I want to evaluate the integral $$\int_{\left | z \right |=3} \frac{e^{z}-1}{z(z-1)(z-i)}\,dz$$ by using the following theorem:

If a function $f$ is analytic everywhere in the finite plane except for a finite number of singularities interior to a positively oriented simple closed curve $C$, then $$\int_{C}f(z) dz=2\pi i\text{Res}\left ( \frac{1}{z^{2}}f(\frac{1}{z}), 0 \right ).$$

I want to calculate this integral by calculating the residue at 0 of $\frac{1}{z^{2}}f(\frac{1}{z})=\frac{1+\frac{1}{2!z}+\frac{1}{3!z^{2}}\cdot \cdot \cdot }{1-(1+i)z+iz^{2}}$ So I did long division but the coefficient of 1/z is 0. I know the method of calculating the residues at each singularities 0, 1, i, respectively. This method gives me the answer $2\pi i(0+\frac{e-1}{1-i}+\frac{e^{i}-1}{i(i-1)})$ which is not 0. Why answers are different?? Why does not hold long division method? Any help please.

  • $\begingroup$ The residue at $i$ is $\frac {e^{i}-1} {i(i-1)}$ $\endgroup$ – Kavi Rama Murthy Jun 8 '18 at 7:18
  • $\begingroup$ Your answer is now correct. What are the different answers you are talking about? $\endgroup$ – Kavi Rama Murthy Jun 8 '18 at 7:51
  • $\begingroup$ one is zero another is not zero.. $\endgroup$ – purecj Jun 8 '18 at 7:54
  • $\begingroup$ Zero is not the answer. What makes you think the answer is 0? $\endgroup$ – Kavi Rama Murthy Jun 8 '18 at 8:39
  • $\begingroup$ I want to solve the integral by using the theorem which is edited above. So I did long division, but the coefficient of 1/z (which is the residue of f) is zero. $\endgroup$ – purecj Jun 9 '18 at 2:02

$$e^z-1=\sum_{n=1}^{\infty}\frac{z^n}{n!}=zg(z)$$ where $g(0) \neq 0$

So $0$ is a removable singularity of $f$.

Now $f$ has poles of order $1$ at $z=1,i \in D(0,3)$.

Thus $$\int_{|z|=3}f(z)dz=\frac{1}{2 \pi i} [Res(f,1)+Res(f,i)]=\frac{1}{2 \pi i}[\frac{e^i-1}{i(i-1)}+\frac{e-1}{1-i}]$$

  • $\begingroup$ Thanks. But I already know this solution. I want to solve the integral by using the theorem which is edited above. $\endgroup$ – purecj Jun 9 '18 at 2:03

We have that for $A=\frac{1+i}{2}$ and $B=\frac{1-i}{2}$, $$\frac{f(1/z)}{z^{2}}=\frac{(e^{1/z}-1)z}{(1-z)(1-iz)}= A(e^{1/z}-1)\cdot\frac{z}{1-z}-iB(e^{1/z}-1)\cdot\frac{iz}{1-iz}\\ =A\sum_{k\geq 1}\frac{1}{k!z^k}\sum_{j\geq 1}z^j-iB\sum_{k\geq 1}\frac{1}{k!z^k}\sum_{j\geq 1}(iz)^j$$ Now the residue at $0$ of the left-hand side, i.e. its coefficient of $1/z$, is obtained by letting $j=k-1$: $$A\sum_{k\geq 2}\frac{1}{k!}-B\sum_{k\geq 2}\frac{i^k}{k!}=A(e-2)-B(e^i-1-i)=\frac{1+i}{2}(e-2)-\frac{1-i}{2}(e^i-1-i)$$ which is exactly the sum of the other residues $\displaystyle 0+\frac{e-1}{1-i}+\frac{e^{i}-1}{i(i-1)}$.

P.S. A first version of this answer appeared in Evaluate the integral $\int_{\left | z \right |=3} (e^{z}-1)dz/(z(z-1)(z-i))$ then I realized that this question had been posted twice and I moved my answer here.

  • $\begingroup$ @purecj Long division is for polynomials and here the numerator is not a polynomial in $z$. The long division applied to $\frac{z}{1-(1+i)z+iz^{2}}$ will give you a series which multiplied by the expansion of $e^{1/z}-1$ gives you the correct result. $\endgroup$ – Robert Z Jun 9 '18 at 8:34
  • $\begingroup$ @purecj I am happy that you appreciate my efforts! $\endgroup$ – Robert Z Jun 9 '18 at 9:30

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.