I was given the a contour integral problem which I am rather unsure about.

First, I was asked to classify the singularities and required branch cuts for


I know that $z=0$ is a pole of order 3, and $z=1$ is a branch point, and thus for the keyhole contour, I need to perform a branch cut of $[1,\infty)$. I also know that the residue at $z=0$ is $1$, which I obtained using a Laurent expansion.

What confuses me is the required phases $e^{i\phi}$ necessary to obtain the integral

$$\int_{1}^{\infty} \frac{\ln(x-1)}{x^3} {\text{d}}x$$

from the contour integral

$$\oint_{\Gamma} f(z){\text{ d}}z,$$

where $\Gamma$ is the keyhole contour.

I know that for normal keyhole contour integrals with positive axis branch cut, for the straight portion of the contour approaching from $x=\infty$ to $x=1$, I need to include a phase of $e^{2i\pi}$. However, I also noticed that the function inside the logarithm is $x-1$ instead of $1-x$, so I suspect I need to include a phase $e^{i\pi}$ representing the minus one somewhere. I am confused as to how to account for them, the signs of the phase for the straight portions of the contour, and where I should even be placing them in the first place. All help is appreciated, thank you!


You'll have 3 integrals for your keyhole contour (I've gone CCW around the point $z=1$), $\int_{\infty}^1 f + \int_{|z-1|=\delta} f + \int_1^{\infty} f$.

The natural logarithm has its usual branch cut along the $-\Re$ axis. The argument (angle) of whatever the natural log is taken of (here, $1-z$) therefore has to stay in the branch: $-\pi < \arg(1-z) \le \pi$.

A point $z$ on the first contour ($\int_{\infty}^1$, running just above the $+\Re$ axis), maps to a $1-z$ in the 3rd quadrant (just below or in the $-\Im$ direction of the $-\Re$ axis--ie an argument of $-\pi+\epsilon$). To allow integration of the logarithm, we'd like this value of $1-z$ to be put in $re^{i\theta}$ form. For the angle $\theta$, it should be easy to see it's $-\pi+\epsilon$. For the modulus, we get $x-1$, where $x=\Re(z)$. This gives $(x-1)e^{-i(\pi-\epsilon)}$ for this first contour (where note we've been careful to keep the angle within the stated branch). Similar reasoning gives $(x-1)e^{i(\pi-\epsilon)}$ for the third contour (epsilon arbitrary). Hope this helps.


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.