[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] finite difference approximation of gradient BFGS
From: |
Patrick Alken |
Subject: |
Re: [Help-gsl] finite difference approximation of gradient BFGS |
Date: |
Mon, 19 Jun 2017 11:29:11 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.1.1 |
I'm not too familiar with the multimin algorithms, but it may be similar
to the nonlinear least squares code, which does implement a
finite-difference Jacobian. Have a look at multifit_nlinear/fdjac.c and
the corresponding documentation here:
http://www.gnu.org/software/gsl/doc/html/nls.html#c.gsl_multifit_nlinear_fdtype
Hope this helps,
Patrick
On 06/17/2017 08:36 AM, Info wrote:
> Hello,
>
> In R, when the gradient function is not provided a finite difference
> approximation is made. I am trying to replicate it with GSL but
> unfortunately I am not able to completely follow the internal code in
> 'optim.c'. I wonder if somebody could share some finite difference
> approximation code for:
> void (* df) (const gsl_vector * x, void * params, gsl_vector * g)
> void (* fdf) (const gsl_vector * x, void * params, double * f,
> gsl_vector * g)
>
> Many thanks.
>
>