Hi,

my compiler (g++ (Debian 4.3.2-1) 4.3.2) is not able to compile the following program:

Code: Select all

#include<Eigen/Sparse>

using namespace Eigen;

int main() {
SparseVector<double,RowMajor> a(10);
SparseVector<double,RowMajor> b(10);
SparseMatrix<double,RowMajor> m(10,10);
a = m * b;
return 1;
}


g++ test.cpp -I/usr/local/include/eigen2
/usr/local/include/eigen2/Eigen/src/Sparse/SparseMatrixBase.h: In member function ‘Derived& Eigen::SparseMatrixBase<Derived>::operator=(const Eigen::SparseMatrixBase<OtherDerived>&) [with OtherDerived = Eigen::SparseProduct<const Eigen::SparseMatrix<double, 1>&, const Eigen::SparseVector<double, 1>&, 3>, Derived = Eigen::SparseVector<double, 1>]’:
/usr/local/include/eigen2/Eigen/src/Sparse/SparseVector.h:242: instantiated from ‘Eigen::SparseVector<_Scalar, _Flags>& Eigen::SparseVector<_Scalar, _Flags>::operator=(const Eigen::SparseMatrixBase<OtherDerived>&) [with OtherDerived = Eigen::SparseProduct<const Eigen::SparseMatrix<double, 1>&, const Eigen::SparseVector<double, 1>&, 3>, _Scalar = double, int _Flags = 1]’
test.cpp:9: instantiated from here
/usr/local/include/eigen2/Eigen/src/Sparse/SparseMatrixBase.h:199: error: no type named ‘InnerIterator’ in ‘class Eigen::SparseProduct<const Eigen::SparseMatrix<double, 1>&, const Eigen::SparseVector<double, 1>&, 3>’


Is calculating the product of a SparseMatrix and a SparseVector not possible?


- wette

Interestingly enough, I can make a slight modification and make this compile:

Code: Select all

#include<Eigen/Sparse>

using namespace Eigen;

int main() {
SparseMatrix<double,RowMajor> a(10,1);
SparseVector<double,RowMajor> b(10);
SparseMatrix<double,RowMajor> m(10,10);
a = m * b;
return 0;
}


It looks like SparseMatrix has an operator=(SparseProduct<,>), but SparseVector does not.

I'd agree that for completeness, one would like to be able to write sparsevector = sparseproduct.

It may be the case that the product won't be sparse, though. But the programmer should know ahead of time if the product is sparse and choose the storage appropriately.

indeed, the pb was a missing overload of operator= in SparseVector, should work now in the devel branch (I did not check though....)