Re: multidimensional root finding by homotopy continuation

[Jerome BENOIT <jgmbenoit at mailsnare dot net>]

Hello,

finally I decided to write down adhoc routines to find my root:
along the homotopy path the hybridsj solver (`gsl_multiroots.h').
Fortunately my routines work fine.
As a matter of fact my plan was to refine my roots with a CAS:
surpprisingly my CAS solver seems too weak to catch the roots of my system
of nonlinear equations:
is there somewhere a multiprecision version of the hybridsj solver 
or an efficent multiprecision mutidimension solver ?

Thanks in advance,
Jerome

Brian Gough wrote:
> Jerome BENOIT writes:
>  > curently I have to solve a parametrized square system of nonlinear equations
>  > for which an exact solution is known for a trivial parametrization:
>  > to get solutions for non trivial parameters, I plane to proceede by homotopy continuation:
>  > did any one try to do it using gsl tools ?
>  > does such an approach to solve square systems have its place in gsl ?
>
> I don't think NAG has any homotopy routines, so perhaps
> it's difficult to provide general purpose framework.