For the spherical pendulum, with constant r = l, we thus have
Thus
is the conserved energy.
where
depends only on
and the constant
. Recalling that
varies from 0 (upright) to
(hanging vertically down), the two component
terms are drawn below.
When
, the gravitational component is the only term and our
equilibrium point is at
(hanging straight down). As
increases, the effective potential is a sum of both potential
terms. Since the
term diverges at
, the
effective potential minimum can no longer be there, for nonzero
. Instead, summing both contributions gives an effective
potential whose minimum
is shifted lower: shifting
further and further to the left (lower values) for
greater and greater values of
, until when the
component totally dominates,
(that is, the
spherical pendulum rotates totally horizontally). Thus as
increases, the constant
motions go from hanging down
stationary (
), to rotating about the z-axis at fixed
, with the angle
rising closer and closer to the
horizontal as the rotation frequency
(hence
)
increases. This effective potential and resulting motions are sketched
below.
(this is seen most easily by
applying
in spherical coordinates, then using
our constraint.) Secondly, in cylindrical coordinates, infinitesimal
distances are
We thus have, applying our constraint
,
This Lagrangian has partial derivatives
giving Euler-Lagrange equations
Thus
is the conserved energy.
The equations on the right just reproduce the relations between
and
,
and
. Using them to eliminate momenta and substitute into the Hamilton's equations on the left gives
reproducing the Euler-Lagrange equations.
which has asymptotes
decreasing from infinity for small
r, and
approaching zero from below as
. In between the decreasing behavior at small r
and the increasing behavior at large r must be a turnover point
where
has a minimum. The potential thus looks as sketched
above, with bound orbits for
and scattering orbits for
.
giving a minimum at
Thus
Thus
for positive
.
thus has a minimum at
Generally
which we simplify to
At
we thus have
which
is always positive for
, indicating stable circular
orbits (since deviations
from
have restoring force
).We thus have a frequency for radial oscillations
But by angular momentum conservation
for a circular orbit at
, so
Thus for our cases we have relations between the periods
and
. This tells us how many cycles of the sinusoidal
variation in r we complete per
-revolution in
, as
drawn below.