This is claimed in the Scientific American article "Q&A: Lawrence Krauss on The Greatest Story Ever Told" published on March 21, 2017.

It says, “I could change the sign of each electric charge in nature locally. But I have to have a rule book.” What's the rule book? In this case, it’s the electromagnetic field.

However looking at explicit $U(1)$ gauge transformations $$\psi \to e^{ia(x)} \psi ,$$ $$ A\to A+ \partial_\mu a(x). $$ I fail to see how such an interpretation is possible

  • 2
    $\begingroup$ The electric charge is observable, i.e., gauge-invariant. I call BS. It's Krauss after all. $\endgroup$ Commented Apr 3, 2018 at 15:19
  • $\begingroup$ @AccidentalFourierTransform good point! But maybe a global phase shift can ne interpreted this way?! $\endgroup$
    – jak
    Commented Apr 6, 2018 at 8:21
  • $\begingroup$ Global phase shifts are generated by global symmetries, which are observable, i.e., non-gauge. $\endgroup$ Commented Apr 6, 2018 at 13:16

1 Answer 1


That really depends on what you mean by ''Gauge Transformation''. I will shortly give the mathematical description, which is also the one used in (almost any) physics-book (but in a less technical setting) and once we have this, it's clear that ''charge'' $\rightarrow$ -''charge'' cannot be a gauge transformation.

For the general setting, let us model electromagnetism not on Minkowski Space $\mathbb{R}^{1,3}$ but on $\mathbb{R}^{1,3} \times \mathrm{U}(1)$. (In the following I will use a little bit of gauge theory, but try to minimize the technical details as best as possible.)

Since $\psi$ does transform under a gauge transformation, which are a change of phase, it doesn't seem too off to view $\psi$ as a function which depends on $(x_{\mu},e^{i\theta})$. The latter parameter just explicilty encodes the phase at each point in spacetime.

Thus, if you change the phase, which is what we will call a gauge transformation we have to specify the way in which the field changes. The general setting would look like this:

$$ \psi (x_{\mu},e^{i \theta} \cdot e^{i\phi(x_{\mu})}) = \Lambda(e^{i \phi (x_{\mu})})\psi (x_{\mu}, e^{i \theta})$$

Where $\Lambda$ is a representation of $\mathrm{U}(1)$ acting on the vector space in which $\psi$ takes it's values (pre second quantisation).

Now let us assume the representation $\Lambda$ be irreducible and one dimensional (by which I mean, it is given by $\Lambda (e^{i \phi (x_{\mu})})\psi = e^{ \lambda (i \phi (x_{\mu}))} \cdot \psi $ for some function $\lambda$.

Then $\lambda$ is given by multiplication with an Integer $e \in \mathbb{Z}$, giving

$$ \psi (x_{\mu},e^{i \theta} \cdot e^{i\phi(x_{\mu})}) = e^{i \cdot e \cdot \phi (x_{\mu})}\psi (x_{\mu}, e^{i \theta}).$$

Finally, $e$ is the elementary (electromagnetic) charge of the particle described by $\psi$.

Thus, changing the charge would be a change of representation and not a gauge transformation, since gauge transformations do not touch the representation, hence, they don't touch the charge.

(going from $e$ to $-e$ would be done by passing from $\Lambda$ to $\Lambda$* which is the complex-conjugate representation)

I'd like to add, that the change of representation described above would correspond to a change of particle, for example passing from the $\pi^+$ to the $\pi^-$-meson.


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.