0
$\begingroup$

$$\int_{0}^{\infty} 1/(x^3+1) dx$$ the question says use complex functions to answer this. I tried using Cauchy residue theorem and this $$1/2\int_{-\infty}^{\infty} 1/(x^3+1) dx=\int_{0}^{\infty} 1/(x^3+1) dx$$ So far i got that the poles are π/3,π.. but only π/3 matters as is it is inside the contour of integration. I then used cauchy residue theorem and did not get the answer of 2π/(3sqrt3). Is this the method i should be using? or is there an alternative way using complex functions. The intuition i used for this exercise was by looking at an example with $$\int_{0}^{\infty} 1/(x^4+1) dx$$.

$\endgroup$
  • $\begingroup$ BTW: if you do \infty you get $\infty$. $\endgroup$ – Dave Dec 26 '18 at 23:01
4
$\begingroup$

You made an error here:

$$ \frac12 \int^\infty_\infty \frac1{1 + x^3} \stackrel{??}{=} \int_0^\infty \frac1{1 + x^3}$$

This is not valid in this case because $\frac1{1 + x^3} \neq \frac1{1 + (-x)^3}$ in general (this is what was used for $1/(1 + x^4)$ - there it works because $x^4$ is even).

The trick I know for doing this integral is to integrate along the border of a circle segment with angle $2 \pi/3$ (see Frank's answer).

$\endgroup$
2
$\begingroup$

Let's generalize this by denoting$$\mathfrak{I}(n)=\int\limits_0^{\infty}\frac {\mathrm dx}{1+x^n}$$where $n\geq2$ and let$$f(z)=\frac 1{1+z^n}$$The trick for $\mathfrak{I}(n)$ is to integrate it along a sector instead of a circle or half circle. Let the radius of the sector be $R$ and the arc integral to be denoted as $\Gamma_R$. Since we only want to enclose one singularity, we'll have the central angle be $\theta=\frac {2\pi}n$ so the only singularity is at $z=e^{\pi i/n}$.

enter image description here

Integrating about the contour, we have$$\begin{align*}\oint\limits_{\mathrm C}\mathrm dz\, f(z) & =\int\limits_0^R\mathrm dx\, f(x)+\int\limits_{\Gamma_R}\mathrm dz\, f(z)+\int\limits_R^0\mathrm dz\, f(z)e^{\pi i/n}\\ & =(1-e^{\pi i/n})\int\limits_0^R\mathrm dx\, f(x)+\int\limits_{\Gamma_R}\mathrm dz\, f(z)\end{align*}$$

The arc integral vanishes as $R$ tends towards infinity. This can be verified using the estimation lemma$$\left|\,\int\limits_{\Gamma_R}\mathrm dz\, f(z)\,\right|\leq ML$$where $L$ is the length of the contour and $|f(z)|$ is bounded by a maximum $M$. The length of the arc can be easily calculated to be $L=\tfrac {2\pi R}n$. And through the triangle inequality, the max can be found$$|z|^n=\left|z^n\right|=\left|z^n+1-1\right|\leq\left|z^n+1\right|+1$$Since $|z|=R$, then we have that$$\left|\,\int\limits_{\Gamma_{R}}\mathrm dz\, f(z)\,\right|\leq\frac {1}{R^n-1}\frac {2\pi R}n\xrightarrow{R\,\to\,\infty}0$$

Hence$$\oint\limits_{\mathrm C}\mathrm dz\, f(z)=(1-e^{\pi i/n})\int\limits_0^{\infty}\mathrm dx\, f(x)$$

The contour integral is equal to $2\pi i$ times the sum of its residues. Fortunately for us, due to our clever thinking in the beginning, we've managed to only encase one singularity of $f(z)$ within our contour: $z=e^{\pi i/n}$. Therefore, the residue can be calculated as$$\begin{align*}\operatorname*{Res}_{z\,=\, e^{\pi i/n}}f(z) & =\lim\limits_{z\to e^{\pi i/n}}\frac {z-e^{\pi i/n}}{1+z^n}\\ & =\lim\limits_{z\to e^{\pi i/n}}\frac z{nz^n}\\ & =-\frac {e^{\pi i/n}}n\end{align*}$$Note that above, I have used L Hopital's rule as a shortcut. Putting everything together$$\begin{align*}\int\limits_0^{\infty}\frac {\mathrm dx}{1+x^n} & =-\frac {2\pi i}n\frac {e^{\pi i/n}}{1-e^{2\pi i/n}}\\ & =\frac {\pi}n\frac {2i}{e^{\pi i/n}-e^{-\pi i/n}}\\ & \color{blue}{\,=\frac {\pi}n\csc\left(\frac {\pi}n\right)}\end{align*}$$

The integral under question is simply the case when $n=3$. Therefore$$\mathfrak{I}(\color{brown}{3})=\int\limits_0^{\infty}\frac {\mathrm dx}{1+x^{\color{brown}{3}}}=\frac {\pi}{\color{brown}{3}}\csc\left(\frac {\pi}{\color{brown}{3}}\right)\color{red}{=\frac {2\pi}{3\sqrt3}}$$

$\endgroup$

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.