Evaluate $\int_0^1 \frac{dx}{(x^2-x^3)^\frac13}$ using contour integration

I try to proceed by first calculating residues of $$f(x) = \frac{1}{(x^2-x^3)^\frac13} = \frac{1}{x\left(\frac1x-1\right)^\frac13}$$

Then answer to integral can be provided by Cachy's Residue theorem = $2i\pi(\text{Residue at } x=0 +\text{ Residue at } x=1)$

Above function has simple pole at $x=0$ but I am unable to find out what kind of singularity $f$ has at $x= 1$.

Is this the right approach? Or is there any way to expand given function(Laurent's series) to find out all residues at once?

  • 2
    $\begingroup$ Maybe take a look at the solution for the equivalent integral here -> math.stackexchange.com/q/1703805/254075 $\endgroup$ – sharding4 Jun 24 '17 at 1:48
  • $\begingroup$ The integral has branch points at $0$ and $1$. You therefore either have to compute the residue at $\infty$ (which requires some care about what value the cube root takes on the contour, and only works in special cases) or use the Pochhammer contour, which is very complicated. $\endgroup$ – Chappers Jun 24 '17 at 1:54

This doesn't work. For one thing, the residue theorem is only valid around closed contours. But the bigger problem is that this function is not analytic near $0$ or $1$.

There is a sneaky way to solve this by making a branch cut along the interval $[0,1]$ and using the residue at $\infty$. Here, you would integrate over a very narrow oval encircling the interval $[0,1]$ and pay close attention to the fact that the values approaching $[0,1]$ on either side of the interval are not quite the same. (They differ by multiplication by $e^{2\pi i/3}$ or $e^{4\pi i /3}$, depending on which way you look at it.)

But it's important that you apply the residue theorem to the outside of this contour, since the function has a branch cut within it.

The easier way is to use the beta function $$B(s,t) = \int_0^1 x^{s-1} (1-x)^{t-1} \, \mathrm{d}x = \frac{\Gamma(s)\Gamma(t)}{\Gamma(s+t)}$$ and the Euler reflection formula $$\Gamma(s)\Gamma(1-s) = \frac{\pi}{\sin(\pi s)},$$ and notice that your integral is $$\int_0^1 x^{-2/3} (1-x)^{-1/3} \, \mathrm{d}x = \frac{\Gamma(1/3) \Gamma(2/3)}{\Gamma(1)} = \frac{\pi}{\sin(\pi/3)}.$$

  • $\begingroup$ This question has been asked from my syllabus itself which does not include Euler reflection formula or beta functions. In fact I have never even heard of them. Never came across branch cutting as well in other questions from same topic(graduation level). Is there any other way around to solve this? $\endgroup$ – LoneCuriousWolf Jun 24 '17 at 1:45
  • $\begingroup$ @AbhishekRana The fact that $\lim_{y \rightarrow 0} f(x+iy)$ is different if you approach from $y > 0$ or $y < 0$ is important and I don't think you'll get anywhere without using this. You don't have to use the language "branch cutting" but it would make it easier. $\endgroup$ – user399601 Jun 24 '17 at 1:49

Another variation on a solution theme in Finding the integral $I=\int_0^1{x^{-2/3}(1-x)^{-1/3}}dx$

Use the logistic map $x=\frac{1}{1+e^w}$, $1-x=xe^w$, $x^2-x^3=x^3e^w$, to map the interval $w\in(-∞,∞)$ to $x\in(0,1)$, moving the branch points to $w=\pm\infty$. Then $dx=x^2e^w\,dw$ so that $$ \int_0^1\frac{dx}{(x^2-x^3)^{1/3}}=\int_{-∞}^∞\frac{x^2e^wdw}{xe^{w/3}}=\int_{-∞}^∞\frac{e^{2w/3}}{1+e^w}dw $$

Now consider $w=u+iv$ restricted to the rectangle bounded by the lines $u=\pm R$, $v=0$, $v=2\pi$ which has one pole at $w=i\pi$ inside with residue $$ \lim_{w\to i\pi}\frac{e^{2w/3}(w-i\pi)}{1+e^w} =\lim_{h\to 0}\frac{e^{2(i\pi+h)/3}h}{1-e^h}=-e^{i2\pi/3} $$ The contributions from the sides $\pm R+i[0,2\pi]$ go to zero with $R\to\infty$ like $e^{-2R/3}$ resp. $e^{-R}$. The integral over the top segment $[-R,R]+i2\pi$ gives a phase-shifted multiple of the integral of the bottom segment $[-R,R]+i0$, $$ \int_{-R+i2\pi}^{R+i2\pi}\frac{e^{2w/3}}{1+e^w}dw =e^{i4\pi/3}\int_{-R}^{R}\frac{e^{2w/3}}{1+e^w}dw $$ so that the residual theorem gives in the limit $R\to\infty$ $$ -i2\pi\,e^{i2\pi/3}=(1-e^{i4\pi/3})\int_{-∞}^∞\frac{e^{2w/3}}{1+e^w}dw $$ or $$ \int_0^1\frac{dx}{(x^2-x^3)^{1/3}}=\frac{i2\pi}{e^{i2\pi/3}-e^{-i2\pi/3}}=\frac{\pi}{\sin(2\pi/3)}. $$


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.