Hope this is the correct way to use this mailing list ! I am a (new) user of UBLAS library and of this mailing list. I am currently developping some basic...
Jean-Marc.MERCIER@...
Dec 4, 2002 9:38 am
305
Hi Jean, ... Yes, you're on topic. ... cholesky, ... attempt ... There's been some motion to reuse LAPACK via bindings (see ...
Dear Joerg Walter, Thanx for your quick answer. ... good new ;-) ... I've taken a look to it. ... I've already look there : i did not find anything, apart from...
Jean-Marc.MERCIER@...
Dec 4, 2002 11:17 am
307
Hi Jean, <snip> ... Good set of linear solvers (like those you've mentioned plus iterative methods, maybe superLU binding) would be very usefull for uBLAS. ...
Hi Alexei, ... My guess is that UBLAS should reach efficiency, generality, and easy writing for iterative method. I will try to implement gauss seidel and...
Jean-Marc.MERCIER@...
Dec 4, 2002 4:18 pm
309
Hi Jean, ... way to ... #include <boost/numeric/ublas/traits.hpp> template <typename T> void f() { // ... boost::numeric::ublas::type_traits<T>::sqrt...
Hi all, this is the first of a series of postings initiated through recent sparse matrix benchmarks done by Alexei Novakov (thanks, Alexei!). He (and his ...
Hi all, this is the second of a series of postings initiated through recent sparse matrix benchmarks done by Alexei Novakov (thanks, Alexei!). In the course of...
Hi Jean-Marc, you wrote ... b ... And your opinion? ... OK, I've also searched and at least found Andrew Lumsdaine's compilation of a todo list here:...
Hi Joerg ... Well, i do confess that I have no opininion, basically because I am a total nOOb to binding concept ;-) Is there some basic documentation that...
Jean-Marc.MERCIER@...
Dec 5, 2002 12:17 pm
314
Hi Jean-Marc, ... Toon wrote initial sketches: http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/boost-sandbox/boost-sandbox/libs/numeric/bindings/index.html and ...
inspired by a code from Joerg Walter (thanks to him) I succed compiling this cholesky routine with UBLAS. template<class M> bool cholesky (M &m) { bool...
Jean-Marc.MERCIER@...
Dec 5, 2002 5:49 pm
317
Hi Jean-Marc, ... total ... I think the discussion boils down to the question, whether we prefer to interface to existing linear algebra (and other numerical)...
Hi Jean-Marc, ... // This doesn't compile, because the code is so old ;-) ... Here one probably could improve using minus_assign(), because lhs and rhs aren't...
Hi Joerg, When trying to swap matrix rows I noticed in `config.hpp' (line 132): // Comeau 4.2.45 seems to have problems with BOOST_UBLAS_FRIEND_FUNCTION I...
... I've also been looking into this but don't find an easy way out. The problem is that I can't find a way to recover the underlying matrix via the...
... OK. In the mean time we've been adding traits for symmetric matrices, did some porting of the traits and added more lapack bindings. I'll integrate your...
Hi Toon, hi all, I've uploaded new version of bindings in file section (bindings_2002_12_10.zip). This file contains only traits and atlas subdirectories, as I...
Hi Kresimir, ... Another possibility is to use some type of meta programming to use the storage or matrix class to generate a functor which represents the...
Dill, John
DillJ@...
Dec 10, 2002 4:01 pm
325
Did I say Command pattern? I meant to say Strategy pattern a la Go4. WWIT? ... John...
Dill, John
DillJ@...
Dec 10, 2002 9:01 pm
326
Hi Kresimir, ... Great! [snip] ... If I understand you correctly, you'd prefer to support the BLAST Forum Standard as much as possible? That sounds reasonable...
Hi Joerg, ... Hmmm ... yes and no. Is there a cblas (Cblas? CBLAS?) standard in BLAST Forum Standard? There are two naming convensions in the same document --...
Hi Kresimir, From your description of bindings, I think that my interpretation was in the opposite direction, but may also be useful :) Attached is three...
Dill, John
DillJ@...
Dec 11, 2002 2:50 pm
330
Hi Kresimir, ... Fixed in my local copy. Thanks Joerg...
Hi Kresimir, ... I see, this is the legacy BLAS. I don't know of any distributed 'new' BLAS implementation (besides the netlib reference implementation(s)), so...
Hi John, ... It seems so... ... Doubtlessly -- I think that your approach can be another, more `user friendly' layer, built on top of the bindings. For...
Hi All, I've just uploaded my current version containing some more changes for 'special' sparse matrix products due to the continuing discussion with Alexei...