Hi,

This seems to be nonsense but somethimes I need the following operations:

-----------------------------------------------------------
Eigen::Matrix<double, 6, 2> B;
Eigen::Matrix<double, 2, 1> b;
B.setConstant(1);
b.setZero();

Eigen::Matrix<double, 6, 1> c = B.rightCols<0>() * b.head<0>();
std::cout << c;
-----------------------------------------------------------------

Logically thinking, multiplication of 6-by-0 matrix to 0-by-1 vector is feasible and
it gives 6-by-1 zero vector. However, Eigen doesn't allow this operation. It signals the assert
"Assertion failed: lhs.cols()>0 && "you are using a non initialized matrix", file d:\.......\eigen\Eigen\src\core\products\coeffvaseproduct.h, line 265".

Is there any way to make this operation feasible?

Thanks in advance.

This is currently not possible because for small matrices the reduction is implemented like this:

res = lhs.coeff(row, 0) * rhs.coeff(0, col);
for(Index i = 1; i < lhs.cols(); ++i)
res += lhs.coeff(row, i) * rhs.coeff(i, col);

Feel free to add a bug entry so that we don't forget about it: http://eigen.tuxfamily.org/bz/