Re: linalg[leastsqrs] in Maple V R4--success!!!

lcw1964 wrote:


The bottom line is thanks to assistance in this NG I have, as a
complete rank amateur, enjoyed a little bit of Maple success, and I
thought an expression of pleasure and gratitude was in order.


After all of this fussing I ended up opting for linalg[leastsqrs] after
all.

It is a lot faster and I have found that if set Digits high enough the
illconditioning in the intermediate matrix calculations seems
adequately compensated for my purposes. The procedure adds a whopping
30 to whatever I set for Digits to begin with, and even with this
compensation the computations are a lot faster and acceptable similar
to the SVD version.

Live and learn!

Les

.