You are on page 1of 2

Additional note for problems 4 and 5

Problems 4 and 5 in the tutorial on phase space deals with obtaining information about
the potential energy and force from a given phase trajectory. In problem 4 we derive the
potential energy for the system from a given phase trajectory. However, this should not
give an impression that it is possible to do so in general. To understand it better, consider
one of the phase trajectories for the quartic potential discussed in problem 3:

From this picture, it is clear that the force vanishes at the origin as well as around the
points 2. Moreover, since the kinetic energy (KE) increases as one moves away from the
origin, the force is repulsive around this point. Whereas the KE decreases as one moves
away from x = 2, thus the force is attractive around this point. This shows that the
origin is a local maximum and x = 2 are local minima, and the potential has no other
extremum in the range 3 < x < 3.
One the other hand, consider a system with potential energy V (x) = x(1 x)2 . The
phase portrait for the system is given by the following figure:

However, it is impossible to deduce even the qualitative behaviour of this potential energy
by looking at the phase trajectory ABC (or any of the oval shaped trajectories with centre

1
around x = 1/3). To obtain the qualitative behaviour of this potential one need to analyse
the entire phase portrait of the system. One can often derive some information on the
potential energy corresponding to a particular phase trajectory. However, this may exhibit
only the local behaviour of the potential energy. To understand its global behaviour in a
qualitative manner, one needs to analyse the complete phase diagram (phase portrait) of
the corresponding physical system.
Likewise, in problem 5, it is incorrect to conclude that the force is nonconservative
purely from the expression for acceleration a(t) = t, as this particular trajectory is obtained
starting from a specific initial condition. For instance, consider a simple harmonic motion
with x(0) = 0, x(0)
= v0 , which has solution x(t) = (v0 /) sin t. Naively, one may
argue that the corresponding force is explicitly time-dependent, as F = m x gives F (t) =
v0 sin t. However, this is not the case; by inverting the original relation, we can write
t = sin1 [x(t)/v0 ]. Substituting into the above expression for F then gives us F = 2 x
as we want it to be. The explicitly time-dependent force we constructed fails to predict
the time evolution of x(t) for any initial condition other than what we started with (e.g.,
check it for x(0) = x0 , x(0)
= 0).
The function t(x) in prolem 5 turns out to be multi-valued in the range x [1/3 : 1/3],
since it is the root of a cubic equation, and therefore F (x) F [t(x)] evaluated using
the above procedure is not a single-valued function. Note that in the SHM example,
by contrast, t(x) is again multi-valued, but F (x) is a single-valued function. In general,
whenever we can express the force as a single valued function of x the system is conservative.
Otherwise it must explicitly depend on time and/or velocity and the system can not be
conservative.
We would like to emphasise that one need not always prove the force to be a multivalued
function in order to deduce if it has explicit time dependence. How then, can we conclude
whether the force here is explicitly time-dependent or not? Two observations that can be
made by looking at the phase trajectory are crucial here:
(a) the phase trajectory crossing itself at the point [0,1]- this can happen only if the
force is multi-valued/time-dependent at that point.
(b) If t is regarded as time itself, we observe that the particle enters the region [-
1/3,1/3] at the point x = 1/3 with a non-zero velocity, turns around at x = 1/3, again
turns around at x = 1/3 before exiting the region at x = 1/3. This behaviour is
inconsistent with a time-independent potential energy U (x) and conserved total energy
E for the particle. Therefore, the force F (t) = t is the only one that can explain the
trajectory as described.
On the other hand, if we rotate the phase plane by 90 , and imagine the transformations
x y and x y, then the phase trajectory can be imagined as the separatrix for the
potential V (y) = (3y 2 2y 3 )/18. This transformation is explained in detail in the last two
pages of the solution.
In general, the phase trajectory need not even cross itself for the system to be non-
conservative as in the case of a damped harmonic oscillator. However, in all these cases, one
can clearly deduce from the corresponding phase trajectories that the KE is a multivalued
function of x. Whenever this happens, the system is non-conservative.

You might also like