Hallo,

matrix_matrix_binary uses restict_traits to identify the type of the
iterators. Thus an expression (dense - banded) gives packed_proxy_tag as
storage category. Shouldn't it give dense_proxy because the result is
essentially dense? This way we could possibly avoid a lot of overhead in the
assignment functions ...

mfg
Gunter

I wonder what is the most efficient way to compute the product J * A *
Jt
where J is a sparse matrix, A is diagonal and Jt is the transpost of J.
In Ublas documentation I can't found any reference to a
diagonal_matrix type even
if it seems to me that such a type exists. If there is support for
this type how should it be used?
Does also anyone know what's the state of SuperLU bindings?


Thanks in advance for any help you will provide.
Regards,
Luca

In the constructor of matrix_range for instance, the precondition checks
are commented out. Now there is no error generated when you create a
matrix_range that is bigger than the matrix is defined on. Anybody an
idea why these precondition checks are commented out?

toon

The following line does not work (it is from a cholesky routine
submitted a long time ago on this list):

UBLAS::project(U.row (i))(UBLAS::range (i + 1, size)) -= UBLAS::prod
( ...

while the following works:

(U.row (i)) (UBLAS::range (i + 1, size)) -= UBLAS::prod (

In the former case I got an error about const correctness while the
latter works smoothly

Hope it is a useful warning
Bye
Antonio

ps where on earth should I post any user question now?

Hi all,

I would like to inform you that a new mailinglist has been created that
is dedicated to Generic Linear Algebra Software (GLAS).

The aim of the glas-ml is to define a generic interface for linear
algebra operations and to provide a reference implementation for this
interface in C++. This forum is intended for people with a strong
interest in numerics and/or C++ (actually the combination of both is
absolutely necessary to achieve our goals).

The list was created a few weeks ago and it has been suggested already
to join the MTL3 (a new rewrite of MTL) effort and this glas effort. So
you might want to dive into the archives already to read some more about
the initial design discussions.

The ml is hosted by the Open Systems Lab of Indiana University (thanks!)
and you can join the list at:
http://lists.boost.org/mailman/listinfo.cgi/glas.

Toon Knapen

Hi Toon,

You accidently posted to yahoo groups. It is too easy to do!

Michael

Could you provide us with a small example?

BTW: use the ublas@... ml instead of the deprecated ml at
yahoogroups.

Stefan Tarrant wrote:
> I use lu_factorize() (the pivoting version) and it still seems to be
> broken as well.
> The break seems to be in the project() function.
>
> I'm using VC++ 7.1.
>
> D:\source\boost\boost\numeric\ublas\lu.hpp(136) : error C2780:
>
'boost::numeric::ublas::matrix_indirect<M,boost::numeric::ublas::indirect_array<\
A>>
> boost::numeric::ublas::project(M &,const
> boost::numeric::ublas::indirect_array<A> &,const
> boost::numeric::ublas::indirect_array<A> &)' : expects 3 arguments - 2
> provided
>
> Great library though, and, in principle, I like the idea of getting rid
> of all of the hacks for broken compilers. There's bound to be a few bugs
> in the process.
>
> Stefan
>
>
>  -----Original Message-----
> *From:* Michael Stevens [mailto:m-stevens@...]
> *Sent:* Saturday, January 15, 2005 12:04 PM
> *To:* ublas-dev@yahoogroups.com
> *Subject:* [ublas-dev] Broken HEAD
>
>     Toon, Karl,
>
>     This is most relevant to you.
>
>     To me it look like the CVS HEAD is still broken. Karl's
>     vector_temporary
>     deduction seems only to be half implemented. Toon is it possible
>     that when
>     you committed the merged uBLAS_pure plus patches you missed
>     something out?
>
>     Michael
>     --
>     ___________________________________
>     Michael Stevens Systems Engineering
>
>     Navigation Systems, Estimation  and
>                      Bayesian Filtering
>         http://bayesclasses.sf.net
>     ___________________________________
>
>     ------------------------------------------------------------------------
>     *Yahoo! Groups Links*
>
>         * To visit your group on the web, go to:
>           http://groups.yahoo.com/group/ublas-dev/
>
>         * To unsubscribe from this group, send an email to:
>           ublas-dev-unsubscribe@yahoogroups.com
>           <mailto:ublas-dev-unsubscribe@yahoogroups.com?subject=Unsubscribe>
>
>         * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
>           Service <http://docs.yahoo.com/info/terms/>.
>
>

