MLH wrote:
Just didn't wann reinvent the rolly thing.
I'll look around at home. The VBA version I wrote is more robust than
standard Gaussian Elimination because it implements full pivoting.
When using matricies for things like linear regression, numerical
instabilities often crop up because those matricies typically have very
small determinants. The pivoting scheme gets around that problem. I
just hope the magnetic charge on that old floppy disk hasn't made it
unreadable. If you don't need pivoting, look under 'Gaussian
Elimination' in any Numerical Methods book to get an algorithm.
James A. Fortune