Use the residue theorem to evaluate $\int_\gamma \frac{z^5}{1-z^3}dz$ where $\gamma$ is the circle $|z|=2$.

I have that $z_0=1$ is a singularity point and taking $g(z)=z^5$ and $h(z)=1-z^3$ and knowing that $g(z_0)\neq 0$ , $h(z_0)=0$ and $h'(z_0)\neq 0$ then $f(z)=\frac{g(z)}{h(z)}$ have simple pole at $z_0$ and $$Res(f;z_0)=\frac{g(z_0)}{h'(z_0)}=-\frac{1}{3}$$

then $\int_\gamma \frac{z^5}{1-z^3}dz=2\pi i* Res(f;z_0)=-\frac{2\pi i}{3}$

but the answer is $-2\pi i$

  • 2
    $\begingroup$ hint: factor $1-z^{3}$ completely $\endgroup$ – Matematleta Jun 11 '15 at 23:36
  • $\begingroup$ @Chilango I tried but I could not factor $\endgroup$ – Roland Jun 11 '15 at 23:37
  • $\begingroup$ This function has 3 different poles of order 1 inside of the region. $\endgroup$ – Braindead Jun 11 '15 at 23:39
  • $\begingroup$ You have discovered one singularity. Why do you think you've discovered them all? $\endgroup$ – zhw. Jun 11 '15 at 23:39
  • $\begingroup$ You can also compute the reside at infinity. $\endgroup$ – Braindead Jun 11 '15 at 23:39

First of all note that $$ 1-z^3=-(z-z_1)(z-z_2)(z-z_3)=-(z-1)\left(z-\frac{1}{2}(-1-i\sqrt{3})\right)\left(z-\frac{1}{2}(-1+i\sqrt{3})\right) $$ Hence, since $|z_j|=1$ for $j=1,2,3$, all three poles of $g/h$ are simple an lie inside the circle $|z|=2$. Therefore by the residue theorem we must have $$ \int_{|z|=2}\frac{z^5}{1-z^3}dz = 2\pi i \sum_{j=1}^3 Res(g/h,z_j ) $$ But, since all the poles are simple ones, $$ Res(g/h,z_1)=\lim_{z\to z_1} (z-z_1)\frac{g(z)}{h(z)} = \lim_{z\to 1} \frac{z^5}{-\left(z-\frac{1}{2}(-1-i\sqrt{3})\right)\left(z-\frac{1}{2}(-1+i\sqrt{3})\right)} =- \frac{1}{3} $$ Similarly $$ Res(g/h,z_2)=\lim_{z\to z_2} (z-z_2)\frac{g(z)}{h(z)} =- \frac{1}{3} $$ and $$ Res(g/h,z_3)=\lim_{z\to z_3} (z-z_3)\frac{g(z)}{h(z)} = - \frac{1}{3} $$ Thus $$ 2\pi i \sum_{j=1}^3 Res(g/h,z_j )= 2\pi i\left(3 \times \frac{-1}{3}\right)=-2\pi i $$


Here is an approach that avoids factorization. Start by noticing $$ \frac{z^5}{1-z^3}=\frac{z^2}{1-z^3}-z^2 $$ Since there are no singularities outside of the unit circle, for any $R\ge2$, we have $$ \begin{align} \int_{|z|=2}\frac{z^5}{1-z^3}\,\mathrm{d}z &=\int_{|z|=2}\frac{z^2}{1-z^3}\,\mathrm{d}z\\ &=\int_{|z|=R}\frac{z^2}{1-z^3}\,\mathrm{d}z \end{align} $$ Furthermore, noticing that $\frac{z^2}{1-z^3}=-\frac1z\frac1{1-1/z^3}$ is very close to $-\frac1z$ when $|z|$ is large, we add the integral of $\frac1z$ in hopes that it will cancel. $$ \begin{align} \left|\,\int_{|z|=R}\frac{z^2}{1-z^3}\,\mathrm{d}z+\int_{|z|=R}\frac1z\,\mathrm{d}z\,\right| &\le\sup_{|z|=R}\left|\,\frac{z^2}{1-z^3}+\frac1z\,\right|2\pi R\\ &\le\sup_{|z|=R}\left|\,\frac1{(1-z^3)z}\,\right|2\pi R\\ &\le\frac{2\pi}{R^3-1} \end{align} $$ Since $R$ is arbitrary, the sum of the integrals on the left must be $0$. Therefore, $$ \begin{align} \int_{|z|=2}\frac{z^5}{1-z^3}\,\mathrm{d}z &=\int_{|z|=2}\frac{z^2}{1-z^3}\,\mathrm{d}z\\ &=-\int_{|z|=2}\frac1z\,\mathrm{d}z\\[6pt] &=-2\pi i \end{align} $$

