1

I am given the equation of a nonuniform oscillator as $$\dot\theta=\omega-a\sin(\theta)\tag{1}\label{eq1},$$ where $a<\omega$, and I'm told that the period, $T$, of this is given by $$T=\frac{2\pi}{\sqrt{\omega^2-a^2}}\tag{2}\label{eq2}.$$ To work this out, I first separate the variables for $\eqref{eq1}$ to get $$\frac{1}{\omega-a\sin(\theta)}d\theta=dt.$$ I know that for the time to increase from $0$ to $T$, then the oscillator has gone from $0$ to $2\pi$. Therefore $$\int_0^{2\pi}\frac{1}{\omega-a\sin(\theta)}d\theta=\int_0^Tdt$$ and so $$T=\int_0^{2\pi}\frac{1}{\omega-a\sin(\theta)}d\theta.$$ This is where the fun begins. I use the Weierstrass substitution of $u=\tan\left(\frac{\theta}{2}\right)$ to evaluate this which changes my integral to $$T=\int_{\theta=0}^{\theta=2\pi}\frac{2}{\omega u^2+\omega-2ua}du.$$ I take the factor of $2$ outside of the integral and complete the square of the quadratic in the denominator to arrive at $$T=2\int_{\theta=0}^{\theta=2\pi}\frac{1}{\omega\left(\left(u-\frac{a}{\omega}\right)^2+1-\frac{a^2}{\omega^2}\right)}du.$$ I factor out $\frac{1}{\omega}$ from the integral, and use the substitutions $x=u-\frac{a}{\omega}$ and $y=1-\frac{a^2}{\omega^2}$ to make this $$T=\frac{2}{\omega}\int_{\theta=0}^{\theta=2\pi}\frac{1}{x^2+y}dx.$$ I then use the substitution $x=\sqrt{y}\tan(v)$, meaning $x^2=y\tan^2(v)$ and $dx=\sqrt{y}\sec^2(v)dv$. This makes my integral $$T=\frac{2}{\omega}\int_{\theta=0}^{\theta=2\pi}\frac{\sqrt{y}\sec^2(v)}{y\tan^2(v)+y}dv.$$ I take out a factor of $\frac{\sqrt{y}}{y}$ and find the following: $$T=\frac{2\sqrt{y}}{\omega y}\int_{\theta=0}^{\theta=2\pi}\frac{\sec^2(v)}{\tan^2(v)+1}dv=\frac{2\sqrt{y}}{\omega y}\int_{\theta=0}^{\theta=2\pi}\frac{\sec^2(v)}{\sec^2(v)}dv=\frac{2\sqrt{y}}{\omega y}\int_{\theta=0}^{\theta=2\pi}1 dv=\frac{2\sqrt{y}}{\omega y}\left[v\right]^{\theta=2\pi}_{\theta=0}.$$ Then I start the process of re-inserting my many substitutions. First $v=\tan^{-1}\left(\frac{x}{\sqrt{y}}\right)$, then $x=u-\frac{a}{\omega}$, then $u=\tan\left(\frac{\theta}{2}\right)$, and finally $y=1-\frac{a^2}{\omega^"}$. This gets me $$T=\frac{2\sqrt{1-\frac{a^2}{\omega^2}}}{\omega\left(1-\frac{a^2}{\omega^2}\right)}\left[\tan^{-1}\left(\frac{\tan\left(\frac{\theta}{2}\right)-\frac{a}{\omega}}{\sqrt{1-\frac{a^2}{\omega^2}}}\right)\right]^{2\pi}_{0}.$$ I can simplify this slightly. Firstly, $$\frac{2\sqrt{1-\frac{a^2}{\omega^2}}}{\omega\left(1-\frac{a^2}{\omega^2}\right)}=\frac{2}{\omega\sqrt{1-\frac{a^2}{\omega^2}}}=\frac{2}{\omega\sqrt{\frac{\omega^2-a^2}{\omega^2}}}=\frac{2}{\sqrt{\omega^2-a^2}},$$ and secondly $$\tan^{-1}\left(\frac{\tan\left(\frac{\theta}{2}\right)-\frac{a}{\omega}}{\sqrt{1-\frac{a^2}{\omega^2}}}\right)=\tan^{-1}\left(\frac{\omega\tan\left(\frac{\theta}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right).$$ So now I have $$T=\frac{2}{\sqrt{\omega^2-a^2}}\left[\tan^{-1}\left(\frac{\omega\tan\left(\frac{\theta}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)\right]^{2\pi}_{0}.$$ I feel like I'm so very close to $\eqref{eq2}$ now, but this is where I become stuck. I know that $\tan\left(\frac{2\pi}{2}\right)=\tan\left(\frac{0}{2}\right)=0$, which means when I evaluate the square brackets between $0$ and $2\pi$ they just cancel out and I'm left with $$T=0.$$

Could someone please tell me where I have gone wrong, or what I am missing here? Been stuck on this one all day to no avail!

