This question is regarding Noether's Theorem in general, but also in the application to an example. The example is:

Find the conserved current for the Lagrangian $$L=\bar{\psi}(\frac{i}{2}\gamma^{\mu}\partial_{\mu}-m)\psi.$$

Is my first step to find a transformation that leaves the action invariant? But isn't there more than one symmetry meaning this question has several answers?

From Peskin and Schroeder, it says we need to find

$$\partial_{\mu}j^{\mu}(x)=0 \; \text{ for } \; j^{\mu}(x)=\frac{\partial L}{\partial (\partial_{\mu}\psi)}\Delta\psi-J^{\mu}.$$

So I think I'm right in saying this $J^{\mu}$ is dependent on the transformation we're making? e.g. if it's just $\psi \rightarrow \psi +a$ then $J^{\mu}=0$.

The Lagrangian changes under the transformation of the field as

$$L \rightarrow L+\alpha \partial_{\mu}J^{\mu}.$$

But this doesn't help me because plugging $\psi \rightarrow \psi +\alpha \Delta \psi$ (an arbitrary transformation of the field $\psi$) into the Lagrangian at the top gives me (after no more than one line)

$$L \rightarrow L+\alpha \bar{\psi}(\frac{i}{2}\gamma^{\mu}\partial_{\mu}-m)\Delta \psi$$

and how do we use this to find $J^{\mu}$?

(Something I've assumed throughout all of this is that $\psi$ and $\bar{\psi}$ are treated completely separate and here we're only considering $\psi$. I hope I am right in doing this.)

I haven't really pin pointed a question here as my understanding breaks down at many points and when I think I finally understand it, I'm given a new Lagrangian and become stuck again. There must be some general procedure?


Firstly, you ask

isn't there more than one symmetry meaning this question has several answers?

Yes! In general, a given theory can have all sorts of symmetries, and each of these symmetries leads to its own conserved quantity via Noether's theorem.

As for what's going on with Noether's theorem and applying it in general, I'd like to strongly encourage you to read my answer here:

Noether's current expression in Peskin and Schroeder

If you still have questions after that, then let me know in the comments to this answer, and I'll gladly add an addendum to address your remaining confusions/questions.

  • 1
    $\begingroup$ Ok that's the best explanation of Noether's theorem I've seen before and it's helped me a lot. A couple of questions that I'm developing as I go through it using problems: First and foremost, step (1) says "Let a particular flow ϕ^(α,x) be given". In the problem above, I'm not given any flow (transformation of the field) and just asked to find the conserved current. So immediately I'm a bit lost. Do I just take an arbitrary flow and work from there? Could you also tell me what I should consider $\bar{\psi}$ as if we're varying $\psi$? $\endgroup$
    – Phibert
    Feb 23 '14 at 0:11
  • $\begingroup$ Incidentally, are you not missing a $\partial_{\mu}$ in the $\delta \mathcal{L}$ term in Lemma 1. of the answer you linked me to? $\endgroup$
    – Phibert
    Feb 23 '14 at 0:54
  • $\begingroup$ @user13223423 Thanks for the careful read; I was indeed missing a $\partial_\mu$! Recall that $\bar\psi = \psi^\dagger\gamma^0$, so any flow on $\psi$ induces a flow on $\bar\psi$ in the natural way; I can write more details about this if "natural way" isn't descriptive enough. Lastly, your problem statement is a bit unclear, but I would guess that it wants you to find the most general flow that transforms the field by both a spacetime transformation, and a target space transformation that is a symmetry. To find this, you can put in a general flow ansatz, and see what symmetry demands. $\endgroup$ Feb 23 '14 at 1:54
  • $\begingroup$ I've only recently come back to Noether's Theorem and I understand it quite well now thanks to you. One more thing though, for the transformation $\phi \rightarrow e^{i\alpha} \phi$, before we even start using Noether's theorem, to show the Lagrangian is invariant under this transformation, do we plug in $\Delta \phi = i \phi$ or $\Delta \phi = i \alpha \phi$ (because we only consider the $\textit{infinitesimal}$ transformation)? The former leaves my Lagrangian invariant, the latter gives $L \rightarrow \alpha^2 L$. (This is for the complex scalar field by the way). $\endgroup$
    – Phibert
    Apr 16 '14 at 18:12
  • $\begingroup$ @user13223423 Yeah so in this case, (using my notation from the linked post), we have $\hat\phi(\alpha, x) = e^{i\alpha}\phi(x) = \phi(x) + i\alpha\phi(x) + O(\alpha^2)$. Now, plug this right hand side into the Lagrangian and expand the result in $\alpha$. Recall that we only want to exhibit infinitesimal invariance which means that to first order in $\alpha$, the Lagrangian doesn't change. So you should find that $L \to L + O(\alpha^2)$ which is what I think you probably mean by $L\to \alpha^2L$. I'm guessing you erroneously only plugged in $i\alpha\phi$ instead of $\phi+i\alpha\phi$? $\endgroup$ Apr 16 '14 at 18:27

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.