[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] GSL Multidimensional Minimizer accepting a matrix instead
From: |
Pritam Chanda |
Subject: |
Re: [Help-gsl] GSL Multidimensional Minimizer accepting a matrix instead of vector ? |
Date: |
Tue, 7 Feb 2012 14:20:42 -0500 |
Hi,
I am planning to do that, but my data matrix is pretty large (say
100000 x 10), packing and unpacking for each minimization will be
expensive.
-Babai
On Tue, Feb 7, 2012 at 2:16 PM, Rhys Ulerich <address@hidden> wrote:
>> I am trying to write a minimizer for a function that takes a data
>> matrix (N x p) and a weight vector (P-dimensional) and a class table
>> vector Y (N -dim). The Multidimensional Minimizer in GSL only allows a
>> vector data type for the functions gsl_multimin_function and
>> gsl_multimin_function_fdf. The two-dimensional paraboloid example
>> shows how to pass a single vector
>> of data (x,y). In my case I have a matrix. Is it possible to do such
>> minimization with gsl ?
>
> Can you pack your matrix, weight vector, and vector into a contiguous
> region of memory and then use a gsl_vector_view
> (http://www.gnu.org/software/gsl/manual/html_node/Vector-views.html)
> to view that memory as a single "vector" to accomplish what you want?
>
> - Rhys