The problem is $$\int_0^{2\pi} \frac{dt}{1+2\cos(t)}.$$ I know it is equal to $$\int\limits_{|z|=1}\frac{2dz}{i(1+z)^2}$$ but I don't know how I should calculate the last integral.
-
4Alternatively, we can use http://en.wikipedia.org/wiki/Weierstrass_substitution – lab bhattacharjee Jun 08 '13 at 11:10
-
@Roze: You can find some good starting points on how to format mathematics on the site here. This AMS reference is very useful. – Zev Chonoles Jun 08 '13 at 11:13
-
2Does this integral exist? – Mark Bennet Jun 08 '13 at 11:15
-
1Note that, $z=-1$ is on the path of integration which is the unit circle. – Mhenni Benghorbal Jun 08 '13 at 18:20
-
Instead of $$\int\limits_{|z|=1}\frac{2dz}{i(1+z)^2}$$ the complex integral should be $$\int\limits_{|z|=1}\frac{dz}{i\left( z+\frac{1-i\sqrt{3}}{2} \right) \left( z+\frac{1+i\sqrt{3}}{2}\right) }. $$ – Américo Tavares Jun 09 '13 at 21:20
4 Answers
Hint: $$ \int\frac{\mathrm{d}z}{i(1+z)^2}=\frac i{1+z}+C $$
As suggested in a comment, the Weierstrass Substitution is often useful in integrals such as the original. $$ \begin{align} \sin(t)&=\frac{2z}{1+z^2}\\ \cos(t)&=\frac{1-z^2}{1+z^2}\\ \mathrm{d}t&=\frac{2\,\mathrm{d}z}{1+z^2} \end{align} $$
Adjusted Hint:
As pointed out by Américo Tavares, $$ \begin{align} \int_0^{2\pi}\frac{\mathrm{d}t}{1+2\cos(t)} &=\int_0^{2\pi}\frac{\mathrm{d}t}{1+e^{it}+e^{-it}}\\ &=\int_0^{2\pi}\frac{e^{it}\,\mathrm{d}t}{e^{2it}+e^{it}+1}\\ &=\oint\frac{-i\,\mathrm{d}z}{z^2+z+1}\\ &=\oint\frac1{\sqrt3}\left(\frac1{z-\omega^2}-\frac1{z-\omega}\right)\,\mathrm{d}z \end{align} $$ where $\omega=\frac{-1+i\sqrt3}{2}$.
Since the sum of the residues at $\omega^2$ and $\omega$ is $0$, if we take a branch cut between $\omega$ and $\omega^2$, we can well-define $$ f(z)=\frac1{\sqrt3}\log\left(\frac{z-\omega^2}{z-\omega}\right) $$ over the rest of $\mathbb{C}$, and $f'(z)=\dfrac1{\sqrt3}\left(\dfrac1{z-\omega^2}-\dfrac1{z-\omega}\right)$. $f$ works the same as $\dfrac{i}{1+z}$ in my previous hint.
- 353,833
-
The trigonometric integral becomes $$ \begin{eqnarray} \int\limits_{|z|=1}\left( \frac{1}{1+2\frac{z+z^{-1}}{2}} \right) \frac{1}{iz}dz &=&\int\limits_{|z|=1}\frac{1}{i}\times \frac{1}{z^{2}+z+1}dz \ &=&\int\limits_{|z|=1}\frac{dz}{i\left( z+\frac{1-i\sqrt{3}}{2} \right) \left( z+\frac{1+i\sqrt{3}}{2}\right) } \end{eqnarray} $$ and not
$$\int\limits_{|z|=1}\frac{2dz}{i(1+z)^2},$$ as computed by OP.
