I'm new to contour integral involving branch point and stuck on this particular integration. Here is the problem:

$$\int_{\mathcal{C}}\log z\,\mathrm{d}z,$$

where $\mathcal{C}$ is a closed square contour connecting the points $-0.5+i$, $-1.5+i$, $-1.5-i$, $-0.5-i$. The branch cut of $\log z$ is chosen as the positive real axis. As far as I understand, the integrand within the contour is analytic so the result is zero. But Mathematica gives me $-2\pi i$.

Could anybody point it out for me where the problem is?

  • $\begingroup$ @mrf: It is a related problem. I did not know that he won't be able to see it. By the way, I can see it. $\endgroup$ – Mhenni Benghorbal Aug 3 '13 at 10:10

You are correct. Mathematica probably uses another branch (and treats the integral as an improper one).

  • $\begingroup$ Thanks @mrf. So do you mean this integral depends on which branch is chosen? Then it is indefinite, is it? $\endgroup$ – user88809 Aug 3 '13 at 10:00
  • $\begingroup$ Taking the principal branch, the cut goes through the contour, so it's not a "standard" i tegral. You had specified the branch though (which you always have to do when dealing with functions having branch points). Once ypu have chosen the branch, the integral has a value. Changing the branch, changes the function and possible the value pf the integral. $\endgroup$ – mrf Aug 3 '13 at 10:03
  • $\begingroup$ OK, I tried to select the negative real axis as branch cut and did obtain the same result as Mathematica did. Thanks again. $\endgroup$ – user88809 Aug 3 '13 at 14:17
  • $\begingroup$ maybe you can help me on an additional problem. Is there a complex version of integration-by-part? I saw someone used it but didn't find it in textbook. I tested integrals $\int_{\mathcal{C}} \frac{\log(x+1)}{x-2}\mathrm{d}x$ and $\int_{\mathcal{C}} \frac{\log(x-2)}{x+1}\mathrm{d}x$, where $\mathcal{C}$ encloses both -1 and 2. But the results do not match. Is it because they are not equal at the first place or I chose the wrong branch cut? $\endgroup$ – user88809 Aug 3 '13 at 14:42
  • $\begingroup$ @user88809 It's better if you ask a new question about that instead. $\endgroup$ – mrf Aug 3 '13 at 22:02

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.