1
$\begingroup$

I'm trying to learn branch cuts and its about $7$ days I'm thinking about doing one of Arfken problems:

Show below relation using the contour:

enter image description here

$$\int_0 ^1 \frac{dx}{(x^2 - x^3)^{\frac{1}{3}}} = \frac{2 \pi}{ 3^{\frac{1}{2}}}$$

I could calculate the integral on the big circle. And I know that because there is no pole in the region, I can put all integrals equal to zero. But I don't know how to calculate other integrations. I really need your help Mathematicians. Any answers are highly appreciated.

  • 0
    please include the image directly2017-01-29
  • 0
    I did. Could you help now please? @tired2017-01-29
  • 0
    I think it is not very hard question and it is a little odd that there is no one in mathematics stack exchange who could solve the problem!2017-01-29
  • 0
    there are enough people who can, but maybe not so many who care, to equate this integral. Especially since the formatting and content of this question is far from optimum.2017-01-30
  • 0
    i give you a small hint: it is very convineant here to take the residue at infinity into account2017-01-30
  • 0
    I didn't underestand what you meat. there is not any residue. It's about branches. Could you please help me to optimize the question? What should I do? I really need help to solve this problem. @tired2017-01-30
  • 0
    the integral around the branchcut $=$ constant times your original integral $=$ $-2\pi i$ times residue at infinity $=$ integral over the big circle in the limit of $R\rightarrow \infty$2017-01-30
  • 0
    Furthermore have a look at example 6 here: https://en.wikipedia.org/wiki/Methods_of_contour_integration#Example_6_.E2.80.93_logarithms_and_the_residue_at_infinity2017-01-30

2 Answers 2

0

$\newcommand{\bbx}[1]{\,\bbox[8px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$

$\ds{\int_{0}^{1}{\dd x \over \pars{x^{2} - x^{3}}^{1/3}} = {2 \pi \over 3^{1/2}}:\ {\large ?}}$.

I'll consider the integral $\ds{\oint_{\mc{DB}}z^{-2/3}\pars{1 - z}^{-1/3}\,\dd z}$. The contour $\mc{DB}$ is the dog-bone one of the OP picture.

$\ds{z^{-2/3}}$ and $\ds{\pars{1 - z}^{-1/3}}$ are given by $$ \left\{\begin{array}{l} \ds{z^{-2/3} = \verts{z}^{\,-2/3}\exp\pars{-\,{2 \over 3}\arg\pars{z}\ic}\,, \qquad -\pi < \arg\pars{z} < \pi\,,\quad z \not= 0} \\[2mm] \ds{\pars{1 - z}^{-1/3} = \verts{1 - z}^{\,-1/3}\exp\pars{-\,{1 \over 3}\arg\pars{1 - z}\ic}\,, \qquad 0 < \arg\pars{1 - z} < 2\pi\,,\quad z \not= 1} \end{array}\right. $$ By multiplying $\ds{z^{-2/3}}$ and $\ds{\pars{1- z}^{-1/3}}$, as given above, we'll see that the product branch-cut is set along $\ds{\bracks{0,1}}$.


\begin{align} \oint_{\mc{DB}}z^{-2/3}\pars{1 - z}^{-1/3}\,\dd z & = \int_{1}^{0}x^{-2/3}\pars{1 - x}^{-1/3}\expo{-2\pi\ic/3}\,\dd x + \int_{0}^{1}x^{-2/3}\pars{1 - x}^{-1/3}\,\dd x \\[5mm] & = 2\ic\expo{-\pi\ic/3}\sin\pars{\pi \over 3}\int_{0}^{1}x^{-2/3} \pars{1 - x}^{-1/3}\,\dd x \\[5mm] & = \ic\root{3}\expo{-\pi\ic/3}\int_{0}^{1}x^{-2/3}\pars{1 - x}^{-1/3}\,\dd x \label{1}\tag{1} \\[1cm] \oint_{\mc{DB}}z^{-2/3}\pars{1 - z}^{-1/3}\,\dd z & = -2\pi\ic\,\mrm{Res}_{\, z = 0}\pars{% -\,{1 \over z^{2}}\,z^{2/3}\bracks{1 - {1 \over z}}^{-1/3}} \\[5mm] & = 2\pi\ic\,\mrm{Res}_{\, z = 0}\pars{% {1 \over z}\,\bracks{z - 1}^{-1/3}} = 2\pi\ic\,\verts{0 - 1}^{-1/3} \expo{-\pi\ic/3} \label{2}\tag{2} \end{align}
\eqref{1} and \eqref{2} lead to $$\bbx{\ds{% \int_{0}^{1}x^{-2/3}\pars{1 - x}^{-1/3}\,\dd x = {2\pi \over 3^{1/2}}}} $$

The whole procedure is explained in detail in a Methods of Contour Integration page.

0

The reason you are having trouble is that the picture is not all that great. Better would be lines extending above and below the branch cut to the large circle. In this case, the contour integral

$$\oint_C dz \, z^{-2/3} (z-1)^{-1/3} $$

is equal to, if the branch cut is the portion of the real axis $x \lt 1$:

$$\int_0^1 dx \, x^{-2/3} \, e^{-i \pi/3} (1-x)^{-1/3} + i R \int_{-\pi}^{\pi} d\theta \, e^{i \theta} \, \left (R e^{i \theta} \right )^{-2/3} \left (R e^{i \theta}-1 \right )^{-1/3} \\ + \int_1^0 dx \, x^{-2/3} \, e^{i \pi/3} (1-x)^{-1/3}$$

The factors $e^{\pm i \pi/3}$ come from the values of the phase $e^{\pm i \pi}$ above and below the branch cut. This is applied when we change the sign of $z-1$ when $z \lt 1$. Note that the contributions to the integral above and below the branch cut cancel when $z \ge 1$.

By Cauchy's theorem the contour integral is zero. You can take the limit as $R \to \infty$ in the second integral (around the large circle) and find it becomes $i 2 \pi$ in this limit - this is the so-called residue at infinity.

The stated result follows.

  • 0
    In the integral from zero to one, why did you used $e^{- i \pi/3}$ instead of $e^{- 2 i \pi/3}$? I mean, $x ^ {- 2 /3}$ and $(1-x) ^ {- 1 /3}$ are both below the branch cut. Why didn't you used two separate phases for them?2017-01-30
  • 0
    @saranj: not sure what you're saying makes any sense. I used $e^{\pm i \pi/3}$ because that arises from the $(-1)^{-1/3}$ when converting $z-1$ to $1-z$.2017-01-30
  • 0
    but in the integral below the branch cut, why should $z - 1$ converge to -1?2017-01-30