On 7/13/07, Daniel Lakeland <dlakelan at street-artists.org> wrote:
> I assume you're looking for numerical results because you mention
> find_root. So you're looking for (s,t) such that
> [x(t),y(t)]=[u(s),v(s)] and t and s are within some known rectangle of
> the s,t plane. Could you perhaps rephrase this as a minimization
> process and use lbfgs?
I believe rephrasing a root-finding problem as a minimization
problem is generally a weak approach, although it might be
worth trying all the same.
The implementation of Newton's method in Maxima (mnewton)
is not very strong, I believe. I think further work on the numerical
solution of equations would be widely useful, so I encourage
all interested parties to consider it.
best
Robert Dodier