Nemon27
  • 199
  • 2
    I haven't read all the details of your question, but since your substitution $u=\tan(\theta/2)$ has a singularity inside the interval of integration (at $\theta=\pi$), I'm thinking that you might be experiencing something similar to what was asked about here: https://math.stackexchange.com/questions/1356523/what-are-the-restrictions-on-using-substitution-in-integration (and in the “linked questions” there). – Hans Lundmark Jan 14 '21 at 20:17
  • @Hans Thank you so much for this suggestion! Do you know if my solution is correct? I would be happy to gift you some rep in return for your feedback! – Nemon27 Jan 14 '21 at 21:25

3 Answers3

3

Note that $$ T= \int_0^{2\pi}\frac{2e^{-i\theta}}{2i\omega-a(e^{i\theta}-e^{-i\theta}) }ie^{i\theta}\,d\theta= \int_0^{2\pi}\frac{2}{2i\omega e^{i\theta}-a(e^{2i\theta}-1) }ie^{i\theta}\,d\theta=\int_\gamma f, $$ with $\gamma\colon[0,2\pi]\to\mathbb C$ given by $\gamma(\theta)=e^{i\theta}$ and $f(z)=2/(-az^2+2i\omega z+a)$. By the Residues' theorem: $$ T=2\pi i \,{\rm Res}\left(f,\frac{i(\omega+\sqrt{\omega^2-a^2})}{a}\right)=2\pi i\frac{2}{-a}\cdot\frac1{-\frac{2i\sqrt{\omega^2-a^2}}{a}}= \frac{2\pi}{{\sqrt{\omega^2-a^2}}} $$ since the poles satisfy $|i(\omega+\sqrt{\omega^2-a^2})/a|<1$ and $|i(\omega-\sqrt{\omega^2-a^2})/a|>1$.

John B
  • 16,854
  • Hi John, unfortunately a lot of the maths you've used here is beyond my own knowledge - I'm not sure how you've inserted $i$ terms into the equation or what Residues' theorem is. Seems terribly clever though so I'll look it up and try to understand it. If you could let me know if the method in my own answer is okay I'd be happy to gift you some rep and accept your answer though! – Nemon27 Jan 15 '21 at 09:50
  • I regret that you don't know this material, but usually it's the best approach to "trigonometric" integrals (really it's very simple after you learn what are isolated singularities of a complex function and the Residues' theorem). Sorry then. Sure, your approach is quite fine, we just need to find where is the mistake. :) – John B Jan 15 '21 at 11:13
  • so my approach of taking limits is incorrect? I'm not sure if you're only refering to the method in my question or the answer I've posted? But thank you for the help! I will definitely investigate the ideas you have suggested, especially if they're widely applicable to other integrals :) – Nemon27 Jan 15 '21 at 11:33
2

$\newcommand{\bbx}[1]{\,\bbox[15px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\on}[1]{\operatorname{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ \begin{align} T & = \bbox[5px,#ffd]{\left.\int_{0}^{2\pi}{\dd\theta \over \omega - a\sin\pars{\theta}} \,\right\vert_{\, \omega\ >\ a\ \geq\ 0}} = \int_{-\pi}^{\pi}{\dd\theta \over \omega + a\sin\pars{\theta}} \\[5mm] & = \int_{0}^{\pi}\bracks{{1 \over \omega + a\sin\pars{\theta}} + {1 \over \omega - a\sin\pars{\theta}}}\dd\theta \\[5mm] & = 2\omega\int_{0}^{\pi}{\dd\theta \over \omega^{2} - a^{2}\sin^{2}\pars{\theta}} = 2\omega\int_{-\pi/2}^{\pi/2}{\dd\theta \over \omega^{2} - a^{2}\cos^{2}\pars{\theta}} \\[5mm] & = 4\omega\int_{0}^{\pi/2}{\dd\theta \over \omega^{2} - a^{2}\cos^{2}\pars{\theta}} = 4\omega\int_{0}^{\pi/2}{\sec^{2}\pars{\theta} \over \omega^{2}\sec^{2}\pars{\theta} - a^{2}}\,\dd\theta \\[5mm] & = 4\omega\int_{0}^{\pi/2}{\sec^{2}\pars{\theta} \over \omega^{2}\tan^{2}\pars{\theta} + \omega^{2} -a^{2}}\,\dd\theta \\[5mm] & = 4\omega\,{1 \over \omega^{2} - a^{2}}\,{\root{\omega^{2} - a^{2}} \over \verts{\omega}} \ \times \\[2mm] & \int_{0}^{\pi/2}{\verts{\omega}\sec^{2}\pars{\theta}/\root{\omega^{2} - a^{2}} \over \bracks{\verts{\omega}\tan\pars{\theta}/\root{\omega^{2} - a^{2}}}^{2} + 1}\,\dd\theta \end{align}


