Home
Reading
Searching
Subscribe
Sponsors
Statistics
Posting
Contact
Spam
Lists
Links
About
Hosting
Filtering
Features Download
Marketing
Archives
FAQ
Blog
 
Gmane
From: Nasos Iliopoulos <nasos_i <at> hotmail.com>
Subject: Re: disabling the matrix expressions form sparse matrices
Newsgroups: gmane.comp.lib.boost.ublas
Date: Sunday 11th March 2012 15:43:28 UTC (over 5 years ago)
I think that avoiding tho derive from matrix_container is a good
compromise. Also, consider that, as long as you feel adventurous  and
depending on the interface requirements, you can locally cast from one type
to another (through dynamic or static_cast, i,e, your compressed_matrix and
ublas' compressed_matrix) to switch between interfaces, since there should
be binary compatibility between the two types. Depending on your
application though, this might be particularly insecure in many ways.

Best,
Nasos

From: [email protected]
To: [email protected]
Date: Sat, 10 Mar 2012 23:30:33 -0500
Subject: Re: [ublas] disabling the matrix expressions form sparse matrices









Hi Naso,
Thx for responding.
I am looking into the compressed_matrix. I have resorted to some 
hackery
where –by hand- defined the type to not derive from the 
matrix_container
-extending this to the constructors, of course. And it works OK – there
is 
a lot
of useful code there.
Now, given that the ublas matrix operations are not really optimized for 
performance
-although a lot of very good work is in place there - it is a bandage I am 
willing to live with.
I thought of deriving privately, but then it was too much work with the 
iterators and all.
The interesting thing is that, when I did it for matrix (actually had to 
hold a matrix and derive from
a matrix range for leading dimension requirement reasons) I had to 
explicitly derive from matrix_container
(vs2010, win7).
Hence I was hoping..
Thx again – especially on a Sat evening,
Petros


 
ps: I would like to point out that, no info 
exists in the compressed_matrix for whether this is, for example,. the
upper 
triangular part
of a symmetric/hermitian matrix. This makes it, 
unavoidably, to be used primarily for storage and element 
insertion/access,
but not ready to be used for apps. I can imagine 
that probably this can be accommodated w/adaptors, but I think this ties
ublas 
inwards
too much for broader use.
Just an opinion..
 
 

From: Nasos Iliopoulos 
Sent: Saturday, March 10, 2012 8:48 PM
To: [email protected] 
Subject: Re: [ublas] disabling the matrix expressions form sparse 
matrices
 

Petro,

If you mean using the uBlas sparse matrices as 
containers only and AFAIK this is not formally possible. I am sure though
you 
can do it through some hackery, forking or wrapping, as long as you can
satisfy 
binary compatibility between uBlas and vendor specific functions. Which
sparse 
matrix type do you need?

Best,
Nasos



> From: [email protected]
> To: 
[email protected]
> Date: Fri, 9 Mar 2012 20:35:39 -0500
> 
Subject: [ublas] disabling the matrix expressions form sparse matrices
> 

> Hi,
> is there a way to disable the matrix expressions of the 
sparse matrices ?
> I would like to use vendor provided functions for my 
calls.
> TIA,
> Petros
> 
> 
_______________________________________________
> ublas mailing 
list
> [email protected]
> 
http://lists.boost.org/mailman/listinfo.cgi/ublas
> Sent to: 
[email protected]



_______________________________________________
ublas mailing 
list
[email protected]
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent 
to: [email protected]

_______________________________________________
ublas mailing list
[email protected]
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: [email protected]
 
CD: 4ms