# Proof of Goldstone's theorem

I am reading Matthew Schwartz's wonderful QFT book, but I have stumbled in my attempts to understand his rather brief proof of Goldstone's theorem. The preliminary information is that $$HQ|\Omega\rangle= E_0Q|\Omega\rangle,\tag{28.7}$$ where $Q$ is the conserved charge corresponding to the spontaneously broken symmetry $$Q\equiv\int d^3x\;j^0(x).\tag{28.5}$$ This is fine. However, from this, he claims that one can create states $$|\pi(\vec{\bf{p}})\rangle\equiv\frac{-2i}{F}\int d^3x e^{-i\,\vec{\bf{p}}\cdot\vec{\bf{x}}}j^0(x)|\Omega\rangle\tag{28.8}$$ that have energy $E(\vec{\bf{p}}) + E_0$. If this is true, then it is clear that the state must obey a massless dispersion relation. However, I do not understand how one determines the energy to be $E(\vec{\bf{p}}) +E_0$.

I have seen a similar proof on p. 228 in Zee's book, but there he shows in footnote 3 that $$\hat{P}^i|\pi(p)\rangle = p^i|\pi(p)\rangle,$$ which is not quite the same although it has the same implication. Any ideas?

• Have you try to use the decomposition of $j_0$ in terms of creation/annihilation operators ? – Adam Apr 3 '17 at 8:32
• Well, I guess I could do it explicitly in terms of the operators for a simple complex scalar field, but he claims this is true in general. – Nick Murphy Apr 3 '17 at 9:41
• – Cosmas Zachos Nov 21 '17 at 15:40

## 2 Answers

I think you got the sign of the exponential in Schwartz's (28.8) wrong here, which I reverse, but I am cavalier with signs and factors, as I assume you just want to see the principle (the forest, not the trees).

The key point is that the SSB current is basically always of the form $j_\mu(x)\propto F \partial_\mu \pi (x) +...$ where the omitted terms are of higher order in the fields (and so do not contribute to the particle state defined); and, of course, group indices have been omitted as well.

This is codified in the appellation "Nambu-Goldstone nonlinear realization of the symmetry". This is the only way to have the v.e.v. of the transform of this Goldstone field not vanish, while the v.e.v.s of all the fields themselves vanish--after shift redefinitions. (See this answer .) It is illustrated later on in (28.13).

The Fourier transform of the Goldstone field π is $$\pi(\vec{\bf{p}}) = \frac{-2i}{F}\int d^3x ~e^{i\,\vec{\bf{p}}\cdot\vec{\bf{x}}}j^0(x),$$
with $$[H,\pi(\vec{\bf{p}})] = E(\vec{\bf{p}}) ~ \pi(\vec{\bf{p}}) .$$ The SSB charge Q is essentially the space integral of the canonical momentum of π, so that $\langle \Omega |[Q,\int d^3 p ~ \pi(\vec{\bf{p}})]|\Omega\rangle \propto F\neq 0$ , cf. his (28.9).

As a consequence, decompressing your text's one liner, $$H|\pi(\vec{\bf{p}})\rangle= ( [H,\pi(\vec{\bf{p}})] + \pi(\vec{\bf{p}}) H ) ~|\Omega\rangle = ( E(\vec{\bf{p}}) + E_0 )\pi(\vec{\bf{p}})|\Omega\rangle= ( E(\vec{\bf{p}}) + E_0 )|\pi(\vec{\bf{p}})\rangle ,$$ so $E(0)\to 0$ .

Still, for the more conventional general proof of the theorem, bypassing your perceptual snag, consider Kibble's summary.

• How would you prove that $[H,\pi(\vec p)]=E(\vec p)\pi(\vec p)$? – AccidentalFourierTransform Nov 21 '17 at 17:57
• This is the standard 2nd quantization expression... I'm sure that text covers it in previous chapters... – Cosmas Zachos Nov 21 '17 at 18:01
• It would be a standard 2nd qn. expression if $\pi(\vec p)$ were a standard creation operator. But here $\pi(\vec p)$ is defined from $j^0$, so it's far from clear whether it satisfies standard expressions or not. In principle, it should be possible to prove that $[H,\pi]=E\pi$ from the properties of $j^\mu$ alone, that is, from $\partial\cdot j=0$ and perhaps something else that I am missing right now. – AccidentalFourierTransform Nov 21 '17 at 18:06
• That's the very point. For a N-G realization, the current is the gradient of a standard field operator times the decay constant F--this must be so. – Cosmas Zachos Nov 21 '17 at 19:14
• Thanks @CosmasZachos! However, I am a bit confused in the same way as @AccidentalFourierTransform. I do not see how this commutation relation follows. – Nick Murphy Nov 24 '17 at 9:53

I guess this follows from the Heisenberg equation of motion for $\pi$: $$i\partial_t \pi = [\pi, H]$$ Calculating it, I found this works if

1) his sign convention is such that the coefficient of E in the exponent is -i so that $i\partial_t$ brings down a factor -E

2) $H|\Omega\rangle =0$

3)Perhaps I'm mistaken, but it looks as if your equation is missing a factor of $Q$ so that $$HQ|\Omega\rangle = E_0 Q|\Omega\rangle$$ -otherwise I have a hard time getting energy eigenstates here.

• yea thanks you are right about the typo. I actually also screwed up the -2i/F factor which I changed now in the question. This explains the messed up sign with the energy. Thanks a lot for the response! However, I don't think H|omega> =0, it should equal E_0|omega>. – Nick Murphy Apr 7 '17 at 15:16
• So $E_0$ is the vacuum energy... What's F? – rwold Apr 7 '17 at 15:24
• Yea exactly E_0 is the vacuum energy. I think F is just a constant with mass dimension 1 to give everything the right dimensions. The current has dimension 2, and I think you want you pi field to have dimension 1 because it should be a scalar. – Nick Murphy Apr 7 '17 at 16:10
• Hey @rwold, looking again at this again, I cannot see how this all works out. Any chance you could be a bit more explicit? Thanks again! – Nick Murphy Oct 16 '17 at 18:52