Boost logo

Ublas :

From: Le Farfadet Spatial (le.farfadet.spatial_at_[hidden])
Date: 2004-12-24 05:12:07

        Hello everybody out there !

        Let me introduce myself : I am Yoann LE BARS and I am a fresh Boost
uBlas user. I am french too, so please excuse my bad english.

        I have got a problem using uBlas. I have post it on Boost-user list and
Toon Kapen invit me to post here. I apologies if you read it for a
second time.

        I have got a problem with Boost uBblas and I have made a little exemple
which show what is my problem (it compiles but I found the behaviour
very strange) :

        If you compile it, you will see that passing M, a 70 * 70 matrix, its
become A, a 50 * 70 matrix. I loose 20 lines and I do not know why. What
is strange is that M and A have got the same memory emplacement.

        Here is a version which works but do not satisfy me :

        The matrix remain a 70 * 70 matrix and when I used it (in the real
code), the values stocked on A are the good ones.

        Why does it not satisfy me ? Well, because I am fulling a sparse matrix
with an identity matrix, the A matrix and the transposed A matrix : it
is very redondant and A can be a very big matrix.

        That is the reason why I have make the LeastSquareMatrix class. I
believe I have done something wrong in it, but what ? What would be
perfect is to have a reference to the matrix on LeastSquareMatrix class
but I prefer not to use a pointer.

        Best regards.

                                        Yoann LE BARS,
                                        alias Le Farfadet Spatial