grinningguppi wrote:
>
> Hi,
> I just upgraded my boost install from 1.31 to 1.32 and tried to simply
> plug the bindings from the sandbox into the 1.32 tree, hoping all
> would be good. But I got tons of compiler errors within the bindings
> when compiling code that worked fine with 1.31+bindings.
> So my question is: Is there a version of the bindings which will work
> with boost 1.32? If so, where or what CVS revisions/date? And if not,
> what do I have to do to get a functional up-to-date version of uBlas +
> bindings?


You can always get the latest version of the bindings from the
boost-sandbox CVS. If you hit trouble, please provide the smallest
test-case possible that shows the problem.

BTW: please use the ublas@... ml instead of the deprecated
ml at yahoogroups.

Hi,
I just upgraded my boost install from 1.31 to 1.32 and tried to simply
plug the bindings from the sandbox into the 1.32 tree, hoping all
would be good. But I got tons of compiler errors within the bindings
when compiling code that worked fine with 1.31+bindings.
So my question is: Is there a version of the bindings which will work
with boost 1.32? If so, where or what CVS revisions/date? And if not,
what do I have to do to get a functional up-to-date version of uBlas +
bindings?

Thanks,
Hendrik

Toon, Karl,

This is most relevant to you.

To me it look like the CVS HEAD is still broken. Karl's vector_temporary
deduction seems only to be half implemented. Toon is it possible that when
you committed the merged uBLAS_pure plus patches you missed something out?

Michael
--
___________________________________
Michael Stevens Systems Engineering

Navigation Systems, Estimation  and
                 Bayesian Filtering
    http://bayesclasses.sf.net
___________________________________

Michael Stevens wrote:
> To me it look like the CVS HEAD is still broken. Karl's vector_temporary
> deduction seems only to be half implemented. Toon is it possible that when
> you committed the merged uBLAS_pure plus patches you missed something out?

Could you be more specific. I don't see what is broken actually? I ran
the regression tests succesfully and we are using it in other internal
projects?

BTW: pay attention to mail to ublas@... instead of the old
yahoogroups ml!

Toon, Karl,

This is most relevant to you.

To me it look like the CVS HEAD is still broken. Karl's vector_temporary
deduction seems only to be half implemented. Toon is it possible that when
you committed the merged uBLAS_pure plus patches you missed something out?

Michael
--
___________________________________
Michael Stevens Systems Engineering

Navigation Systems, Estimation  and
                  Bayesian Filtering
     http://bayesclasses.sf.net
___________________________________

Toon,

I just noticed the thread was/is actually being sent to the old yahoogroups
list. I guess we need to be careful to post to lists.boost.org

Michael

Michael Stevens wrote:
> I guess I need to start working through some of Karl issue and bug reports for
> uBLAS_pure!

Karl's patches are already integrated in the HEAD. This is currently
thus the different between the uBLAS_pure branch and the CVS HEAD

Toon,

On Tuesday 04 January 2005 15:56, Toon Knapen wrote:
> I just performed the merge (I polled about this a few weeks ago and
> since there were no (negative) reactions ..)
>
> The uBLAS_pure branch contained a lot of cleaning up done by Michael
> Stevens that was done before the 1.32 release. So now we'll continue
> from there.
>
I guess I need to start working through some of Karl issue and bug reports for
uBLAS_pure!

I am programming every day (work!) at the moment so this will take some time.

Michael

I just performed the merge (I polled about this a few weeks ago and
since there were no (negative) reactions ..)

The uBLAS_pure branch contained a lot of cleaning up done by Michael
Stevens that was done before the 1.32 release. So now we'll continue
from there.

toon

David Abrahams wrote:

>
> We can either go back and forth about what Lars is likely to believe, or
> you can just make the request and see what happens.  You could point him
> at the announcement if you're worried.
>


As I already posted, I have contacted Lars. And in the mean time he
acknowledge that the necessary actions were taken by him.

