In the Appendix of Peskin & Schroeder's "An Introduction to Quantum Field Theory" there is a list of integrals in Minkowski space. Of particular interest to me is the integral (A.44): $$ I(\Delta) = \int \frac{d^{d}\ell}{(2\pi)^d} \frac{1}{(\ell^{2} - \Delta)^{n}} = \frac{(-1)^{n} i }{(4\pi)^{d/2}} \frac{\Gamma(n-\tfrac{d}{2})}{\Gamma(n)} \left( \frac{1}{\Delta} \right)^{n-d/2} $$

$\Delta$ doesn't depend on $\ell$, and Peskin uses the metric $(+---)$. Also, $d$ is the dimension of the space we're looking at, obviously. Peskin derives this integral by Wick rotating to Euclidean space such that $\ell^{0} = i \ell^{0}_{\mathrm{E}}$, and $\ell^{2} = - \ell^{2}_{\mathrm{E}}$ (this is shown in more detail in Chapter 6.3).

(Context: I need to look at the above integral for $n=3$. Supposedly this diverges in $d=4$, which is what I am looking at)

I have two questions about this integral:

  • Because of the Wick rotation used, is the above integral exact or does it just work some of the time, or as an approximation? I always thought that it was exact, but recently I had a discussion with someone who said this is not true and a Wick rotation doesn't necessarily yield the exact result. I am confused and would like some clarification on this.

  • Secondly, Schroeder uses the $(+---)$ metric (aka the wrong metric for those East Coast fans like myself). This means that $\ell^{2} = (\ell^{0})^2 - ||\boldsymbol{\ell}||^{2}$. I am wondering, does the above integral change in its result if we use the metric $(-+++)$? Mostly I am anticipating that an extra minus sign floats around somewhere since now we'd have $\ell^{2} = -(\ell^{0})^2 + ||\boldsymbol{\ell}^{2}||$.

  • 1
    $\begingroup$ Be wary when using equations from P&S as editions have errors detailed on their site. I spent hours questioning a result only to find it was because of an error in P&S. In this case, the integral is perfectly fine. $\endgroup$
    – JamalS
    Mar 10 '17 at 23:00
  • $\begingroup$ Really appreciate the heads up! I never knew that P&S has a website with errata, I'll have to check that out $\endgroup$ Mar 11 '17 at 1:51
  1. The Euclidean integral $$ I_E(\Delta) ~:=~\int \! \frac{d^d\ell_E}{(2\pi)^d} \frac{1}{(\Delta+\ell_E^2)^n}~\stackrel{(A.44)}{=}~\frac{1}{(4\pi)^{d/2}}\frac{\Gamma\left(n-\frac{d}{2}\right)}{\Gamma(n)} \Delta^{\frac{d}{2}-n}\tag{E}$$ is integrable iff $$n>\frac{d}{2}.$$ See also the superficial degree of divergence.

  2. For $n<\frac{d}{2}$ the Euclidean integral $I_E(\Delta)$ is declared in dimensional regularization to be equal to the rhs. of eq. (E) via analytic continuation.

  3. The Minkowski integral in the $(\mp,\pm,\ldots,\pm)$ convention is defined via Wick rotation $$\ell_M^0~=~i\ell_E^0 \tag{A.43}$$ to be given by the Euclidean integral $$I_M(\Delta) ~:=~\int \! \frac{d^d\ell_M}{(2\pi)^d} \frac{1}{(\Delta\pm\ell_M^2)^n} ~:=~i I_E(\Delta),\tag{M}$$ respectively.

  • $\begingroup$ In 3, you're saying that the integral has the same result regardless of which $(\pm,\mp,\ldots,\mp)$ is used? $\endgroup$ Mar 11 '17 at 1:50
  • $\begingroup$ Yes, for the way that I've defined it. $\endgroup$
    – Qmechanic
    Mar 11 '17 at 6:23

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.