  • $\begingroup$ Why you sum $\frac{1}{z}$? $\endgroup$ – Roland Jun 12 '15 at 0:07
  • 1
    $\begingroup$ @askazy: because near $z=\infty$, $\frac{z^2}{1-z^3}=-\frac1z\frac1{1-1/z^3}$ looks like $-\frac1z$. $\endgroup$ – robjohn Jun 12 '15 at 0:16
  • 2
    $\begingroup$ @askazy: that is a good question, and pretty much the whole idea to this approach, so I have incorporated my comment into my answer. $\endgroup$ – robjohn Jun 12 '15 at 0:23
  • $\begingroup$ Very nice approach +1 !! $\endgroup$ – Alonso Delfín Jun 12 '15 at 0:31
  • $\begingroup$ I think it's a very innovative approach. Very nice. $\endgroup$ – Braindead Jun 12 '15 at 2:25

If $a\ne 1$ denotes a root of $z^3-1$, then $$ \int_{|z|=2}\frac{z^5}{1-z^3}\,dz=2\pi i\left[\frac{(a^0)^5}{-3(a^0)^2}+\frac{a^5}{-3a^2}+\frac{(a^2)^5}{-3(a^2)^2}\right]=-\frac{2\pi i}{3}(1+a^3+(a^2)^3)=-2\pi i. $$

  • $\begingroup$ Nice and easy, avoids factorization and uses that $a$ is a root of unity (+1) $\endgroup$ – Alonso Delfín Jun 12 '15 at 0:52

All three poles are inside of the curve. In this situation, rather than computing the residue at each pole, you can calculate the residue at infinity:

$\mathrm{Res}_{z=\infty} f(z) = - \mathrm{Res}_{w=0} \dfrac{f(1/w)}{w^2}$

Using a simple change of variables $z\mapsto 1/w$, can be shown that

$$\int_{C} f(z) dz = - \mathrm{Res}_{z=\infty} f(z) =\mathrm{Res}_{w=0} \dfrac{f(1/w)}{w^2},$$ if $C$ encloses all finite singular points.

In this case,

$$\dfrac{f(1/w)}{w^2} = \dfrac{1}{w^3-1}\dfrac{1}{w^4}$$

So now the residue of this new expression at $w=0$ can be computed in one of the two ways:

  1. This is a pole of order 4. So we can take the 3rd derivative of $1/(w^3-1)$ at $0$ then divide by 3!.

  2. Using the geometric series expansion of $1/(w^3-1)$. You can use the relationship between the Taylor series coefficients and derivatives. Alternatively, you can look at the $w^{-1}$th term in the series expansion of $\dfrac{1}{w^3-1}\dfrac{1}{w^4}$ for the residue.

No matter what, the residue of $\dfrac{1}{w^3-1}\dfrac{1}{w^4}$ at $w=0$ is $-1$. So the original integral is $-2\pi i$.


I think your mistake is, that you forgot 2 singularities: $- \frac{1}{2} + i\sqrt{\frac{3}{4}}$ and $- \frac{1}{2} - i\sqrt{\frac{3}{4}}$. The way to find these singularities is to set $z^3-1=0$ and then factor your solution $z=1$ out, this leads you to $z^2+z+1=0$. And this equation has the two complex solutions i mentioned above.

The nice thing is that (if you calculate it) the residue at these points are $-\frac{1}{3}$ for each of them, so in total the solution $-2 \pi i$ is correct.


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.