1

I am trying to solve a problem in which I have to prove that for a particle with mass $m$ in a potential $U=A|x|^n$, has a period of oscillation given by

$$\tau =\frac{2}{n} \sqrt\frac{2\pi m}{E} \left ( \frac{E}{A}\right)^{1/n} \frac{\Gamma(1/n)}{\Gamma(1/2+1/n)}$$

What I have tried is to write down the second law by using $F = -\frac{d}{dx}U$, and I got the form

$$\frac{d^2 x}{dt^2}+\frac{A}{m}n x|x|^{n-2} = 0 $$

At this point I am thinking in taking cases, for example when $n$ is an even positive integer with $n>2$ we can solve this, I used the Laplace transformation knowing the transformation

$$ L\{x^{n-1}\} = \Gamma(n)s^{-n} $$

also it's inverse have a similar form, but as a result I got a non-periodic function

$$x(t) = x(0)-\frac{A}{m}n\Gamma(n) \frac{t^{n+1}}{\Gamma(n+2)} $$

and I am kind of confused because I wasn't expecting to get this, I would like to know if this approach is correct or someone knows a betters way to tackle this problem.

  • 2
    Just an idea: from energy considerations one can deduce that the particle will oscillate, but not in a simple harmonic fashion. So to get the period one way is to take an initial condition of at rest at some $x > 0$, and then don't solve for the whole trajectory, but just aim to find out how long it takes to reach the location $x=0$. That will be one quarter of the period. – Andrew Steane Mar 17 '21 at 17:02
  • 1
    ... or (same idea another way) start at $x = 0$ with non-zero velocity, and find out how long it takes for the velocity to reach zero. – Andrew Steane Mar 17 '21 at 17:04
  • The Lplace transform of derivative shoudl involve the intital conditions, the intial position and intial velocity. – ytlu Mar 17 '21 at 17:05
  • @ytlu I made the assumption that $ x'(0) = 0 $ – Adán González Mar 17 '21 at 17:06
  • There is a neat way to deal with this kind of problem: the action variable. In the phase space $(x,p)$, the particle orbits around a closed trajectory $p^2/2m + A|x|^n = E$ clockwise and encloses an area $J \propto \sqrt{2mE}\cdot (E/A)^{1/n} \propto E^{1/2+1/n}$. Then the formula $\tau = \partial J/\partial E$ implies that $\tau \propto E^{1/n-1/2}$. Finding the precise expression for the proportional constant is an easy exercise. Tried to explain more details by posting as an answer, but failed since it is closed. You may refer to: https://physics.stackexchange.com/a/389454/100521 – L-C Nov 01 '21 at 22:17

2 Answers2

5

Assume the total energy of the particle is $E$, a conservative quantity: $$ E = \frac{1}{2}m v^2 + A|x|^n. $$

We can obtain the velocity as function of position: $$ v = \pm \sqrt{\frac{2(E-A|x|^n)}{m}} $$

Fot the periodic motion, the particle will run back and forth in between two turning points,

$$ x_r = \pm \left( \frac{E}{A}\right)^{1/n}. $$

The integral for the period $\tau$:

$$ \tau = 4 \int_0^{x_r} \frac{dx}{v(x)} = 4 \int_0^{x_r} \frac{\sqrt{m} dx}{\sqrt{2(E-A x^n)}} $$ The integral is over postive $x$, thus the absolute symbole may be removed.

Simplify the integrand: $$ \tau = 4 \frac{\sqrt{m}}{\sqrt{2E}} \int_0^{x_r} \frac{dx}{\sqrt{1-\frac{A}{E} x^n}} = 2 \frac{\sqrt{2m}}{\sqrt{E}} \left(\frac{E}{A}\right)^{1/n} \int_0^{1} \frac{d\xi}{\sqrt{1-\xi^n}} $$

WHere $\xi = \left(\frac{A}{E}\right)^{1/n} x$. It should have little problem to go from the last equation to the answer.

ytlu
  • 4,196
2

I think that you can obtain analytical solution for the period $~T$

follow the solution from @ytlu

with A and m equal 1 and omit the abs() function

the energy E is:

$$E=\frac{1}{2}\,v^2+x^n$$ with: $$E_0(v=0~,x=x_0)=E(v~,x)$$ you obtain $$v=\sqrt{2(\,x_0^n-x^n)}$$ thus for $~v=0~,x_T=x_0$ and the period T is: $$\frac{dx}{dt}=v(x)$$ $\Rightarrow$ $$T=4\,\int_{0}^{x_0}\left(\frac{1}{v(x)}\right)\,dx\tag 1$$

you can choose the initial condition arbitrary $~x_0=1~$

the evaluation of equation (1) : \begin{align*} &n=2 \qquad T=\sqrt(2)\,\pi\\ &n=3 \qquad T=\frac{2}{3}\sqrt{2}\,\mathcal{B}\left(\frac{1}{3}~,\frac{1}{2}\right)\\ &n=4 \qquad T=\frac{1}{2}\sqrt{2}\,\mathcal{B}\left(\frac{1}{4}~,\frac{1}{2}\right)\\ &n=5 \qquad T=\frac{2}{5}\sqrt{2}\,\mathcal{B}\left(\frac{1}{5}~,\frac{1}{2}\right)\\ &n=6 \qquad T=\frac{2}{6}\sqrt{2}\,\mathcal{B}\left(\frac{1}{6}~,\frac{1}{2}\right)\\ \end{align*} from here you can deduce that \begin{align*} &\text{if $n~$ is even}\qquad, T=\frac{1}{2}\sqrt{2}\,\mathcal{B}\left(\frac{1}{n}~,\frac{1}{2}\right)\\ &\text{if $n~$ is odd}\qquad, T=\frac{2}{n}\sqrt{2}\,\mathcal{B}\left(\frac{1}{n}~,\frac{1}{2}\right)\\ &\text{where}\\ &\mathcal{B}(x,y)=\frac{\Gamma(x)\,\Gamma(y)}{\Gamma(x+y)} \end{align*}

Eli
  • 11,878
  • 1
    You are right, I got the general solution by doing the integral (1) using the substitution $u=1-\xi^n$ and identify $B(x,y)$ after some simplification. But thanks anyway! – Adán González Mar 18 '21 at 18:04
  • @AdánGonzález you see the solution just after you evaluate the integral for different n. If you try to integrate your function $f(\xi)~$ you don't get a solution – Eli Mar 18 '21 at 18:09