1

This post might appear as a duplicate of the following:

Over an integral arising from Kepler's problem [also: generally useful integral]

So recalling quickly:

$$\Phi(\epsilon) = \frac{1}{2\pi}\int_0^{2\pi}\ \frac{\sin^2\theta\ \text{d}\theta}{(1 + \epsilon\cos\theta)^2}$$

Note: The integral arises in studying the Kepler's problem, which is why $\epsilon\in [0;\ 1]$.

Necessary Note I tried to solve it with complex analysis, as you can see in the above link, but in my book it's solved by series with lots of unknown passages, which I will write down here.

$$ \begin{align*} \frac{1}{2\pi}\int_0^{2\pi}\ \frac{\sin^2\theta\ \text{d}\theta}{(1 + \epsilon\cos\theta)^2} & = \sum_{n = 1}^{+\infty}\ (-1)^n\epsilon^n(n+1)\cdot\frac{1}{2\pi}\int_0^{2\pi}\ (1 - \cos\theta)\cos^n\theta\ \text{d}\theta \\\\ & = \sum_{n = 0}^{+\infty}\ \epsilon^{2n}(2n+1)\left[\frac{1}{2^{2n}} \binom{2n}{n} - \frac{1}{2^{2n+2}}\binom{2n+2}{n+1}\right] \\\\ & = \sum_{n = 0}^{+\infty}\ \epsilon^{2n}\frac{2n+1}{2^{2n+2}} \frac{(2n)!}{(n!)^2} \left[4 - \frac{(2n+1)(2n+2)}{(n+1)^2}\right] \\\\ & = \sum_{n = 0}^{+\infty}\ \epsilon^{2n} \frac{(2n+1)!}{2^{2n+1}(n!)^2 (n+1)} \\\\ & = \frac{1}{\epsilon^2}\ \sum_{n = 0}^{+\infty} \epsilon^{2(n+1)}(-1)^{n+1}\binom{-1/2}{n+1} \\\\ & = \frac{1}{\epsilon^2}[(1 - \epsilon^2)^{-1/2} - 1] \end{align*} $$

Can anybody help with that? It seems really messy..

P.s. I wrote down here exactly the passages of the book.

What I need

I am not familiar with the passage from the first row to the second one (namely: where the integral has been solved, in which we pass to the binomial coefficient). In addition to that, I didn't understand how the very last series has been summed.

Enrico M.
  • 26,114
  • what exactly do u find messy? i find this derivation quiet nice and transparent – tired Jan 24 '16 at 18:23
  • 1
    if u really want solve this by complex analysis: consider the map from $(0,2\pi]$ to $S^1$ induced by $z=e^{i x}$ and apply the residue theorem . – tired Jan 24 '16 at 18:25
  • What do you mean by "Can anybody help with that"? Is there a step you don't understand? Unless you clarify where you're having trouble, it's unclear what you're asking. –  Jan 24 '16 at 18:26
  • @tired altrady did with complex analysis. Your comments are not constructive, if you find that really clear then please, enlighten me! – Enrico M. Jan 24 '16 at 18:26
  • @KimPeek i asked u ''what exactly do you find messy?''. could u enlighten me what is ''not constructive'' with this comment? – tired Jan 24 '16 at 18:28
  • 1
    I suspect this can be written as a Laplace coefficient. – Alice Jan 24 '16 at 18:44
  • Are you sure that you have copied this exactly? On the first line in this chain of equalities, inside the integral on the r.h.s., isn't it $(1-\cos^2 \theta)$ (note the square!)? – Alex M. Jan 24 '16 at 19:29
  • @AlexM. Nope, I copied exactly what the book wrote! – Enrico M. Jan 24 '16 at 20:08
  • @Alice Laplace coefficients? I don't know if I'm familiar with that method.. Any reference? ^^ – Enrico M. Jan 24 '16 at 20:09
  • @KimPeek: Well, this means that there are many mistakes in the book. The square should have been there, because it comes from $\sin ^2 = 1 - \cos ^2$ ($\sin ^2$ being in the numerator). Also, the series on the first line should start at $n=0$ like the ones on the other lines (I was wondering where the first term had disappeared, but now it's clear: it's yet another typo). – Alex M. Jan 24 '16 at 20:11
  • @tired Saying the derivation is "nice and transparent" is not that constructive, isn't it? I mean sure, You may find it so, but I don't. I wrote what are my doubts in the editing! – Enrico M. Jan 24 '16 at 20:11
  • @KimPeek then ask me what u don't understand (as i offered) instead of piss me off... – tired Jan 24 '16 at 20:16
  • @tired You're right about that, I acted in a stupid way and I'm sorry. I gave you a little upvote ^^ – Enrico M. Jan 24 '16 at 20:18
  • @KimPeek Don't worry, things like this happen! Thanks for the apology (and the upvote ;) ) – tired Jan 24 '16 at 20:19

