1

This question was asked in a masters of mathematics exam for which I am preparing.

Compute $\int\limits_0^{\infty} \frac{x^{1/3}}{1+x^{2}} dx$.

I could only think of substituting $y^3 = x$ and that does not change much.

Could somebody post a solution using residues or in ways besides here in this link?: How to compute the integral $\int_{0}^{\infty} \frac{x^{1/3}}{1+x^{2}} \ dx$

Edit : I am interested in the answers which use contour integration and residue calculus.

5 Answers5

6

Evaluating $$\oint_C \frac{z^{\alpha-1}}{1+z} dz$$ we see that there is a branch cut along the positive $x-$axis and a pole at $z=-1$.

Take $C$ to be a keyhole contour consisting of a segment from $\epsilon$ to $R$, a circle of radius $R$, a segment from $R$ to $\epsilon$ and a small circle of radius $\epsilon$ surrounding the origin.

The result is:

$$\int_0^\infty \frac{ x^{\alpha-1}}{1+x} = \frac{\pi}{\sin \pi \alpha} \quad \text{when } 0<\alpha<1.$$

With the substitution $y^{1/2} = x$, our integral becomes

$$\bbox[5px, border: 1pt solid blue]{\int_0^\infty \frac{x^{1/3}}{1+x^2} dx = \frac{1}{2} \int_0^\infty \frac{y^{-1/3}}{1+y} dy = \frac{\pi}{2\sin \frac{2\pi}{3}}=\frac{\pi}{\sqrt{3}}.}$$

UPDATE:

In response to J.G.'s question:

The residue at $z=-1$ is $$b=\text{Res}_{z=-1} \frac{z^{\alpha-1}}{1+z}=e^{\pi i (\alpha-1)}.$$

So $$\oint_C \frac{z^{\alpha-1}}{1+z} dz = 2\pi i b$$

On the first segment (from $\epsilon$ to $R$), $z^{\alpha-1}=x^{\alpha-1}$, on the return trip, $z^{\alpha-1}=(e^{2\pi i} x)^{\alpha-1}.$

The integrals along the circles go to zero as $\epsilon\to0$, $R\to0$.

$$\int_0^\infty \frac{x^{\alpha-1}}{1+x}dx - \int_0^\infty \frac{e^{2\pi i (\alpha-1)} x^{\alpha-1}}{1+x}dx = 2\pi i e^{\pi i (\alpha-1)}$$

$$\begin{aligned} \int_0^\infty \frac{x^{\alpha-1}}{1+x}dx &=\frac{2\pi i e^{\pi i (\alpha-1)}}{1-e^{2\pi i (\alpha-1)}}\\ &= \frac{2\pi i}{e^{-\pi i (\alpha-1)}-e^{\pi i (\alpha-1)}} \\ &=\frac{\pi}{\sin \pi(1-\alpha)} \\ &= \frac{\pi}{\sin \pi \alpha}. \end{aligned}$$

mjw
  • 8,647
  • 1
  • 8
  • 23
  • 1
    Yes, "keyhole contour" is the critical point. This is one of those iconic problems that is also useful now and then in real life. – paul garrett Sep 02 '20 at 19:10
  • (+1) The classical keyhole contour works like a charm. – Mark Viola Sep 02 '20 at 19:26
  • Could you show how this gets us to $\pi\csc(\pi\alpha)$? – J.G. Sep 02 '20 at 20:02
  • @J.G., yes, edited my answer to show these details. Thanks. – mjw Sep 02 '20 at 22:17
  • @mjw can you please add more details to your update to J.G. 's question. I am having difficulties in understanding the proof but I really want to understand it as its's really nice. –  Sep 19 '20 at 16:06
  • @Ben, I would be happy to add more details. Which step or steps need further elaboration? Are you stuck at any point, or are any calculations unclear? – mjw Sep 21 '20 at 01:24
  • @mjw sorry for late reply, can you please write what are the curves ( their equation) that form the complete contour , it will help me to understand the answer. I know about contours of Radius $\epsilon $ and in general but I don't understand what " Keyhole contour" means as it has not been covered in our class, so kindly tell equation of curves in your answer. Your answer is more conceptual than others. So, I want to understand your answer. –  Oct 02 '20 at 08:24
  • "Another update" has the parameterizations.${}{}{}{}{}{}{}{}$ – mjw Oct 05 '20 at 21:31
5

With $x=\tan t$ it becomes $\int_0^{\pi/2}\tan^{1/3}tdt$. This can be evaluated in terms of the Beta function. In particualar, $\int_0^{\pi/2}\tan^{2s-1}tdt=\tfrac12\pi\csc\pi s$ implies your integral is $\tfrac12\pi\csc\frac{2\pi}{3}=\frac{\pi}{\sqrt{3}}$.

J.G.
  • 115,835
3

$\newcommand{\bbx}[1]{\,\bbox[15px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ With Ramanujan's Master Theorem: \begin{align} \int_{0}^{\infty}{x^{1/3} \over 1 + x^{2}}\,\dd x & \,\,\,\stackrel{x^{2}\ \mapsto\ x}{=}\,\,\, {1 \over 2}\int_{0}^{\infty}{x^{\color{red}{2/3} - 1} \over 1 + x}\,\dd x \end{align} Note that $\ds{{1 \over 1 + x} = \sum_{k = 0}^{\infty}\pars{-1}^{k}x^{k} = \sum_{k = 0}^{\infty}\color{blue} {\Gamma\pars{k + 1}}{\pars{-x}^{k} \over k!}}$.

