[astro] success!

[Travis Desell]

I found a bug in the MPI code.  The sum of squares example is now converging
in 19 iterations (for 22 different parameters), as opposed to 700+ like 
it used to using
the conjugate gradient descent.  Gradient descent takes over 1500.  I'm 
still trying to
get the conjugate gradient hessian working, but this should be good 
enough for Nate
so long as he doesn't mind using the MPI version.

Once i get the hessian version working i'll port the code over into the 
salsa version, so
hopefully that will be working tomorrow night.

cheers,

-- 
----------------------------------------------------------------------------------------------------------
Travis Desell
<deselt @ cs.rpi.edu>
1-518-878-7424
Worldwide Computing Laboratory ( http://wcl.cs.rpi.edu/ )
Rensselaer Polytechnic Institute, 110 8th Street, Troy NY 12180, USA
----------------------------------------------------------------------------------------------------------