Sets $\ds{t = \verts{\omega}\tan\pars{\theta}/\root{\omega^{2} - a^{2}}}$: \begin{align} T & = \bbox[5px,#ffd]{\left.\int_{0}^{2\pi}{\dd\theta \over \omega - a\sin\pars{\theta}} \,\right\vert_{\, \omega\ >\ a\ \geq\ 0}} = {4\on{sgn}\pars{\omega} \over \root{\omega^{2} - a^{2}}}\ \underbrace{\int_{0}^{\infty}{\dd t \over t^{2} + 1}} _{\ds{\pi/2}} \\ & = \bbx{2\pi \over \root{\omega^{2} - a^{2}}} \\ & \end{align}
Felix Marin
  • 89,464
  • Hi Felix, I think I just about follow your method until the end but then I get lost, I'm not familiar with the $\text{sgn}$ function. I would be happy to accept your answer, even gift some rep, if you could let me know if the method in my own answer is mathematically sound? – Nemon27 Jan 15 '21 at 12:08
  • @Nemon27 $\operatorname{sgn}\left(x\right) = \left{\begin{array}{rcl} {\displaystyle -1} & \mbox{if} & {\displaystyle x < 0} \ {\displaystyle 0} & \mbox{if} & {\displaystyle x = 0} \ {\displaystyle 1} & \mbox{if} & {\displaystyle x > 0} \end{array}\right.$ – Felix Marin Jan 16 '21 at 19:59
  • @Nemon27 Problem with your answer : Inverse Trigonometric functions are not, in general, a "one to one" functions unless they are confined to some intervals like, for instance, $\displaystyle\left[0,\pi/2\right]$. That's the reason I started to move everything toward that interval. In such a case, $\displaystyle\arcsin\ \mbox{or/and}\ \arccos$ behave as an "one to one" function as I already mentioned above. – Felix Marin Jan 16 '21 at 20:05
0

Taking on board the comment of @Hans Ludmark, I believe I have found the answer but I'm not entirely sure if it is rigorous. Considering the discontinuity at $\theta=\pi$ I now re-examine the equation

$$T=\frac{2}{\sqrt{\omega^2-a^2}}\left[\tan^{-1}\left(\frac{\omega\tan\left(\frac{\theta}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)\right]^{2\pi}_{0}.$$

I have split the brackets to be evaluated into two. I now write this as $$T=\frac{2}{\sqrt{\omega^2-a^2}}\left(\left[\tan^{-1}\left(\frac{\omega\tan\left(\frac{\theta}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)\right]^{\pi}_{0}+\left[\tan^{-1}\left(\frac{\omega\tan\left(\frac{\theta}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)\right]^{2\pi}_{\pi}\right),$$ where the bracket on the left is to the left of $\theta=\pi$ and the bracket on the right is to the right of $\theta=\pi$. I then write this as $$T=\frac{2}{\sqrt{\omega^2-a^2}}\left(\tan^{-1}\left(\frac{\omega\tan\left(\frac{\pi}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)^--\tan^{-1}\left(\frac{\omega\tan\left(\frac{0}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)+\tan^{-1}\left(\frac{\omega\tan\left(\frac{2\pi}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)-\tan^{-1}\left(\frac{\omega\tan\left(\frac{\pi}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)^+\right)$$ where a '$-$' superscript denotes the term on the left of $\pi$ and a '$+$' superscript denotes the term on the right of $\pi$. The terms $$-\tan^{-1}\left(\frac{\omega\tan\left(\frac{0}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)$$ and $$+\tan^{-1}\left(\frac{\omega\tan\left(\frac{2\pi}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)$$ cancel since $\tan\left(\frac{0}{2}\right)=\tan\left(\frac{2\pi}{2}\right)=0$ This leaves $$T=\frac{2}{\sqrt{\omega^2-a^2}}\left(\tan^{-1}\left(\frac{\omega\tan\left(\frac{\pi}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)^--\tan^{-1}\left(\frac{\omega\tan\left(\frac{\pi}{2}\right)-a}{\sqrt{\omega^2-a^2}}\right)^+\right)$$ For the term on the left of $\pi$, we can take the limit to find $$\lim\limits_{x\to\frac{\pi}{2}^-}\tan(x)=\infty,$$ and for the term on the right of $\pi$ we take the limit to find $$\lim\limits_{x\to\frac{\pi}{2}^+}\tan(x)=-\infty.$$ This reduces our equation to $$T=\frac{2}{\sqrt{\omega^2-a^2}}\left(\tan^{-1}(\infty)-\tan^{-1}(-\infty)\right).$$ Taking limits again we find that $$\lim\limits_{x\to\infty}\tan^{-1}(x)=\frac{\pi}{2}$$ and $$\lim\limits_{x\to-\infty}\tan^{-1}(x)=-\frac{\pi}{2}.$$ Hence: $$T=\frac{2}{\sqrt{\omega^2-a^2}}\left(\frac{\pi}{2}--\frac{\pi}{2}\right)=\frac{2\pi}{\sqrt{\omega^2-a^2}}.$$

Could someone let me know if this is a sound method?

Nemon27
  • 199