Hello, This email message is a notification to let you know that a file has been uploaded to the Files area of the ublas-dev group. File :...
ublas-dev@yahoogroups...
Jan 2, 2003 10:34 pm
368
I am in the early stages of looking at whether uBLAS might be applicable for some work I'm doing in real time filtering and estimation with Kalman filters and,...
Hi Simon, have u tried the simplest P = prod(F,prod(Q,trans(F)) ? where P,Q are symmetric_matrix<double> typed and F sparse_matrix<double> ? another idea to...
Jean-Marc.MERCIER@...
Jan 6, 2003 10:23 am
370
Hi Simon, ... / ... Q ... sparse_matrix is too obvious ;-( I'd recommend to try compressed_matrix. BTW, are you using the CVS version already? ... Assuming...
Dear Jean-Marc, ... I think I tried writing it that way, but I don't recall now. I'll definitely try it though. ... F is actually a state transition matrix (or...
Dear Joerg, Many thanks for the suggestions. ... I was running the tests about 3 weeks ago, and I was using the version in boost. Should I continue to use the...
Dear Joerg, I was implementing your suggestion and I've run into some issues with axpy_prod. Since the function doesn't seem to be in the boost version of...
Hi Simon, ... Good. ... Yep. You probably need to specify the return type of the axpy product: pPred->assign(axpy_prod<symmetric_matrix<...> >(F, temporary) +...
Hi all, I replaced `bindings_2002_12_10.zip' in `Files' sections with `bindings_2003_01_07.zip'. Only significant (observable ;o) change (wrt. previous ...
Hi Simon, ... boost::numeric::ublas::row_major, ... Done. I've attached a sample containing some scenarios (I love to pester mailing list with source code ;-)....
Dear Joerg, You're keen! I was going to email you back to say that your suggestion work, but I see you beat me to it.... I just tried compiling the code under...
Hi Simon, ... OK. I'm working on a Linux box with an 1.7 GHz Pentium IV. I compiled the sample with GCC 3.2.1, default boost.build options: <snip> g++ -c...
Hello Joerg, hello fellow uBLAS people, I hope you all had a good break. I've just got back to reading the uBLAS list after a long break. Moving to Kassel,...
This is the second of 4 postings regarding issues introduced or current with 2002-12-30. The equals function is used by uBLAS container operations to check for...
This is the first of 4 postings regarding issues introduced or current with 2002-12-30. Since the boost_1_29_0 release there have been changes to the way in ...
This is the fourth and last of 4 postings regarding issues introduced or current with 2002-12-30. While I was testing containers with integral types, I...
This is the third of 4 postings regarding issues introduced or current with 2002-12-30. Best illustrated with this simple test case. void test_index_check () {...
Hi Joerg, hi Simon, ... [...] ... [...] ... (Just a sanity check ;o) Results on 133MHz Pentium III, Linux, GCC 3.2: case 1 a 3.47 s case 2 a 4.42 s case 3 a...
Dear Joerg, Ah-ha! I forgot to add -DNDEBUG to my compilation options. I get the same results now. Perhaps there should be a UBLAS_CHECK_FOR_IDIOTS option.... ...
Hi Simon, Hi Joerg, ... I see similar results to Joerg case 1 a 2.98 s case 2 a 2.67 s case 3 a 6.39 s case 4 a 5.92 s case 5 a 2.68 s case 6 a 0.76 s case 1 b...
Joerg, This is what I mentioned before. Maybe it makes sence to have DEBUG macro (instead of NDEBUG) to compile debug version, while release wouldn't require...
Hi Michael, ... They are particularly efficient for sparse matrices as they significantly reduce the need for begin(), end() calls in that case. ... Thanks! ...
Hi Michael, ... Yes, that change was intentional. Kresimir pointed out the problem here: http://groups.yahoo.com/group/ublas-dev/message/225 (near the end of...
Hi Michael, ... Exactly. A new scalar type has to be accompanied by a corresponding type_traits specialization. ... Because it seems to work now ;-) ... Ouch...
Hi Michael, ... Yes, this seems plausible. But to use such a scalar type with all of ublas the udt needs to meet a couple of other requirements, too. ... I...
Hi Michael, ... whereas v[2] = 1 fails? I think this is an old idiosyncrasy of struct row_major and struct column_major. I'll try to fix that and see what my ...
Hi Alexei, ... [snip] NDEBUG is due to a very old discussion with Dave Abrahams about detecting, if a compiler builds something like a debug mode executable....
... Joerg, I understand this to mean that you want to model matrices over a field rather than matrices over a ring or an algebra. This will make matrix<...
Dear Joerg (and everybody else!), Many thanks for the suggestions! I implemented them and ran them in a simple benchmark. (Files attached - I'm following...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
