Hi.
There seems to be some issues with the maxima lsquares module.
1) The examples in the maxima manual are not correct. They are missing
load("mnewton") and there seems to be a problem with
keepfloat setting in mnewton.mac (this is a bug probably somewhere else
though). I can get it to work if I use keepfloat:false instead of
the original true in mnewton.mac. I reported to the bugs page as well;
under Linux with 5.12.0 and CVS / clisp and gcl 2.6.7. Also lsquares.mac
has the mnewton load commented out at the beginning (why?).
2) I decided to hack the lsquares module a bit in order to use it with
BFGS and have a nicer output from it. The behavior can be controlled
with parameter settings (see the modified lsquares.mac attachment). The
default is to do things the old way (to be compatible with earlier apps).
More specifically, it is now possible to get the fitted function and the
parameters separately as a list.
Jussi Eloranta
Cal State Northridge
ps. For give me, if I am missing something - I am a newbie to maxima
programming...
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: lsquares.mac
Url: http://www.math.utexas.edu/pipermail/maxima/attachments/20070520/0ae9df5c/attachment.diff