0
$\begingroup$

$\gamma(\theta) = 2e^{i\theta}$ is a circle centered at $(0,0)$ with radius $2$, so $z = 1$ is inside this path and thus we have to use Cauchy's integral formula for $\int\limits_{\gamma} \frac{1}{z-1}$.

So Cauchy's integral formula says $\int\limits_{\gamma} \frac{f(z)}{(z-a)^{n+1}} = \frac{2\pi i}{n!} f^{n}(z)$.

If I choose $f(z) = 1$, then is this function still not holomorphic because the point $z = 1$ is on the path? If it is holomorphic, then the value of the integral is easy; it's just $2\pi i$. However, the value of the integral is $2\pi i$ for $0 \leq \theta \leq 2\pi$ over the entire path of the circle. How do I get the value for $0 \leq \theta \leq \frac{\pi}{2}$? Can I just divide $2\pi i$ by $4$ since $\frac{\pi}{2}$ is $\frac{1}{4}$ of the circle?

$\endgroup$
  • $\begingroup$ You need to use the parametrization you have been given to evaluate the integral $\endgroup$ – science Apr 17 '15 at 0:54
2
$\begingroup$

For the first question: you can use Cauchy. The formula applies to the integral $$\int_C \frac{f(z)}{(z-a)^{n+1}}\,dz$$ where $f(z)$ is holomorphic on and inside $C$. Note carefully: that's $f(z)$, not $f(z)/(z-a)^{n+1}$. The latter is not holomorphic, except in trivial cases, but it doesn't matter. Your $f(z)$ is just $1$, so it's holomorphic and you can use Cauchy.

Second question: if $\theta$ goes from $0$ to $\pi/2$ then you do not have a closed path and definitely cannot use Cauchy. You need to parametrise the path, or use an appropriate antiderivative. The integral will not be a quarter of the previous integral, except possibly by coincidence.

$\endgroup$
0
$\begingroup$

You need to go this way

$$ \int_{0}^{\pi/2} \frac{2i e^{i\theta} d\theta}{2e^{i\theta}-1}d\theta=\dots\,. $$

$\endgroup$
0
$\begingroup$

Reparametrize the curve as $$ \Gamma(t) = 1 + R(t) e^{it} \quad 0 \leq t \leq \pi- \arctan 4= \alpha $$ Note that, $R(t) = |2e^{it} -1|$, so no problem with continuity or differentiability wrt $t$. Now under this parametrization, $$ dz = (R'(t) + i R(t)) e^{it} dt , \qquad z-1 = R(t) e^{it}$$ So this leads to, $$ \int_\Gamma \frac{1} {z-1} dz = \int_0 ^{\alpha} (i + \frac{R'}{R})dt $$ $$\qquad = \big[i t + \log R(t)\big]_0 ^{\alpha}=i(\pi -\arctan 4)+ \log \sqrt5 $$

$\endgroup$

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.