Angus Leeming wrote:
> David Abrahams wrote:
>>>>> And what will the gmane name be?
>>>>
>>>> You should ask Lars to just move the existing newsgroup to the new
>>>> list.
>>>
>>> That would be better coming from one of the maintainers of the new list.
>>
>> There's no need to make work for people who are already doing this as
>> volunteers.  Lars accepts requests from anyone.
>
> That wasn't my intention.
>
> What I meant was that the gmane interface exists already. If some random
> Joe pops up and says, "Please redirect the existing list to /dev/null and
> replace it with my new list", then I'd certainly treat the request with
> suspicion. A request from someone connected with the lists would carry
> more credence.

We can either go back and forth about what Lars is likely to believe, or
you can just make the request and see what happens.  You could point him
at the announcement if you're worried.

--
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com

David Abrahams wrote:
>>>> And what will the gmane name be?
>>>
>>> You should ask Lars to just move the existing newsgroup to the new
>>> list.
>>
>> That would be better coming from one of the maintainers of the new list.
>
> There's no need to make work for people who are already doing this as
> volunteers.  Lars accepts requests from anyone.

That wasn't my intention.

What I meant was that the gmane interface exists already. If some random
Joe pops up and says, "Please redirect the existing list to /dev/null and
replace it with my new list", then I'd certainly treat the request with
suspicion. A request from someone connected with the lists would carry
more credence.

Regards,
Angus

Angus Leeming wrote:
> David Abrahams wrote:
>>>> The list on yahoo should still be read but I suggest to send or cc all
>>>> replies to ublas@....
>>>
>>> And what will the gmane name be?
>>
>> You should ask Lars to just move the existing newsgroup to the new list.
>
> That would be better coming from one of the maintainers of the new list.

There's no need to make work for people who are already doing this as
volunteers.  Lars accepts requests from anyone.


--
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com

> Lars Magne Ingebrigtsen <larsi+gmane -AT- gnus.org>
>
> I'd say a simple mail would suffice saying that the mailing list has
> changed but that the gmane interface should remian as is.

Done

Toon Knapen wrote:
>> That would be better coming from one of the maintainers of the new list.

> FYI, the list administrator is Gunther Winkler with Michael Stevens and
> myself as backup. If you can let Gunther or me know who we need to
> contact exactly (and his email address) then we will take care of it.
>
> (Maybe it's best to send Lars' email address via private mail)

Hi, Toon.

His contact details are to be found here: http://gmane.org/contact.php, so
I don't think he's too worried about spambots finding it. I repeat it
below, although you should replace -AT-...

Lars Magne Ingebrigtsen <larsi+gmane -AT- gnus.org>

I'd say a simple mail would suffice saying that the mailing list has
changed but that the gmane interface should remian as is.

Regards,
Angus
k

Angus Leeming wrote:
> David Abrahams wrote:
>
>>>>The list on yahoo should still be read but I suggest to send or cc all
>>>>replies to ublas@....
>>>
>>>And what will the gmane name be?
>>
>>You should ask Lars to just move the existing newsgroup to the new list.
>
>
> That would be better coming from one of the maintainers of the new list.
>

FYI, the list administrator is Gunther Winkler with Michael Stevens and
myself as backup. If you can let Gunther or me know who we need to
contact exactly (and his email address) then we will take care of it.

(Maybe it's best to send Lars' email address via private mail)

toon

David Abrahams wrote:
>>> The list on yahoo should still be read but I suggest to send or cc all
>>> replies to ublas@....
>>
>> And what will the gmane name be?
>
> You should ask Lars to just move the existing newsgroup to the new list.

That would be better coming from one of the maintainers of the new list.

Regards,
Angus

Angus Leeming wrote:
> Gunter Winkler wrote:
>
>>
>> Hello,
>>
>> after the recent administration difficulties of this list we decided to
>> start a new one. I collected some email adresses of active posters of the
>> last few weeks and send them an invitation to the new list. Further I
>> invite all members of this list to subsribe to ublas@... at
>>
>> http://lists.boost.org/mailman/listinfo.cgi/ublas
>>
>> in order to keep in touch with ublas developement or in order to get
>> general help on the usage of ublas. If you experience any difficulties to
>> join the new list, feel free to contact me.
>>
>> The list on yahoo should still be read but I suggest to send or cc all
>> replies to ublas@....
>
> And what will the gmane name be?

