In the book Quantum field theory by Mark Srednicki (chapter 93, pages 575-576) in order to compute winding number, $n$, in a 4-dimensional space with coordinates $x = (x_1, x_2, x_3, x_4)$ and such that

$$\hat{x} = (\sin\chi \sin\psi \cos\phi, \sin\chi \sin\psi \sin\phi, \sin\chi \cos\psi, \cos\chi), \quad \sum_\mu \hat{x}_\mu \hat{x}_\mu = 1$$

$n$ is given by

$$ n = -\frac{1}{24\pi^2}\int_0^\pi d\chi\int_0^\pi d\psi \int_0^{2\pi} d\phi\ \epsilon^{\alpha\beta\gamma}tr\{(U\partial_\alpha U^\dagger) (U\partial_\beta U^\dagger) (U\partial_\gamma U^\dagger)\}, \quad \epsilon^{\chi\psi\phi} = +1 $$

Where $U$ is only dependent on $\hat{x}$, belongs to $SU(2)$ and has the winding number $n$ associated. $tr$ represents the trace.

But suddenly Srednicki says that you can write $n$ as an integral over the surface of this 4-dimensional space of the form

$$ n = \frac{1}{24\pi^2}\int dS_\mu\ \epsilon^{\mu\nu\sigma\tau}tr\{(U\partial_\nu U^\dagger) (U\partial_\sigma U^\dagger) (U\partial_\tau U^\dagger)\}, \quad \partial_\nu = \partial/\partial x^\nu\ {\rm and\ so\ on} $$

I don't understand how you can go from one expression of $n$ to the other.

  • $\begingroup$ I asked this in Math's site but I couldn't find any answer, so taking into account that there is Physics related to Winding number and my interest on it comes from theta vacua and instantons, I have cross-posted it $\endgroup$
    – Vicky
    May 30, 2019 at 5:54
  • $\begingroup$ Crossposted from math.stackexchange.com/q/3179213/11127 $\endgroup$
    – Qmechanic
    May 30, 2019 at 5:59

1 Answer 1


I don't have Sredniki to hand, but it seems to be that your "four dimensional" space of the $x^\mu$ at the beginning of the question is really the three dimensional space forming the surface of a three sphere and the integral is over this three-dimensional space. Your $U\in {\rm SU}(2)$ is also parametrized by a three-sphere, so you have a winding number of 3-sphere about 3-sphere.

In the second equation, however the $x^\mu$ are clearly intended to be an actual four dimensional space and one is choosing some general (no longer necessarily a sphere) three dimensional surface embedded in that space. If you choose a surface $x^0=0$ for example then $\epsilon^{0abc}= \epsilon^{abc}$ and you are back to the original integral. Even if the new surface is not compact, you will still get an integer value for $n$ if $U\to {\rm identity}$ at infinity.


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.