Re: [ublas] zero_vector broken?

From: Hongyu Miao (jackymiao_at_[hidden])
Date: 2007-12-10 10:43:03

Yes, I agree with that. There are also some similar things: column(matrix) should act "like" a vector, which means
     matrix m1;
         ...
     vector v1 = column(m1,0);
should work without forcely convert the data type.

------------------				 
Hongyu Miao
2007-12-10

-------------------------------------------------------------
From:Neal Becker
Date:2007-12-10 10:37:46
To:ublas_at_[hidden]
CC:
Subject:Re: [ublas] zero_vector broken?

Sebastian Gesemann wrote:

> On Dec 10, 2007 4:07 PM, Neal Becker <ndbecker2_at_[hidden]> wrote:
> 
>> So, we actually _want_ iterators of a zero vector to skip over the zeros?
>>  I
>> guess I really don't understand this.
>>
> 
> The thing is that a zero vector doesn't actually store the zeros. So,
> there's nothing to iterate over. Same thing with sparse vectors and
> matrices. Skipping (unstored) zeros is a good thing during calculations
> (i.e. a dot product) to speed things up, I guess.
> 
> Cheers!
> SG

I understand that it doesn't store the zeros - that's why I used it.  I
thought it was supposed to act _as if_ it had stored the zeros.

_______________________________________________
ublas mailing list
ublas_at_[hidden]
http://lists.boost.org/mailman/listinfo.cgi/ublas