Re: [ublas] is something seriously wrong with zero_vector?

From: Michael Stevens (mail_at_[hidden])
Date: 2005-10-06 13:20:32

On Thursday 06 Oct 2005 16:26, Neal Becker wrote:
> I noticed that creating a zero_vector with some size, and then iterating
> over it produced strange results.  Then I noticed this:
>         BOOST_UBLAS_INLINE
>         const_iterator begin () const {
>             return const_iterator (*this);
>         }
>         BOOST_UBLAS_INLINE
>         const_iterator end () const {
>             return const_iterator (*this);
>         }

Funny coincidence, I was looking at the same code today.

> Excuse me?  Doesn't this mean begin() == end() and effectively size is
> always zero?

The iterators in zero_vector work fine. They do this by defining themselves as 
Sparse Bidirectional Iterators.

This allows uBLAS to choose more effecient algorithms when zero_matrix is 
assigned to a target. The target will be simply 'clear'ed and then nothing 
copied.

identiy_vector uses the same mechanism. It iterates over the single element 
which is 1.

Michael
-- 
___________________________________
Michael Stevens Systems Engineering
34128 Kassel, Germany
Phone/Fax: [phone]
Navigation Systems, Estimation  and
                 Bayesian Filtering
    http://bayesclasses.sf.net
___________________________________