David Thanks a lot for your reply. I actually did catch the fact that I was accidentally storing zeros after I had sent the message, and, you're right: it definitely makes the error message understandable!
I'll try your suggestions. Best regards Tom -----Original Message----- From: David Bateman [mailto:dbate...@dbateman.org] Sent: Thursday, January 22, 2009 5:17 PM To: Thomas M. Ortiz Cc: octave-dev@lists.sourceforge.net Subject: Re: [OctDev] trouble copying sparse matrix Thomas M. Ortiz wrote: > > Hello > > I'm using Octave 3.0.1, configured for "i686-pc-msdosmsvc". I've > linked the libraries into my own program using VS 2005 SP1. > > I am trying to copy a TNT::Sparse_Matrix<double> into an Octave > SparseMatrix object using the following code: > > void TNTOctaveInterface::exportSparseMatrix(TNT::Sparse_Matrix<double> > &sparseTNTMatrix, SparseMatrix *sparseOctaveMatrix) > > { > > try > > { > > for (int i = 0; i < sparseTNTMatrix.num_rows(); i++) > > { > > for (int j = 0; j < sparseTNTMatrix.num_cols(); j++) > > { > > sparseOctaveMatrix->elem(j,i) = (sparseTNTMatrix)[i][j]; > > } > > } > > return; > > } > > catch (...) > > ... > That is a ugly way of filling a sparse matrix.. You are having to store even the zeros, which probably explains the crash. One complexity of the TNT sparse matrix conversion to Octave's format is that TNT uses compressed row format and Octave uses compressed column. Another complexity, if the header for this class I found on the web is right, is that TNT stores the matrix as a set of sparse vectors, and the matrix doesn't let you get at the vector. If it did some like void TNTOctaveInterface::exportSparseMatrix (const TNT::Sparse_Matrix<double> &sparseTNTMatrix, SparseMatrix &sparseOctaveMatrix) { octave_idx_type nc = sparseTNTMatrix.num_cols(); octave_idx_type nr = sparseTNTMatrix.num_cols(); octave_idx_type nz = sparseTNTMatrix.num_nonzeros(); // TNT uses compressed row and Octave compressed column. Store the transpose // and then transpose it yet again. sparseOctaveMatrix = SparseMatrix (nc, nr, nz); octave_idx_type jj = 0; sparseOctaveMatrix.xcidx(0) = jj; for (octave_idx_type j = 0; j <nc; j++) { for (TNT::Sparse_Vector<double>::const_iterator p = TNT::Sparse_Matrix::S_[j].begin(); p <TNT::Sparse_Matrix::S_[j].end(); p++) { sparseOctaveMatrix.xdata(jj) = p->value();; sparseOctaveMatrix.xridx(jj++) = p->index();; } sparseOctaveMatrix.xcidx(j + 1) = jj; } sparseOctaveMatrix = sparseOctaveMatrix.transpose (); } will do what you want.. Maybe it'll give you ideas for your solution.. D. -- David Bateman dbate...@dbateman.org 35 rue Gambetta +33 1 46 04 02 18 (Home) 92100 Boulogne-Billancourt FRANCE +33 6 72 01 06 33 (Mob) ------------------------------------------------------------------------------ This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
