2
$\begingroup$

How can I calculate the integrals $$\int_0^\pi \frac{dx}{a+b\sin(x)}$$ and $$\int_0^\pi \frac{dx}{a+b\cos(x)}$$ using contour integrals ?

I know the residue theorem , and my idea is to choose the line from $0$ to $\pi$ on the $x$-axis followed by the upper half-circle connecting the points $(\pi/0)$ and $(0/0)$.

But I do not know the singularities of the functions $f(z)=\frac{1}{a+b\sin(z)}$ and $g(z)=\frac{1}{a+b\cos(z)}$.

The other problem is that I do not know whether the integral over the half-circle can be calculated by hand.

Is there a way, or must I begin with a substitution, for example $t=\tan(\frac{x}{2})$ ?

$\endgroup$
  • 3
    $\begingroup$ The usual approach is to put $z=e^{ix}$ and use Euler's formulas to rewrite everything as an integral of a rational function along the unit circle. (Here you need some extra argument to get from $[0,\pi]$ to $[0,2\pi]$.) See for example math.stackexchange.com/questions/1610037 for something similar. $\endgroup$ – mrf Oct 20 '16 at 11:32
  • 1
    $\begingroup$ as an expericend user you should be able to find dozens of posts which explain the procedure properly. For example my first MSE post: math.stackexchange.com/questions/955363/… $\endgroup$ – tired Oct 20 '16 at 11:36
  • 1
    $\begingroup$ So, everything I have to do is writing $\sin(x)$ or $\cos(x)$ in terms of $e^{ix}$ and substitute $z=e^{ix}$ to calculate the residue ? $\endgroup$ – Peter Oct 20 '16 at 11:39
  • 1
    $\begingroup$ in the case of cosine it is obvious how to extend the interval of integration to a complete period. the sine case might be more complicated but might be also doable $\endgroup$ – tired Oct 20 '16 at 11:58
  • 1
    $\begingroup$ ok..in the case of sine, i think one has to also perform straight line integrals $(-1,1)$ which aren't too complicated $\endgroup$ – tired Oct 20 '16 at 12:01
1
$\begingroup$

The integral over the sine is a bit tricky because the integration region cannot be extended to $2 \pi$, so if you want to used the residue theorem, a unit circle is out. Rather, we use a semicircle. Thus consider the complex integral

$$\oint_C \frac{dz}{b z^2+i 2 a z-b} $$

where $C$ is the unit semicircle in the upper half-plane along with the diameter along the real axis. Thus, the contour integral is equal to

$$\frac12 \int_0^{\pi} \frac{d\theta}{a+b \sin{\theta}} + \int_{-1}^1 \frac{dx}{b x^2+i 2 a x-b}$$

The contour integral is also equal to $i 2 \pi$ times the sum of the residues of the poles inside $C$. In this case, let's assume $a \gt b \gt 0$; the poles are at

$$z_{\pm}=-i \frac{a}{b} \pm i \sqrt{\frac{a^2}{b^2}-1} $$

Note that both poles are outside of the interior of the contour $C$; thus, the contour integral is zero. Accordingly,

$$\begin{align} \int_0^{\pi} \frac{d\theta}{a+b \sin{\theta}} &= -2 \int_{-1}^1 \frac{dx}{b x^2+i 2 a x-b} \\ &= -\frac{2}{b} \frac1{z_+-z_-} \int_{-1}^1 dx \left (\frac1{x-z_+} - \frac1{x-z_-} \right ) \\ &= - \frac{2}{i 2 \sqrt{a^2-b^2}} \left [\log{\left (\frac{1-z_+}{-1-z_+} \right )} - \log{\left (\frac{1-z_-}{-1-z_-} \right )} \right ]\\ &= - \frac{2}{i 2 \sqrt{a^2-b^2}} \left [\log{\left (\frac{1-z_+}{1+z_+} \right )} - \log{\left (\frac{1-z_-}{1+z_-} \right )} \right ]\\ &= - \frac{2}{i 2 \sqrt{a^2-b^2}} i 2 \left [ \arctan{\left ( \frac{a}{b} - \sqrt{\frac{a^2}{b^2}-1}\right )} - \arctan{\left ( \frac{a}{b} + \sqrt{\frac{a^2}{b^2}-1}\right )} \right ] \\ &= \frac{2}{\sqrt{a^2-b^2}} \arctan{\left (\frac{\sqrt{a^2-b^2}}{b} \right )} \end{align}$$

Things work out similarly when $b \lt a$. Accordingly, when $a \gt 0$ and $b \gt 0$,

$$\int_0^{\pi} \frac{d\theta}{a+b \sin{\theta}} = \begin{cases}\frac{2}{\sqrt{a^2-b^2}} \arctan{\left (\frac{\sqrt{a^2-b^2}}{b} \right )} & a \gt b \\ \frac{1}{\sqrt{b^2-a^2}} \log{\left (\frac{b+\sqrt{b^2-a^2}}{b-\sqrt{b^2-a^2}} \right )} & a \lt b \end{cases}$$

$\endgroup$

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.