[SciPy-user] Numerical Recipes robust fit implementation

[dmitrey]

my 2 cents:
openopt ralg solver works good to (currently unconstrained only) 
problems like F(x) = sum(abs(f[i](x))) (or other non-smooth funcs; 
convexity is highly preferable)
and it can use F(x) 1st derive if provided by user

however, ralg is intended for medium-scale problems only (nVars up to 
1000), and it is missing patterns yet, I guess it could be nice for your 
problem

HTH, D.

Matthieu Brucher wrote:
> Hi,
>  
>
>     Ultimately I hope to determine the probability of a
>     a given data set being exponentially distributed, 
>
>
>
>
> In this case, you could use a non quadratic fit function, with a 
> pseudo norm1 cost ( for instance sqrt(eps + (y-f(x))²) with a small 
> eps so that the cost is derivable and a gradient descent should do the 
> trick).
>
> Matthieu
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>