Subject: finding multiple roots of non-polynomials
From: Keith Wald
Date: Fri, 19 Aug 2005 12:43:42 -0700 (PDT)
Hey folks,
Thanks very much for quickly sending your advice!
Yeah, I realize the general problem is hard. I was
thinking in terms of the situation where the system is
stable, and the zeros are "almost periodic", usually
with some multiple of pi, like the usual
transcendentals that come from simple boundary value
problems. Typically, initial seed guesses of
something*pi +/- epsilon converge nicely with N-R or
even a simple minded binary search.
I'm not afraid to work a bit; I was just curious if
there was a canned routine, as that's usually
preferred.
Thanks,
Keith
--- Robert Dodier wrote:
> > These zeroes are generally transcendental
> > numbers. If you want to find approximate
> > zeros of functions (functions that maxima can
> numerically
> > evaluate), you can use some kind of iteration
> > (secant or newton) to converge to a root from
> > an initial guess.
>
> the script mnewton.mac in
>
http://cvs.sf.net/viewcvs.py/maxima/maxima/share/contrib/
> implements newton's method. it worked for me in some
> problems. a bug has been fixed in the cvs version,
> so don't use the one bundled with maxima 5.9.1.
>
> fwiw,
> robert dodier
>
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com