Re: [Help-gsl] solution of 3d quadratic algebraic system

[Rhys Ulerich]

Illes,

> The 3 variables represent biochemical concentrations normalized to the
> interval [0,1]. Can I tell the solver (or another solver) that it should
> stay inside this interval? I keep receiving solutions outside this interval,
> which are mathematically OK, but biochemically really nonsense.

Ah... constrained Newton iteration is another matter.  I'd suggest
doing some reading to find appropriate numerical techniques.  I am
uncertain if GSL has a canned solution for you.  Googling or hitting
the journals for "Newton iteration inequality constraint" should give
you lots of places to look.

>From reading other responses to this message, it does seem like
futzing by adding penality terms could get somewhere.  If you do get
it to work, you'll have a scheme that'll solve your specific
three-concentration problem.  If you learn and implement the general
constrained approach, then you'll know a scheme that'll solve the
whole class of problems.

As a last thought, C and GSL probably aren't the best tools if you're
exploring algorithms.

- Rhys