Back

Action-Angle Variables

For a bound one-degree-of-freedom system the natural coordinates are not position and momentum but the action J=12πpdqJ=\tfrac1{2\pi}\oint p\,dq (the phase-space area the orbit encloses, divided by 2π2\pi) and an angle θ\theta that simply winds at the constant rate θ˙=ω(J)=H/J\dot\theta=\omega(J)=\partial H/\partial J. The left panel is the phase orbit with that enclosed area shaded; the right is the action-angle picture, where the harmonic orbit becomes a circle of radius 2J\sqrt{2J} swept uniformly. The harmonic oscillator is isochronous (ω=ω0\omega=\omega_0 for every amplitude); the pendulum is anharmonic (ω\omega falls as the swing grows). Ramp ω0\omega_0 slowly and JJ barely moves while the energy tracks ω\omega: adiabatic invariance.

Figure 1. The J(t) strip (flat = the action is conserved), the phase orbit with its enclosed action area, and the action-angle loop of radius sqrt(2J). Potentials include a Kepler radial orbit; the ramp-speed slider perturbs the parameter, slow stays adiabatic, fast breaks the invariance. Method: contour action integral; velocity-Verlet evolution; the uniformly winding angle variable.
potential
energy E0.60
omega0 / L1.00
ramp speed0.00

WHAT TO TRY

  • The left phase orbit and the right action-angle loop are the same motion in two coordinate systems: the canonical transformation turns any bound orbit into a circle of radius sqrt(2J) that the angle sweeps at constant rate.
  • Push the ramp slider up to drive the frequency: slow ramps keep the bottom J(t) line flat (the adiabatic invariant holds), but a fast ramp makes it drift and the verdict turns red.
  • Switch the potential to pendulum, quartic, or the Kepler radial orbit: the phase orbit changes shape, yet the action-angle picture is still a clean circle swept uniformly.