http://math.nist.gov/javanumerics/jama/

[uemcre@unc.edu]

I looked at the Cholesky program to see how it handled error conditions
and found that it only tests that the matrix is symmetric.  If the rest of
the package is anything like this, it certainly is amateurish and not what
I would expect from the Bureau.  

1. There is no reason to store the whole matrix and thus no reason to
check for symmetry.

2. You MUST check for singularity and positive-definiteness

The program fails on the matrix
 0 1
 1 1