A one or two sentence summary of the paper

This paper describes a method for fluid simulation that doesn't require
very small time steps to remain stable, at the cost of some accuracy

A one or two sentence description of the problem the paper is trying to
solve.

How to keep fluids stable when their timesteps aren't proportional to the
distance between points (that is, they are larger than what's guaranteed
to be stable)

A one of two sentence description of the method used to solve the
problem.

Their method extended a normal simulation by, rather than simply

generating new densities and velocities by simple algebra, following the
current velocity backwards in time to find the simulation point the point
under investigation came from.

A list of short (approx one sentence) descriptions of the key ideas of
the paper.

Mathematically generating new densities and velocities based on the
current densities, velocities, and forces at every point led to unstable
fluids when time steps were large, due to numerical error accumulation
between simulation steps.  To counter this, this paper introduced a way
of using only values that already existed in the simulation, ensuring
that the total volume of the fluid is at least close to what it was at
the last time step.

Any questions that the paper raises for you. This might be some flaw that
they do not address, something that is unclear, some piece of background
knowledge that you don't have that would help you understand the paper,
...

My question was whether the step where they removed divergence from the
field already existed before this paper, but Yu-Chi answered that for me
today (it did).