Then, \begin{align} \int_{0}^{\infty}{x^{1/3} \over 1 + x^{2}}\,\dd x & = {1 \over 2}\,\Gamma\pars{\color{red}{2 \over 3}} \color{blue} {\Gamma\pars{1 + \bracks{-\,\color{red}{2 \over 3}}}} \\[2mm] &= {1 \over 2}\,{\pi \over \sin\pars{\pi/3}} = \bbx{{\root{3} \over 3}\,\pi} \\ & \end{align}

Felix Marin
  • 89,464
2

First, let's do a substitution to reduce the number of poles to be computed. In this problem, it wouldn't be necessary, since the function is fairly easy and there aren't too many poles, however it's a good a practice since in the future you may encounter more intricated problems.

$$\int_{0}^{\infty}{\frac{x^\frac{1}{3}}{x^2+1}dx}=\frac{1}{2}\int_{0}^{\infty}{\frac{z^{-\frac{1}{3}}}{z+1}dz}$$

Now, let’s define our function and integration path. Bear in mind that since we are working with complex functions $z^{-\frac{1}{3}}=\exp{\left(-\frac{Log\left(z\right)}{3}\right)}$, wich means that we must take care of the branch point. Hence, the selection of the keyhole contour. $$f(z)=\frac{z^{-\frac{1}{3}}}{z+1}$$

enter image description here

$$\oint f(z)dz=\left(\int_{0+ir}^{R+ir}+\int_{\Gamma}+\int_{R-ir}^{0-ir}+\int_\gamma \right)f(z)dz$$

1)Let's start working with the two integrals in the positive axis of the Real Line: $$\displaystyle{\lim_{R\rightarrow \infty\\ r\rightarrow 0}}\left(\int_{0+ir}^{R+ir}\int_{R-ir}^{0-ir} \right)f(z)dz$$

Remember that: $$Log(z)=\log|z|+i\text{Arg}(z)$$

$$\int_0^\infty \frac{\exp\left(-\frac{\log\left(z\right)+0i}{3}\right)}{z+1}dz+\int_\infty^0 \frac{\exp\left(-\frac{\log\left(z\right)+2\pi i}{3}\right)}{z+1}dz=\left(1-e^{-\frac{2\pi i}{3}}\right)\int_0^\infty\frac{z^{-\frac{1}{3}}}{z+1}dz$$

  1. To compute the others integrals let's make use of the the Estimation Lemma: $$\left\lvert\int_\Gamma f(z)dz\right\rvert\leq\displaystyle{\lim_{R\rightarrow \infty}}\left\lvert\int_0^{2\pi}\frac{\left(Re^{it}\right)^{-\frac{1}{3}}}{Re^{it}+1}Rie^{it}dt\right\rvert\leq\displaystyle{\lim_{R\rightarrow \infty}}2\pi R^{-\frac{1}{3}}\rightarrow 0$$

$$\left\lvert\int_\gamma f(z)dz\right\rvert\leq\displaystyle{\lim_{r\rightarrow 0}}\left\lvert\int_{2\pi}^{0}\frac{\left(re^{it}\right)^{-\frac{1}{3}}}{re^{it}+1}rie^{it}dt\right\rvert\leq\displaystyle{\lim_{r\rightarrow 0}}2\pi r^{\frac{2}{3}}\rightarrow 0$$

  1. Finally, let's compute the residue: $$\oint f(z)dz=2\pi i\displaystyle{\lim_{z\rightarrow -1}}\frac{z^{-\frac{1}{3}}}{z+1}(z+1)=2\pi ie^{-\frac{\pi i}{3}}$$

Gathering all results: $$\int_0^\infty\frac{z^{-\frac{1}{3}}}{z+1}dz=\frac{2\pi ie^{-\frac{\pi i}{3}}}{1-e^{\frac{-2\pi i}{3}}}=\frac{\pi }{\frac{e^{\frac{\pi i}{3}}-e^{-\frac{\pi i}{3}}}{2i}}=\frac{\pi}{\sin\left(\frac{\pi}{3}\right)}=\frac{2\pi}{\sqrt{3}}$$

Hence: $$\boxed{\int_{0}^{\infty}{\frac{x^\frac{1}{3}}{x^2+1}dx}=\frac{\pi}{\sqrt{3}}}$$

Teruo
  • 1,860
  • Why in point 2 of your answer : when you are making use of estimation lemma, Then how did you estimated 1st inequality in 1st line of 2. I think the inequality must be opposite –  Feb 12 '21 at 06:40
0

Substitution $$x=y^{3/2}$$ allows to write $$I=\int\limits_0^\infty\dfrac{\sqrt[\large3]x\,\text dx}{1+x^2} = \dfrac32\int\limits_0^\infty\dfrac{y\,\text dy}{1+y^3}.$$ At the same time, substitution $$y=\dfrac1z$$ gives $$I= \dfrac32\int\limits_0^\infty\dfrac{\text dz}{1+z^3} = \dfrac32\int\limits_0^\infty\dfrac{\text dy}{1+y^3}.$$ Therefore, $$I= \dfrac34\int\limits_0^\infty\dfrac{(1+y)\,\text dy}{1+y^3} = \dfrac34\int\limits_0^\infty\dfrac{\text dy}{1-y+y^2} =3\int\limits_0^\infty\dfrac{\text dy}{3+(2y-1)^2}.\tag1$$ Integral $(1)$ does not require the residue approach: $$I=\dfrac{\sqrt3}2\arctan\dfrac{2y-1}{\sqrt3}\bigg|_0^\infty = \dfrac{\sqrt3}2\left(\dfrac\pi2+\dfrac\pi6\right),$$ $$\color{brown}{\mathbf{I=\dfrac\pi{\sqrt3}.}}$$