Fluid Polygons and Polyhedra

Take a look at these fluid structures @ http://www-math.mit.edu/~bush/bells.html
























Presumably, they can be theoretically predicted & described by the Navier-Stokes equations. But they weren't; they were discovered experimentally.

There is clearly a lot of structure in these equations and neat new phenomena waiting to be discovered.

There must be more efficient ways of extracting structures out of the Navier-Stokes equations than the haphazard methods of searching for solutions that have been employed over the last 200 years.

Machine Learning any one?