– Américo Tavares Jun 11 '13 at 08:34 -
1@AméricoTavares: thanks for noticing this. The corrected integral can be handled in a similar way to the incorrect one, as I have amended to my answer. – robjohn Jun 11 '13 at 15:45
If you write $$ \begin{equation*} z=e^{it}\qquad \left( 0\leq t\leq 2\pi \right), \end{equation*} $$
since $$ \begin{equation*} \cos t=\frac{z+z^{-1}}{2},\qquad dt=\frac{dz}{iz}, \end{equation*} $$
the integral takes the form $$ \begin{eqnarray*} \int\limits_{|z|=1}\left( \frac{1}{1+2\frac{z+z^{-1}}{2}} \right) \frac{1}{iz}dz &=&\int\limits_{|z|=1}\frac{1}{i}\times \frac{1}{z^{2}+z+1}dz \\ &=&\int\limits_{|z|=1}\frac{dz}{i\left( z+\frac{1-i\sqrt{3}}{2} \right) \left( z+\frac{1+i\sqrt{3}}{2}\right) } \end{eqnarray*} $$ and not
$$\int\limits_{|z|=1}\frac{2dz}{i(1+z)^2}.$$
Since $\cos \frac{2\pi }{3}=\cos \frac{4\pi }{3}=-\frac{1}{2}$, hence $1+2\cos \frac{2\pi }{3}=1+\cos \frac{4\pi }{3}=0$, the given integral has singularities at $t\in\{\frac{2\pi, }{3},\frac{4\pi }{3}\}$. We thus split it as follows: $$ \begin{equation*} \int_{0}^{\frac{2\pi }{3}}\frac{dt}{1+2\cos t}+\int_{\frac{2\pi }{3}}^{\frac{4\pi }{3}}\frac{dt}{1+2\cos t}+\int_{\frac{4\pi }{3}}^{2\pi }\frac{dt}{1+2\cos t}. \end{equation*} $$
Each integral is a divergent improper integral of the second kind. Using the Weirstrass substitution$^1$ $x=\tan \frac{t}{2}$, for instance the first integral becomes
$$ \begin{eqnarray*} I_{1} &=&\int_{0}^{\frac{2\pi }{3}}\frac{1}{1+2\cos t}dt,\qquad x=\tan \frac{t}{2} \\ &=&\int_{0}^{\sqrt{3}}\frac{2}{\left( 1+2\frac{1-x^{2}}{1+x^{2}}\right) \left( 1+x^{2}\right) }\,dx \\ &=&\int_{0}^{\sqrt{3}}\frac{2}{3-x^{2}}\,dx,\qquad x=\sqrt{3}u \\ &=&\frac{2\sqrt{3}}{3}\int_{0}^{1}\frac{1}{1-u^{2}}\,du \\ &=&\left. \frac{2\sqrt{3}}{3}\operatorname{arctanh}u\right\vert _{0}^{1}=\infty . \end{eqnarray*} $$
--
$^1$ The Weierstrass substitution is a universal standard substitution to evaluate an integral of a rational fraction in $\sin t,\cos t$, i.e. a rational fraction of the form
$$R(\sin t,\cos t)=\frac{P(\sin t,\cos t)}{Q(\sin t,\cos t)},$$
where $P,Q$ are polynomials in $\sin t,\cos t$
$$ \begin{equation*} \tan \frac{t }{2}=x,\qquad t =2\arctan x,\qquad dt =\frac{2}{1+x^{2}}dx \end{equation*}, $$
which converts the integrand into a rational function in $x$. We know from trigonometry (see this answer) that
$$\cos t =\frac{1-\tan ^{2}\frac{t }{2}}{1+\tan ^{2}\frac{ t}{2}}=\frac{1-x^2}{1+x^2},\qquad \sin t =\frac{2\tan \frac{t }{2}}{1+\tan ^{2} \frac{t }{2}}=\frac{2x}{1+x^2}.$$
- 39,180
Open,$\cos(t)$ as $1-2\sin^2(t/2)$,
The denominator of your integral then becomes,$3-4\sin^2(t/2)$.
Multiply $\sec^2(t/2)$ in the Numerator and Denominator.
Your integrand then becomes,$\frac{\sec^2(t/2)}{3\sec^2(t/2)-4\tan^2(t/2)}$
Substitute,$z=\tan(t/2)$
Your integrand now modifies to(ignoring constants),$\frac{1}{3-z^2}$,
which is of known form and can be resolved using numerous methods(one of them being partial fractions)
- 337
-
Ok,I just realized(after looking at the other answers) that the definite integral diverges,but you could nevertheless use the indefinite integral in cases where the limits imply convergence of the integral. – Raghav Jun 08 '13 at 13:21
First, note that, the original integral is an improper integral since the integrand has singularities in the interval of integration $[0,2\pi]$. Namely, $t=\frac{2}{3}\pi$ and $\frac{4}{3}\pi$. You can find these singularities by solving the equation
$$ 1 + 2\cos(t) = 0. $$
So, you should know how to handle this integral. Here is the value of the indefinite integral
$$ \frac{2}{\sqrt {3}}\,\,{\it \tanh^{-1}} \left( \frac{1}{\sqrt{3}}\,\tan \left( \frac{t}{2} \right) \right) .$$
For the complex integral, you can see that you have a pole on the path of integration as I pointed out in my comment.
- 48,099