Grand Challenge Problems

[David G. Hough on validgh]

Tim Peters mentioned these problems which are supposed to be keeping supercomputer
vendors in business for the next few years (supercomputer == system with massive
memory and I/O bandwidth => massively parallel).   Taxpayers will
fund attacks on the following:

	climate modeling
	semiconductor and superconductor modeling
	quantum chromodynamics
	mapping the human genome

NSF is asking for $638 million to fund these activities in FY 92.  It's in the
May 1991 SIAM News.

There's also another article with more details on quantum chromodynamics (QCD).
Turns out the underlying operators are 3x3 complex matrices, although I don't
recall that the QCD program in the PERFECT suite depends on complex arithmetic
much; it must all have been expanded into real arithmetic in that program.  Integrals
of functions with exponential factors all figure in, too.  Because of the size
of the lattices of points to be integrated, however, memory and I/O bandwidth
look like the limiting factors on QCD calculations.