On 7/21/07, Shaikh Surabuddin <ssuhrab at yahoo.com> wrote:
> (%i1) load("mnewton")$
> (%i2) load("lsquares")$
> (%i3) lsquares(matrix([1.1,7.1],[2.1,13.1],[3.1,25.1],[4.1,49.1]), [x,y], y=a*b^x+c, [a,b,c], [5,5,5]);
> quotient is not exact
It appears that this error message is generated from linsolve
(which is eventually called from lsquares).
A little experimentation seems to show that if the keepfloat
flag is false (its default value) then the error goes away.
You can edit lsquares.mac and mnewton.mac (which are somewhere
in your Maxima installation files) to assign keepfloat:false or just cut
out the keepfloat assignment.
Or you could apply rat to the data (which changes all floats to
rationals) before calling lsquares; I didn't try that.
I am working on a revision of lsquares which, I hope, will be
an improvement on the existing version.
Hope this helps,
Robert Dodier