0
$\begingroup$

I have a Hamiltonian with two degrees of freedom, and when I change to action angle variables, one of the action variables does not appear in the final Hamiltonian. The reason seems to be because the redundent action variable shifts the area of the second variable and therefore does not appear in the final energy expression.

I have something of the form

$$H(x_1,x_2,p_1,p_2)=2-\cos\left(\frac{2\pi}{L_1}p_1+\frac{2\pi\alpha x_2}{L_2}\right)-\cos\left(\frac{2\pi }{L_1}p_2\right)$$ where $p_i\in[0,L_1)$, $x_i\in[0,L_2).$ Since $x_1$ is cyclic $p_1=\kappa$ is constant therefore the first action variable is therefore $$J_1=\frac{1}{2\pi}\oint p_1dx_1=\frac{\kappa L_2}{2\pi}.$$ Because of the periodic nature of the problem the orbits hit the boundary and continue to form a loop. The turning points are $$x_m^{\pm}=-\frac{L_2\kappa}{a L_1}\pm\frac{L_2}{2\pi a}\arccos(1-E)+\frac{L_2 m}{a}.$$ The left and right turning points for an orbit with energy $E=0.44$ are $x_0^-$ and $x_0^+$ respectively. Calculating the remaining action variable we find $$J_2=\frac{1}{2\pi}\oint p_2dx_2=2L_1(x_0^+-x_0^-)-\int_{x_0^-}^{x_0^+}p'_2dx_2$$ where $p'_2=\frac{L_1}{2\pi}\arccos(2-E-\cos(\frac{2\pi}{L_1}\kappa+\frac{2\pi a}{L_2}x_2)).$ On then see that changing variables, $\kappa$ is eliminated from $J_2$. Therefore, $$H(J_1,J_2)=H(J_2)=E.$$ Do I then treat this as a one dimensional classical system? From $$\omega_1=\frac{\partial H}{\partial J_1}=0,$$ this does seem that the motion is indeed around the $J_2$ torus. Is this correct? Does anybody know of a similar problem in classical mechanics? I am also interested in the quantum spectra of such a system, this seems to imply that the energy contour has no curvature, this would affect the eigenvalues available, and cause degeneracy... Any feedback would be greatly appreciated.

$\endgroup$
2
  • $\begingroup$ by $x_0^-$ and $x_0^+$ do you mean $x_1^-$ and $x_1^+$? what is the subscript $m$ in your expression for $x_m^\pm$? Do you not mean $\omega_2=\partial H/\partial J_2$? $\endgroup$ Commented Aug 29, 2020 at 14:12
  • $\begingroup$ The $x^\pm$ could be for different $m$s but I think the answer is the same. The $J_1$ co-ordinate still comes out of the $J_2$. Since it is a two dimensional system, compute the periodic orbits, $\omega_1$ and $\omega_2$ needs to be calculated. But $\omega_1=0$ and $\omega_2\neq 0$. My question is basically, after the canonical transformation to action angle variables can I not just treat this system as a one dimensional system, now depending on one action variable $J_2$? And what would this mean for the eigenvalues? I think degeneracy? But not sure how to go about doing this... $\endgroup$
    – lpphd
    Commented Aug 29, 2020 at 15:40

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.