I am reading up about Quantum Field theory and the integral of the following function pops up: $$\frac{1}{\sqrt{z^2 + m^2}}$$The details are actually explained in this question.

The book says that there are two branch cuts, $[-\infty, -im]$ and $[im, +\infty]$.


1) Since I can write $\frac{1}{\sqrt{z^2 + m^2}} = \frac{1}{\sqrt{z + im}\sqrt{z-im}}$, which branch cut corresponds to which $\sqrt{z\pm im}$ ?

2) Is the branch cut of $\sqrt{z - im}$, $[-\infty, im]$ or $[im, +\infty]$?

3) If it were possible to choose the branch cuts for $\sqrt{z - im}$ to be $[-\infty, im]$ , and for $\sqrt{z + im}$ to be $[-im, +\infty]$, how do I treat the interval $[-im, im]$?


I didn't look at the link post. I am going off the information giving that the function is $\frac{1}{\sqrt{z^2+m^2}}$.

I don't see why infinite branch cuts are used when we can use a single finite branch cut. If we let $z_1 = z - im = r_1e^{i\theta_1}$ and $z_2 = z + im=r_2e^{i\theta_2}$, we have $$ \frac{1}{\sqrt{z^2+m^2}} = \frac{1}{\sqrt{r_1r_2}\exp(i(\theta_1+\theta_2)/2)} $$ If we wind around each branch point separately, the function is multi-valued since we have $$ \frac{1}{\sqrt{r_1r_2}\exp(i(\theta_1+\theta_2+2\pi)/2)} = -\frac{1}{\sqrt{r_1r_2}\exp(i(\theta_1+\theta_2)/2)} $$ If we wind around both, we have $$ \frac{1}{\sqrt{r_1r_2}\exp(i(\theta_1+\theta_2+4\pi)/2)} = \frac{1}{\sqrt{r_1r_2}\exp(i(\theta_1+\theta_2)/2)} $$ so the function is single valued. Therefore, we can define the branch cut from $[-im, im]$

  • $\begingroup$ so why are they taking the branch cuts to be $\mathbb{C}\backslash([-im, im].?$ $\endgroup$ – SuperCiocia Dec 31 '14 at 10:05
  • $\begingroup$ @SuperCiocia I have no idea why they chose infinite branch cuts when a finite branch cut will suffice. It could be because they are Physicists not Mathematicians or they have a reason for it. I would suggest asking a Physicist why that was used. With the branch cut I suggested, we can use a dogbone contour. There a few post on how to do so as well. $\endgroup$ – dustin Dec 31 '14 at 16:33
  • $\begingroup$ I'm not sure this works since the argument isn't defined in the image of $\mathbb{C}\backslash [-im,im]$ under the function $z^2+m^2$. $\endgroup$ – hjhjhj57 Dec 31 '14 at 20:23
  • $\begingroup$ @SuperCiocia The more I think about it the more I believe this isn't right. First: It is possible to define a branch of the argument in a region iff it is possible to define a branch of the logarithm in it. Second: The image of $\mathbb{C}\backslash[-im,im]$ (imaginary interval) under $z^2+m^2$ is $\mathbb{C}\backslash[0,m]$ (real interval). Third: A branch of the logarithm (argument) is only defined in a simply connected region of $\dot{\mathbb{C}}$ (punctured plane) and $\mathbb{C}\backslash[0,m]$ isn't a simply connected region of the punctured plane. 1/2 $\endgroup$ – hjhjhj57 Jan 1 '15 at 4:48
  • $\begingroup$ I believe the problem is that from the beginning you assume there is a branch of the argument defined ($\theta_i=\arg z_i$). 2/2 $\endgroup$ – hjhjhj57 Jan 1 '15 at 4:50

A branch of the square root is only defined in a branch of the logarithm, i.e., in a simply connected region of $\dot{\mathbb{C}}$. So in fact you have an infinite amount of branches: any ray from either $im$ or $-im$ to infinity defines one.

In particular, observe that if $f_{\pm}(z)=z\pm im$, then $$ f_+(-\infty,-im] = (-\infty,0]\quad\text{and}\quad f_-[im,\infty) = [0,\infty), $$ (these intervals are imaginary!) so we can define two different branches of the logarithm (hence two branches of the square root), one in $f_+\left(\mathbb{C}\backslash (-\infty,-im]\right) = \mathbb{C}\backslash(-\infty,0]$ and the other in $f_-\left(\mathbb{C}\backslash[im,\infty)\right) = \mathbb{C}\backslash[0,\infty)$. In order to be able to define a common branch of the logarithm you need a branch that works in both regions (the principal branch works perfectly in this case).

Also observe that in this case the function $f:=\sqrt{f_-(z)f_+(z)}$ will be defined in: $$ \mathbb{C}\backslash(-\infty,-im]\cap\mathbb{C}\backslash[im,\infty). $$ This should've already answered (1) and (2).

For (3) observe that the points in the interval are just as any other point in the plane: $$ \begin{align} f_-(-im,im) = (-2im,0), \\ f_+(-im,im) = (0,2im). \end{align} $$ (these are imaginary intervals!) Which are contained in the region where the square root is defined.


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.