by Terence Tao

Cast a stone into a still lake. There is a large splash, and waves begin radiating out from the splash point on the surface of the water. But, as time passes, the amplitude of the waves decays to zero.
This type of behavior is common in physical waves,
and also in the partial differential equations used in
mathematics to model these waves. Let us begin with the
classical wave equation
The wave equation can be viewed as a special case of
the more general linear Klein–Gordon equation

An important way to capture dispersion mathematically is through the
establishment of dispersive inequalities that assert, roughly
speaking, that if a solution
One can represent any solution
On the other hand, solutions to the linear Schrödinger
equation

To reconcile this fact with the dispersive estimate, we
observe that solutions to dispersive equations such as
the linear
Schrödinger equation spread out in space as time
goes to infinity (much as the ripples on a pond do),
allowing the
The above analysis of the linear Schrödinger equation
relied crucially on having an explicit fundamental solution
at hand. What happens if one works with nonlinear (and
not completely integrable) equations, such as
Recall that in 1961, Morawetz proved the decay of
solutions to the classical wave equation in the presence
of a star-shaped obstacle. Morawetz used the “Friedrichs
abc method,” in which one multiplied both sides of a PDE
such as
In 1968, Morawetz applied this technique to study
solutions

Once one has some sort of decay estimate for a
dispersive equation, it is often possible to “bootstrap”
the estimate to obtain additional decay estimates. For
instance one might use the decay estimate one already
has to bound the right-hand side of a nonlinear PDE such as
An early result of this type was developed by Morawetz and
Strauss
in 1975. They showed that for any finite energy solution
In the decades since Morawetz’s pioneering work, many additional
Morawetz inequalities have been developed. For instance, in 1978,
Lin
and Strauss developed Morawetz inequalities for the nonlinear
Schrödinger equation, and Morawetz herself discovered further such
estimates for the wave equation outside of an obstacle. In more recent
years, “interaction Morawetz inequalities” were introduced, which
could control correlation quantities such as

One way to view Morawetz inequalities is as an assertion
of monotonicity of the radial momentum, which takes the
form
On the other hand, under hypotheses such as finite energy, this radial
momentum should be bounded. So by the fundamental theorem of
calculus, the time derivative of the radial momentum should have a
bounded integral in time. Intuitively, one expects this time
derivative to be large when the solution has a strong presence near
the origin, but not when the solution is far away from the origin.
Far from the origin the radial vector field
The Morawetz inequalities are indispensible as an ingredient in controlling the long-time behavior of solutions to a wide array of dispersive defocusing equations, including a number of energy-critical or mass-critical equations in which the analysis is particularly delicate and interesting; see for instance the texts [e1], [e3], [e4] for detailed coverage of these topics. They have also been successfully applied to many equations in general relativity (such as Einstein’s equations for gravitational fields), for instance to analyze the asymptotic behaviour around a black hole. The fundamental tools that Morawetz has introduced to the field of dispersive equations will certainly underlie future progress in this field for decades to come.