Re: [ublas] Compilation error in assignment

From: Sourabh (sourabh_at_[hidden])
Date: 2007-02-28 06:16:02

On Wed, 28 Feb 2007, Gunter Winkler wrote:

> On Wednesday 28 February 2007 11:19, Sourabh wrote:
> > #include <boost/numeric/ublas/matrix_sparse.hpp>
> > #include <boost/numeric/ublas/io.hpp>
> >
> > using namespace boost::numeric::ublas;
> >
> > int main () {
> >     typedef mapped_matrix<double> Matrix;
> >     typedef mapped_matrix<Matrix> MyMatrix;
> >     MyMatrix m;
> >     MyMatrix n;
> >     MyMatrix l;
> >
> >     MyMatrix k = m + n;
> > }
> >
> > This program, on compilation gives the following error:
> 
> The error is thrown, because there is no comparison operator for matrices. 
> However, the sparse assignement must be able to decide whether an element of 
> MyMatrix is "zero" or not.
> 
> The construction of nested matrices is not supported by ublas. In order to 
> work, you Matrix-class has to behave like a scalar:
> 
> * comparison: operators ==, != (for sparse types)
> * useful default constructor (which has to construct the zero element)
> 
> mfg
> Gunter
> 
> 
> 
> 
Hi Gunter,
Thanks for replying.

I did not understand your statement (MyMatrix has to behave like a 
scalar). Can't ublas infer the type (vector) and drive the corresponding 
operator on the elements ?

Please explain.

 -- 
-- Sourabh