Action variables for the Kepler problem

  • Thread starter Thread starter etotheipi
  • Start date Start date
  • Tags Tags
    Kepler Variables
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
2 replies · 3K views
etotheipi
Homework Statement
Express the Hamiltonian of the orbit in terms of the action variables ##I_{\phi}## and ##I_r## and hence show that the orbit is closed
Relevant Equations
N/A
Let's consider the Hamiltonian $$H = \frac{1}{2m} p_r^2 + \frac{1}{2mr^2} p_{\phi}^2 - \frac{k}{r}$$where the generalised momenta are here ##p_r = m\dot{r}## and ##p_{\phi} = mr^2 \dot{\phi}## conjugate to the coordinates ##r## and ##\phi##. Since ##p_{\phi}## does not depend on ##\phi## it can easily be integrated to obtain the action variable ##I_{\phi} = p_{\phi}## (which is also by construction a constant of motion), however the action variable corresponding to ##r## proves more difficult. Since ##\partial H / \partial t = 0##, we have ##H \equiv E = \text{constant}## because ##dH/dt = -\dot{p}_i \dot{q}_i + \dot{p}_i \dot{q}_i + \partial H / \partial t = 0##. The integral for ##I_r## becomes$$I_r = \frac{1}{2\pi} \oint p_r dr = 2\times \frac{1}{2\pi} \int_{r_{\text{min}}}^{r_{\text{max}}} \sqrt{2mE + \frac{2mk}{r} - \frac{I_{\phi}^2}{r^2}} dr$$We must show this equals$$I_r = k\sqrt{\frac{m}{2|E|}} - I_{\phi}$$Two problems, I can't see how to do that integral, and we'd also need to somehow eliminate ##r_{\text{min}}## and ##r_{\text{max}}##. Anyway, the integral is probably more important right now... let ##A \equiv 2mE##, ##B \equiv 2mk## and ##C \equiv -I_{\phi}^2##, and then$$I_r = \frac{1}{\pi} \int_{r_{\text{min}}}^{r_{\text{max}}} \frac{\sqrt{Ar^2 + Br + C}}{r} dr$$I wonder if there's supposed to be a nice way of solving it, because all I tried was this$$I_r = \frac{1}{\pi} \int_{r_{\text{min}}}^{r_{\text{max}}} \frac{1}{r} \sqrt{A \left(r+ \frac{B}{2A} \right)^2 + \left(\frac{C}{A} - \frac{B^2}{4A^2} \right)} dr = \frac{1}{\pi} \int_{u(r_{\text{min}})}^{u(r_{\text{max}})} \frac{\sqrt{Au^2 + D}}{u - \frac{B}{2A}} du$$with ##D \equiv \left(\frac{C}{A} - \frac{B^2}{4A^2} \right)##, which does not seem to help. Maybe I could let ##u \equiv \sqrt{D/A} \sinh{\theta}##, in which case$$I_r = \frac{1}{\pi} \int_{\theta(r_{\text{min}})}^{\theta(r_{\text{max}})} \frac{\sqrt{Au^2 + D}}{u - \frac{B}{2A}} du = \frac{D}{\pi \sqrt{A}} \int_{\theta(r_{\text{min}})}^{\theta(r_{\text{max}})} \frac{\cosh^2{\theta}}{\sqrt{\frac{A}{D}} \sinh{\theta} - \frac{B}{2A}} d\theta$$but this just gets messier and messier. Is there a trick or something that would make the integral easier? Thank you in advance

EDIT: By the way, I forgot to write that also we're told that a useful result is$$\int_{r_{\text{min}}}^{r_{\text{min}}} \sqrt{ \left(1-\frac{r_{\text{min}}}{r} \right) \left(\frac{r_{\text{max}}}{r} - 1 \right) } dr = \frac{\pi}{2}(r_{\text{min}} + r_{\text{max}}) - \pi \sqrt{r_{\text{min}}r_{\text{max}}}$$although it's not obvious to me at all how that helps...
 
Last edited by a moderator:
on Phys.org
etotheipi said:
$$I_r = \frac{1}{\pi} \int_{r_{\text{min}}}^{r_{\text{max}}} \frac{\sqrt{Ar^2 + Br + C}}{r} dr$$

##r_\text{min}## and ##r_\text{max}## correspond to the roots of the argument of the square root in the integrand. Thus, the argument may factored to be proportional to ##(r - r_{\text{max}})(r - r_{\text{min}})##. Be careful with signs. For example, ##A## is a negative number.
 
  • Like
  • Love
Likes   Reactions: Delta2 and etotheipi
Ah perfect, that's the key insight! I suppose that would be because the roots of ##p_{r} = m\dot{r} = 0## correspond to ##r = r_{\text{max}}## and ##r = r_{\text{min}}## because these two are the local maximum and minimum with respect to ##t## in the coordinate ##r = r(t)##. So, using a funny letter for the proportionality constant to keep things interesting,$$2mEr^2 + 2mkr - I_{\phi}^2 = \Xi(r-r_{\text{min}})(r-r_{\text{max}}) = \Xi r^2 - \Xi(r_{\text{min}} + r_{\text{max}}) r+ \Xi r_{\text{min}} r_{\text{max}}$$Since ##r^2##, ##r## and ##1## are linearly independent functions and this holds for all ##r \in [r_{\text{min}}, r_{\text{max}}]## we have ##\Xi = 2mE = (-2mk)/(r_{\text{min}} + r_{\text{max}}) = (-I_{\phi}^2)/(r_{\text{min}} r_{\text{max}})##, which we keep for later. Back to the integral, since ##|r- r_{\text{max}}| = r_{\text{max}} -r \geq 0##,$$I_r = \frac{\sqrt{|\Xi|}}{\pi} \int_{r_{\text{min}}}^{r_{\text{max}}} \frac{1}{r} \sqrt{|(r-r_{\text{min}})(r-r_{\text{max}})|} dr = \frac{\sqrt{|\Xi|}}{\pi} \int_{r_{\text{min}}}^{r_{\text{max}}} \sqrt{\left(1-\frac{r_{\text{min}}}{r}\right) \left(\frac{r_{\text{max}}}{r} -1 \right)} dr$$And using the magic hint$$I_r = \frac{\sqrt{|\Xi|}}{\pi} \left(\frac{\pi}{2}(r_{\text{min}} + r_{\text{max}}) - \pi \sqrt{r_{\text{min}}r_{\text{max}}} \right) = \frac{\sqrt{|\Xi|}}{2}\left( r_{\text{min}} + r_{\text{max}}\right) - \sqrt{|\Xi| r_{\text{min}}r_{\text{max}} }$$and after a little more re-writing, and remembering that ##\Xi = - |\Xi| < 0##, as well as that ##E = -|E| < 0## and ##I_{\phi} = |I_{\phi}| > 0##,$$I_r = \frac{\sqrt{-2mE}}{2} \frac{-2mk}{2mE} - \sqrt{I_{\phi}^2} = k\sqrt{\frac{m}{2|E|}} - I_{\phi}$$Hence the Hamiltonian takes the form$$H = -\frac{mk^2}{2(I_r + I_{\phi})^2}$$which is symmetric in the two action variables, and thus the rate of change of both of the corresponding angle variables are equal, i.e. $$\dot{\theta}_r = \frac{\partial H}{\partial I_r} \equiv \frac{\partial H}{\partial I_{\phi}} = \dot{\theta}_{\phi}$$and since the frequencies of both of these variables in the phase space are the same, the orbit is closed.

Thanks you!
 
Last edited by a moderator:
  • Like
Likes   Reactions: TSny