mm} \gamma^{\mu} \hspace{1mm} (1 + \frac{i}{4}\sigma_{\alpha \beta} \omega^{\alpha \beta} = (\delta^{\mu}_{\nu} - \omega^{\mu}_{\nu}) \gamma^{\nu} $$

Multiplying the terms, discarding higher than order 1 terms, and subtracting the $\gamma^{\mu}$ from each side, we get:

$$ \frac{i}{4} \big( \gamma^{\mu} \sigma_{\alpha \beta} \omega^{\alpha \beta} - \sigma_{\alpha \beta} \omega^{\alpha \beta} \gamma^{\mu} \big) = - \omega^{\mu}_{\nu} \gamma^{\nu} $$

We can change the above equation to get:

$$ \frac{i}{4} \big( \gamma^{\mu} \sigma_{\alpha \beta} - \sigma_{\alpha \beta} \gamma^{\mu} \big) \omega^{\alpha \beta} = \delta^{\mu}_{\beta} \gamma_{\alpha} \omega^{\alpha \beta} $$

If we switch the indices $\alpha$ and $\beta$ and then use the antisymmetric property of $\omega^{\alpha \beta}$, then we get:

$$ \frac{i}{4} \big( \gamma^{\mu} \sigma_{\alpha \beta} - \sigma_{\alpha \beta} \gamma^{\mu} \big) \omega^{\alpha \beta} = -\delta^{\mu}_{\alpha} \gamma_{\beta} \omega^{\alpha \beta} $$

So if I add these two and do the appropriate algebra, then I arrive at the above commutation relation:

$$ [\gamma^{\mu},\sigma_{\alpha \beta}] = 2i(\delta^{\mu}_{\alpha}\gamma_{\beta} - \delta^{\mu}_{\beta}\gamma_{\alpha}) $$

However, this can all only be done if I assume that the $\omega^{\alpha \beta}$ commute with the $\gamma^{\mu}$. But, on the other hand, it is obvious from the result of this derivation that the $\sigma_{\alpha \beta}$ don't commute with $\gamma^{\mu}$. Why is this the case? Why should I assume that? It seems like a magical property that the $\omega^{\alpha \beta}$ have to get some relations that we like. Does it have to do with them being infinitesimal?

  • 1
    $\begingroup$ $\omega$ is a set of 6 numbers whereas $\sigma$ is a set of 6 matrices. $\endgroup$
    – Prahar
    Commented Feb 14, 2022 at 5:00

1 Answer 1


Look again at the defining equation for the $\sigma_{\mu\nu}$ and $\omega^{\mu\nu}$: $$ S = 1 -\frac{\mathrm{i}}{4}\sigma_{\mu\nu}\omega^{\mu\nu}$$ Here, $S$ is a matrix, and we're expanding it in term of some basis of matrices $\sigma_{\mu\nu}$ with coefficients $\omega^{\mu\nu}$. So the $\omega^{\mu\nu}$ are just real numbers, while the $\sigma_{\mu\nu}$ are matrices. Real numbers commute with everything, matrices don't.


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.