1
$\begingroup$

When Zee computes the Grassmann path integral in Section II.5 of his QFT book, he uses an algebraic step I can't follow. I follow this part:

\begin{align} \text{Tr}\ln\!\big[\gamma^5\big(-i\gamma^\mu\partial_\mu-m\big)\gamma^5\big]&=\text{Tr}\!\left\{\ln\gamma^5+\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)+\ln\gamma^5 \right\}\\[4pt] &=\text{Tr}\!\left\{\big(\ln\gamma^5+\ln\gamma^5\big)+\ln\!\big(-i\gamma^\mu\partial_\mu-m\big) \right\}\\[4pt] &=\text{Tr}\!\left\{\ln\!\big(\gamma^5\big)^{\!2}+\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)\right\}\\[4pt] &=\text{Tr}\!\left\{\ln1+\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)\right\}\\[4pt] &=\text{Tr} \ln\!\big(-i\gamma^\mu\partial_\mu-m\big) ~~. \end{align}

Then Zee writes

\begin{align} \text{Tr} \ln\!\big(-i\gamma^\mu\partial_\mu-m\big)&=\frac{1}{2}\text{Tr}\!\left\{\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)+\ln\!\big(i\gamma^\mu\partial_\mu-m\big)\right\}\\[4pt] &=\frac{1}{2}\text{Tr}\ln\!\big(\partial^2+m^2\big)~~, \end{align}

but I do not see how he got the complex conjugate on the first line. It appears to me that he did something like

\begin{align} \text{Tr} \ln\!\big(-i\gamma^\mu\partial_\mu-m\big) &=\text{Tr}\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)^{2\times\frac{1}{2}} \\[4pt] &=\frac{1}{2}\text{Tr} \ln\!\big(-i\gamma^\mu\partial_\mu-m\big)^{2} \\[4pt] &=\frac{1}{2}\text{Tr}\!\left\{\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)+\ln\!\big(-i\gamma^\mu\partial_\mu-m\big)\right\}~~, \end{align}

but I do not see where the complex conjugate of $(-i\gamma^\mu\partial_\mu-m)$ comes from.

$\endgroup$
6
  • $\begingroup$ The argument of any logarithm has to be > 0 so we take z*z $\endgroup$
    – joseph h
    Commented Nov 30, 2020 at 4:42
  • $\begingroup$ @Drjh in this case it doesn't even make sense to talk about > 0 because it's the logarithm of an operator $\endgroup$ Commented Nov 30, 2020 at 5:10
  • $\begingroup$ The arg of the log needs to be positive. $\endgroup$
    – joseph h
    Commented Nov 30, 2020 at 5:36
  • $\begingroup$ @Drjh How can you determine the arg of an operator? $\endgroup$ Commented Nov 30, 2020 at 9:30
  • 2
    $\begingroup$ Since the thing in the brackets of log(...) is an operator, I am unable to determine its sign. Can you please enlighten me on how you would determine the sign of the operator in the brackets? What is it that you think I am confused about? If you think I am confused about the answer to the question I asked, then you are exactly right and that should be obvious because if I was not confused, then I would not have posted the question. $\endgroup$ Commented Dec 1, 2020 at 15:03

1 Answer 1

2
$\begingroup$

The crux of the argument is proving that $\text{Tr}\ln(i\gamma^\mu\partial_\mu-m)$ = $\text{Tr}\ln(-i\gamma^\mu\partial_\mu-m)$. While it might be possible to prove this using the fact that the trace of the log of an operator is the logarithm of the product of its eigenvalues, I wasn't able to do this myself.

Start with $\text{Tr}\ln(i\gamma^\mu\partial_\mu-m)$, then using the cyclic property of the trace, its linearity, and the expansion of the logarithm of an operator, you get $$ \text{Tr}\ln(\mathbb{I}_4\ (i\gamma^\mu\partial_\mu-m)) \\ = \text{Tr}\ln((\gamma^5)^2 \ (i\gamma^\mu\partial_\mu-m)) \\ = \text{Tr}\ln(\gamma^5(i\gamma^\mu\partial_\mu-m)\gamma^5) $$ Then using the fact that $\{\gamma^5, \gamma^\mu\} = 0$, the $\gamma^5$'s act on the $\gamma^\mu$ by conjugation to yield $-\gamma^\mu$, while $m$ commutes with the $\gamma^5$, yielding $$ \text{Tr}\ln(-i\gamma^\mu\partial_\mu-m) $$

So, using the linearity of the trace again, we can break up $\text{Tr}\ln(i\gamma^\mu\partial_\mu-m)$ into $\frac12 \text{Tr} [ \ln(i\gamma^\mu\partial_\mu-m) + \ln(i\gamma^\mu\partial_\mu-m)]$ and rewrite the second one using the previous identity.

$\endgroup$
1
  • $\begingroup$ Can you really use the cyclic propery of the trace through the logarithm? That does not seem right. One could, perhaps, use $0 = \ln(1) = \ln(\gamma^5 \gamma^5) = \ln(\gamma^5) + \ln(\gamma^5)$, and add on both sides instead? $\endgroup$ Commented Sep 23, 2021 at 8:55

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.