[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] non-linear least square fitting
From: |
Patrick Alken |
Subject: |
Re: [Help-gsl] non-linear least square fitting |
Date: |
Sat, 18 Aug 2018 09:16:40 -0600 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 |
GSL does not currently support constrained optimization directly. You
could, however, introduce a large penalty if your parameters are outside
the desired range. In your residual function, check if x is outside
[x_min,x_max] and set the output f_i to a large value. This will
discourage the solver from exploring that part of parameter space.
Patrick
On 08/18/2018 05:14 AM, Brijesh Upadhaya wrote:
> Hi,
>
> I am new to GSL and I want to use non-linear least-square fitting to fit a
> data set with 2000 data points. The function has 14 parameters in total. I
> tried using gsl_multifit_nlinear_driver () to solve the problem.
>
> It works nice but I want the parameters to be in certain range: x_min <=
> xpar <= x_max. How to fix the lower and upper bounds (Or I missed something
> in the gsl-2.5 documentation!). Some of the parameters in my model don't
> have any physical meaning if not bounded.
>
> I have another question as well!. I want to do a multi-objective
> optimization using GSL-2.5 but I couldn't locate any routine in the
> documentation. I have 40 data set having 2000 points each and I need to
> obtain 14 parameters (Global minimum).
>
> with kind regards,
> Brijesh