2 Answers2

5

There is an elementary path: $$I = \frac{1}{2\pi}\int_0^{2\pi}\frac{\sin^2\theta}{(1+\varepsilon\cos\theta)^2}\,d\theta =\frac{1}{2\pi\varepsilon}\int_0^{2\pi}\sin\theta\,d\left(\dfrac{1}{1+\varepsilon\cos\theta}\right)$$ By parts: $$ I = \frac{1}{2\pi\varepsilon}\dfrac{\sin\theta}{1+\varepsilon\cos\theta}\,\biggr|_0^{2\pi} - \dfrac{1}{2\pi\varepsilon}\int_0^{2\pi}\frac{\cos\theta\,d\theta}{1+\varepsilon\cos\theta} = 0 - \frac{1}{2\pi\varepsilon^2}\int_0^{2\pi}\frac{\varepsilon\cos\theta+1-1}{1+\varepsilon\cos\theta}\,d\theta$$$$ = -\frac{1}{2\pi\varepsilon^2}\theta\,\biggr|_0^{2\pi} + \frac{1}{2\pi\varepsilon^2}\int_0^{2\pi}\frac{d\theta}{1+\varepsilon\cos\theta} = -\frac{1}{\varepsilon^2} + \frac{1}{\pi\varepsilon^2}\int_0^{\pi}\frac{d\theta}{1+\varepsilon\cos\theta}.$$ Applying universal trigonometric substitution,
$$\theta = 2\arctan u,\quad d\theta = \dfrac{2}{1+u^2}du,\quad\cos\theta = \dfrac{1-u^2}{1+u^2},$$
we get: $$I=-\dfrac{1}{\varepsilon^2}+\dfrac{2}{\pi\varepsilon^2}\int_0^\infty\frac{d\theta}{1+\varepsilon+(1-\varepsilon)u^2},$$ $$I=-\dfrac{1}{\varepsilon^2}+\dfrac{2}{\pi\varepsilon^2(1-\varepsilon)}\sqrt{\dfrac{1-\varepsilon}{1+\varepsilon}}\arctan\sqrt{\dfrac{1+\varepsilon}{1-\varepsilon}}u\,\biggr|_0^\infty,$$ $$ \boxed {I = \dfrac{1}{\varepsilon^2}\left(\dfrac{1}{\sqrt{1-\varepsilon^2}}-1\right)}$$

2

To understand the first line, note that the Taylor series expansion of $\frac 1 {(1 + a \epsilon)^2}$ around $0$ gives

$$\frac 1 {(1 + a \epsilon)^2} = \sum \limits _{n=0} ^\infty (-1)^n (n+1) a^n \epsilon ^n .$$

In this, take $a = \cos \theta$ and multiply the whole equality by $\sin ^2 \theta$ (note that inside the parantheses in the right-hand side there should be $1 - \cos^2 \theta$, produced by replacing $\sin ^2 \theta$, and that summation starts from $n=0$).

Note that, save for some multiplicative factors, the integral in the right-hand side is essentially $\int \limits _0 ^{2 \pi} \cos^n \theta {\rm d} \theta - \int \limits _0 ^{2 \pi} \cos^{n+2} \theta {\rm d} \theta$. In order to understand the passage to the second line, you'll have to evaluate these integrals. Fortunately, I don't have to type that, somebody else did it for me, just take a look at the solutions found there.

The following lines are just elementary algebraic manipulations, you shouldn't have any trouble with them.

Finally, to understand the passage from the last-but-one line to the last one, write first the Taylor expansion of $(1+y)^a$ around $0$ (note that $a$ need not be an integer, that's the whole point in fact!). You'll get that the $n$th derivative ($n \ge 1$) in $0$ is $a (a-1) (a-2) \dots (a-n+1)$ so that

$$(1+y)^a = 1 + \sum \limits _{n=1} ^\infty \frac 1 {n!} a (a-1) (a-2) \dots (a-n+1) y^n .$$

Now do the ultra-smart trick of letting $y = - \epsilon ^2$ and $a = -\frac 1 2$. Plug the resulting expression into the last line of your chain of equalities and after doing plenty of algebraic simplifications you should get the expression on the line preceding it.

The details of these calculations are not difficult, but numerous, and typing them here is not my idea of a nice Sunday evening. In any case, if you get really stuck, leave a comment - but please don't ask for pre-calculus-level help!

Alex M.
  • 35,207
  • Don't worry, I will do all by myself, I'm not a lazy person ^^ Thank you really much, you clarified me a couple of points! – Enrico M. Jan 24 '16 at 20:41