Re: Contributing a kron function

[heberf]

I haven't done any of that unfortunately but I'd be happy to help.

Heber

On Fri, 22 Oct 1999, etienne grossmann wrote:
>   Hello,
> 
>   let's say I were to contribute a builtin kron function, based on
> Heber's code :
> 
> ======================================================================
> #include <octave/oct.h>
> #include "config.h"
> #include "oct-obj.h"
> #include "oct-map.h"
> #include "defun-dld.h"
> #include "toplev.h"
> 
> DEFUN_DLD (newkron, args, ,"Faster Kronecker Product")
> {
> 
>   octave_value_list retval;
> 
>   Matrix A = args(0).matrix_value();
>   Matrix B = args(1).matrix_value();
>   
>   int ra = A.rows();
>   int ca = A.columns();
>   int rb = B.rows();
>   int cb = B.columns();
>   int r = ra*rb;
>   int c = ca*cb;
>   Matrix tmpmat(rb,cb,0);
>   Matrix C(r,c,0);
>   for (int i=0; i<ra; i++)
>     {
>       for (int j=0; j<ca; j++)
>         {
>           tmpmat = A(i,j)*B;
>           C = C.insert(tmpmat,i*rb,j*cb);
>         }
>     }
>   retval(0) = octave_value(C);
> return retval;
> 
> }
> ======================================================================
> 
>   I guess I would have to add another function for 2 ComplexMatrix
> arguments. Right? 
> 
>   And put it in a "src/DLD-FUNCTIONS/kron.cc"
> 
>   And add kron.cc to the definition of  DLD_XSRC in src/Makefile.in
> 
>   Does that sound correct? Is there a better place (e.g. in data.cc)
> to define such a function?
> 
>   Other things that could be useful would include a little test suite,
> I guess (any preferred format for that?).
> 
>   Heber, have you already done a all/some of the above?
> 
>   Cheers,
> 
>   Etienne
>