In my text Schwartz writes: since $\bar{\psi} \gamma^\mu \psi$ transforms like a 4-vector, we can deduce that

$$\Lambda_s^{-1} \gamma^\mu \Lambda_s = (\Lambda_V)^{\mu \nu} \gamma^\nu, \tag{10.78}$$

where $\Lambda_s$ and $\Lambda_V$ are the Lorentz transformation acting on Dirac spinors and 4-vectors, respectively.

In this argument, it seems that the equality

$$\overline{\Lambda_s \psi} = \bar{\psi} \Lambda_s^{-1}$$

is implied. I am having trouble seeing why this is true. Expanding this out, we need

$$\psi^\dagger \Lambda_s^\dagger \gamma^0 = \psi^\dagger \gamma^0 \Lambda_s^{-1}$$

so this is equivalent to showing that

$$\Lambda_s^\dagger \gamma^0 \Lambda_s = \gamma^0.$$

In my attempt I used the Weyl representation, although I assume it must be true in general too. In the Weyl representation we have

$$\Lambda_s = \exp(i \theta_{\mu \nu} S^{\mu \nu}), \tag{10.71}$$ $$S^{ij} = \frac{1}{2} \epsilon_{ijk} \begin{bmatrix}\sigma_k & 0 \\ 0 & \sigma_k\end{bmatrix}, \; S^{0i} = -\frac{i}{2} \begin{bmatrix}\sigma_i & 0 \\ 0 & -\sigma_i\end{bmatrix}. \tag{10.72}$$

Now I can verify the equation for certain $\Lambda_s$, namely those generated by a single $S^{\mu \nu}$. For example with $S^{01}$,

$$\Lambda_s = \exp(2i \beta S^{01}) = \exp \begin{bmatrix} 0 & \beta & 0 & 0 \\ \beta & 0 & 0 & 0 \\ 0 & 0 & 0 & -\beta \\ 0 & 0 & -\beta & 0 \end{bmatrix} = \begin{bmatrix} \cosh\beta & \sinh\beta & 0 & 0 \\ \sinh\beta & \cosh\beta & 0 & 0 \\ 0 & 0 & \cosh\beta & -\sinh\beta \\ 0 & 0 & -\sinh\beta & \cosh\beta \end{bmatrix}$$

and the desired $\Lambda_s^\dagger \gamma^0 \Lambda_s = \gamma^0$ can be easily verified. Similarly for any other specific $S^{\mu \nu}$. But this isn't enough to show it in general. The form of $\exp(i \theta_{\mu \nu} S^{\mu \nu})$ gets complicated for general $\theta_{\mu \nu}$. And we can't decompose it like $$\exp(i \theta_{01} S^{01}) \exp(i \theta_{02} S^{02}) \cdots$$ since the $S^{\mu \nu}$ components don't commute in general. So I'm stuck with trying to figure out why $\Lambda_s^\dagger \gamma^0 \Lambda_s = \gamma^0$ for general $\Lambda_s$.


Why $\Lambda_s^\dagger \gamma^0 \Lambda_s = \gamma^0$ for general $\Lambda_s$?

Expanding infinitesimal form of $\Lambda_s = \exp{\left(i \theta_{\mu\nu}S^{\mu\nu}\right)}$ and keeping only first order terms gives this:

\begin{align} \Lambda_s^\dagger \gamma^0 \Lambda_s &= \left(1-i \theta_{\mu\nu}S^{\mu\nu}\right) \gamma^0 \left(1+i \theta_{\mu\nu}S^{\mu\nu}\right) \\ & \approx \gamma^0 - {i \gamma^0 \theta_{\mu\nu}S^{\mu\nu}} + i \gamma^ 0 \theta_{\mu\nu}S^{\mu\nu} \\ & = \gamma^0 \end{align}

  • $\begingroup$ This works for infinitesimal Lorentz transformations but not in general, right? $\endgroup$ – jcai May 25 '20 at 8:51

I found that part of Schwartz's derivations on the next page explains this retroactively. As preliminaries, note that $(\gamma^0)^2 = \mathbf{1}$ and $\gamma^{0\dagger} = \gamma^0$, as well as

$$\gamma^0 (S^{\mu\nu})^\dagger \gamma^0 = S^{\mu \nu}.$$

These can be derived from the Dirac algebra or checked explicitly with the Weyl representation. Then we can derive,

$$(\gamma^0 \Lambda_s \gamma^0)^\dagger = \gamma^0 \exp(i\theta_{\mu \nu} S^{\mu \nu})^\dagger \gamma^0 = \exp(-i\theta_{\mu \nu}\gamma^0 S^{\mu\nu\dagger}\gamma^0) = \exp(-i\theta_{\mu\nu}S^{\mu\nu}) = \Lambda_s^{-1}.$$

The only nontrivial step is the use of the identity

$$\gamma^0 \exp(M) \gamma^0 = \exp(\gamma^0 M \gamma^0)$$

for any matrix $M$. We can check this by Taylor expanding:

$$\exp(\gamma^0 M \gamma^0) = \sum_{n=0}^\infty \frac{1}{n!} (\gamma^0 M \gamma^0)^n = \sum_{n=0}^\infty \frac{1}{n!} \gamma^0 M^n \gamma^0 = \gamma^0 \exp(M) \gamma^0,$$

where we got $(\gamma^0 M \gamma^0)^n = \gamma^0 M^n \gamma^0$ by repeatedly contracting $\gamma^0 \gamma^0 = \mathbb{1}$.

To nudge the result to the desired form, we have,

$$\gamma^0 \Lambda_s^\dagger \gamma^0 = (\gamma^0 \Lambda_s \gamma^0)^\dagger = \Lambda_s^{-1}$$

$$\Lambda_s^\dagger \gamma^0 \Lambda_s = (\gamma^0)^{-1} = \gamma^0,$$

as desired.


In general, $\overline{AB}=\bar B \bar A$. Hence $\overline{\Lambda_{s}\psi}=\bar \psi \Lambda ^{-1}_{s}$ since $\Lambda_{s}$ is real and $\Lambda^{t}_{s}=\Lambda^{-1}_{s}.$

  • $\begingroup$ I don't understand what you mean. The $\bar{\psi} \equiv \psi^\dagger \gamma^0$ notation only applies to Dirac spinors. Writing $\overline{\Lambda_s \psi} = \bar{\psi} \bar{\Lambda}_s$ would be nonsensical because $\bar{\Lambda}_s$ isn't well-defined. Even if you did define $\bar{\Lambda}_s = \Lambda_s^\dagger \gamma^0$, the equation $\overline{\Lambda_s \psi} = \bar{\psi} \bar{\Lambda}_s$ would be incorrect. $\endgroup$ – jcai May 25 '20 at 3:50
  • $\begingroup$ Also $\Lambda_s$ isn't always real. $\endgroup$ – jcai May 25 '20 at 3:52
  • $\begingroup$ $\overline{\Lambda_s \psi} = \bar{\psi} \Lambda_s^{-1}$ was in the title of the post - and you asked it again in the body. You're only entitled to one question per post and checking your calculations is off topic. And $\Lambda_s$ is the Lorentz transformation which is real - the exponential form is a complex exponential but it's not denoted by $\Lambda_s$. $\endgroup$ – Cinaed Simson May 25 '20 at 19:59

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.