1

I need help with integrating the following: $$\int_0^\infty \frac{x^\lambda}{x+1} \;dx \qquad \text{for }-1<\lambda<0$$

There is also a hint to place the branch cut of the integrand along the positive real axis. I'm not entirely sure how to use this hint, because if we place the branch cut along the real axis, then I can't perform contour integration involving the real axis...

I know at least this much:

  • There is a pole at $z=-1$.
  • The branch cut should start from the origin, and stretch out along the real axis to $+\infty$.

I suspect that I should be taking a path, just above the real axis, encircling the branch point at $z=0$, and perhaps a path back along just below the real axis. But once, again, I am not entirely sure whether that yields me the integral of interest. Any help to get me started will be appreciated.

Troy
  • 561
ksgj1
  • 175
  • 3
    http://math.stackexchange.com/questions/1968756/calculating-int-0-infty-frac1x-alpha-frac11x-rm-dx/1968759#1968759 – Ron Gordon Nov 29 '16 at 13:46
  • 2
    @RonGordon Hi, sorry I'm new to this. How do I search for duplicates, especially if the question is purely symbols, integral signs etc.? – ksgj1 Nov 29 '16 at 13:51
  • 2
    Well...you hit upon the biggest problem here in M.SE. I like to think that at some point, we will have serious capability to do searches through LaTeX/MathJax, but right now it is either too clunky or simply nonexistent. At this point, unless it is screaming obvious, I would just post your question and let others point out its duplicity if that is the case. – Ron Gordon Nov 29 '16 at 13:53

1 Answers1

2

When you have integrals of that form, you have a branch point and you have to use the well known Residues formula with the Cauchy Principal Value:

$$\mathcal{P} \int_0^{+\infty} z^{\lambda}R(z)\ \text{d}z = \frac{2\pi i}{1 - e^{2i\lambda \pi}} \sum_k\ \text{Res}[f(z), z_k]$$

Where $f(z) = z^{\lambda}R(z)$ and $R(z) = \frac{P(z)}{Q(z)}$ a rational function with certain conditions.

In your case, there is a simple pole at $z = -1$, hence

$$\text{Res} = \lim_{z\to 1} (z-1)f(z) = (-1)^\lambda = e^{i\pi\lambda}$$

hence

$$I = \frac{2\pi i}{1 - e^{2i\lambda \pi}} e^{i\pi\lambda}$$

Manipulate a bit

$$\frac{2\pi i}{-2i\sin(\lambda\pi)} = -\frac{\pi}{\sin(\lambda\pi)}$$

And with easy trigonometry, you can write it as

$$-\pi\ \text{cosec}(\lambda\pi)$$

Enrico M.
  • 26,114
  • why you use the principal value? the singularity in the question is clearlt integrable – tired Nov 29 '16 at 13:44
  • @tired whoops I just wanted to write the general formulation lol, it's obviously useless xD – Enrico M. Nov 29 '16 at 13:45
  • 1
    @tired: there's nothing technically wrong with having a principal value integral there....what bothers me is that the formula is plucked from nowhere and is called "well-known." (Maybe true to some, but not to the OP.) – Ron Gordon Nov 29 '16 at 13:59
  • @RonGordon it is a well known formula from residues calculus! You better than almost everyone else shall know this. – Enrico M. Nov 29 '16 at 14:00
  • 1
    @RonGordon...i know it is just not needed here and makes the problem (for the inexperienced reader) just unnecessarcily more complicated then it is. And thumbs up for your second point – tired Nov 29 '16 at 14:01
  • 1
    @AlanTuring: it's not about me; it's about the OP. He wants to be shown how to make use of the branch cut to come up with a formula like that. The statement of the formula itself is not useful here. – Ron Gordon Nov 29 '16 at 14:02
  • @tired oh please solve it in a Ron Gordon way! I'd love it! – Enrico M. Nov 29 '16 at 14:02
  • 1
    @AlanTuring: see my comment to the question; I did it out that way previously. – Ron Gordon Nov 29 '16 at 14:03
  • 1
    @AlanTuring no,no, just go ahead. it is a good exercise for you – tired Nov 29 '16 at 14:04
  • @tired What exercise? I know how to derive the formula! You mean it's a good exercise to solve it with non -residues technique? – Enrico M. Nov 29 '16 at 14:22