Re: can BGFS be made to return the approximate hessian?

[John Lamb <J dot D dot Lamb at btinternet dot com>]

Steven Scott wrote:
> For nonlinear statistical models the BGFS algorithm is often used to
> find the mode of the likelihood function (and thus the MLE point
> estimate).    The BGFS algorithm constructs an approximate hessian
> matrix internally, but the GSL implementation of BGFS does not seem to
> expose the approximate hessian to users.  

If you personally need this, I have some C++ functions that may help. 
One separately creates an initial Hessian estimate and the other does a 
perturbed Cholesky decomposition. Both work with gsl_matrix pointers. 
I've used these without problems on a large number of problems with 
matrices with as many as 1800 rows.

--
JDL