Re: Non linear equations: howto?

[Søren Hauberg]

søn, 19 10 2008 kl. 16:40 +0200, skrev Fabrice DELENTE:
> On Sun, Oct 19, 2008 at 03:54:22PM +0200, Søren Hauberg wrote:
> > As a starting guess you use [-10; 10] which is a 2-vector. Your starting
> > guess should be a 5-vector, otherwise things just don't add up.
> 
> Ok, it works if I give [0;0;0;0;0] as a starting guess.
> 
> However, for different starting guesses, I get very different answers.
> Furthermore, the plot of f(x) with the guessed numbers doesn't go through
> the point I want...

Well, it's an iterative technique so, starting guesses matter. Does the
algorithm actually converge? If you call [x, val, info] = fsolve (@f,
x0), what value of 'info' do you get? 

If you're equation has many solutions, then fsolve will essentially find
the one closest to the initial guess.

I'm curious: f is a function of 5 parameters, so how do you plot it?

Søren
> 
> Any hint why? Thanks!
>