Matrix problems

[Ashton Shortridge <ashton@geog.ucsb.edu>]

Hello,

I'm performing gaussian simulation using Gstat. I'm having a problem with matrix
issues. If I set max above 4, Gstat will process for a bit and then crash,
giving me a non-positive definite matrix error. The command file looks like the
following:

#  Conditional Simulation on a mask map
#
data(elev): 'e:\envstats\s100.xyz', x=1, y=2, v=3,
  sk_mean=20, max=8, radius=200;
variogram(elev): 0.5 Nug(0) + 3 Sph(200);
mask:                 'e:\envstats\grd100.asc';
method: gs;
set nsim=10;
predictions(elev): 'e:\envstats\sims\k100sim';

The mask is simply a 100x100 grid. Interestingly, if I set max to, say, 4
(meaning only the nearest 4 points are chosen for identifying the value), the
script appears to run without any trouble. I'd appreciate any thoughts and
suggestions.

Regards,
Ashton Shortridge
ashton@geog.ucsb.edu