You should ask Lars to just move the existing newsgroup to the new list.

--
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com

Gunter Winkler wrote:

>
> Hello,
>
> after the recent administration difficulties of this list we decided to
> start a new one. I collected some email adresses of active posters of the
> last few weeks and send them an invitation to the new list. Further I
> invite all members of this list to subsribe to ublas@... at
>
> http://lists.boost.org/mailman/listinfo.cgi/ublas
>
> in order to keep in touch with ublas developement or in order to get
> general help on the usage of ublas. If you experience any difficulties to
> join the new list, feel free to contact me.
>
> The list on yahoo should still be read but I suggest to send or cc all
> replies to ublas@....

And what will the gmane name be?

Angus

Hello,

after the recent administration difficulties of this list we decided to start
a new one. I collected some email adresses of active posters of the last few
weeks and send them an invitation to the new list. Further I invite all
members of this list to subsribe to ublas@... at

http://lists.boost.org/mailman/listinfo.cgi/ublas

in order to keep in touch with ublas developement or in order to get general
help on the usage of ublas. If you experience any difficulties to join the
new list, feel free to contact me.

The list on yahoo should still be read but I suggest to send or cc all replies
to ublas@....

regards
Gunter

Daniel L Elliott wrote:
> Hello,
>
> I am completely stumped.  Using the cvs head 1.32 version of boost with the
> following program:

This is again the old include-policy problem. Previously the headers of
ublas were not self-contained. This has been fixed but only on a branch
(and thus not in the 1.32 release). I plan to merge this branch also to
the HEAD.

For now, you just need to include storage.hpp before you include lu.hpp.
In that case, on line 24 of lu.hpp will it recognise unbounded_array
(which is defined in storage.hpp)


>
> #include<lu.hpp>
> #include<vector.hpp>
> #include<io.hpp>
>
> #include<iostream>
>
> using std::cout;
> using std::endl;
> using namespace boost::numeric::ublas;
>
> int main(){
>
>   // create a permutation matrix
>   permutation_matrix<size_t> P( size );
>   // default contructions gives the identity-permutation
>
>   cout << P << endl;
>
>   return 0;
> }
>
> I get the following errors:
> g++ -c -I.
>
-I/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/\
ublas
> -Wall -g3 -I.
>
-I/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/\
ublas
> -o permutationTest.o permutationTest.cpp
> In file included from permutationTest.cpp:1:
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:24:
> error: parse
>    error before `,' token
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:29:
> error: `
>    vector_type' is not a class or namespace
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:29:
> error: `
>    size_type' is not a class or namespace
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:29:
> error: `
>    vector_type' fails to be a typedef or built in type
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:29:
> error: ISO
>    C++ forbids declaration of `size_type' with no type
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:29:
> error: template
>    declaration of `typedef int boost::numeric::ublas::size_type'
>
/s/chopin/a/grad/dane/thesis/code/boost-1.32/include/boost-1_32/boost/numeric/ub\
las/lu.hpp:29:
> confused by earlier errors, bailing out
>
> Any ideas as to what might be wrong here?
>
> Thank you.
>
> -- dan elliott
> **************************
> Dan Elliott
> 402-210-6429 (me)
>      or
> 402-210-6735 (amy)
> dan_elliott@...
> **************************
> _______________________________________________
> Boost-users mailing list
> Boost-users@...
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>

Michael Stevens wrote:
> Toon,
>
>>Indiana university is willing to host the uBLAS mailinglist in the
>>future (just like the currently host the boost-ml too). Now we also need
>>a list-owner or administrators for the ml. Any volunteers ?
>
> Sure count me in. I think we need at least 3 people.

Gunther already offered to do it and I thus proposed myself as being the
backup to the sys admin at IU. I will contact him and ask if he can add
you too.

t

Toon,
> Indiana university is willing to host the uBLAS mailinglist in the
> future (just like the currently host the boost-ml too). Now we also need
> a list-owner or administrators for the ml. Any volunteers ?
Sure count me in. I think we need at least 3 people.

Thanks for organising this,
	 Michael