Hi think the documentation lacks of information on usage of matrices defined with std::complex scalar type.

For example, how can I extract the imaginary part and save it on a new matrix starting from

Code: Select all

Matrix< complex<float>,3,3 > A=Matrix< complex<float>,3,3 >::Random();


without explicitly looping over the matrix with for loops?

Code: Select all

Matrix< float,3,3 > I;
for (int i=0; i<9; i++)
I(i)= std::imag(A(i));


A.imag()
A.real()

if I had two sparse matrix A and B, could I obtain a third matrix C as:
complex<double> C(A,B) ?
If guess that's not working well.
Any idea?

In two steps:

MatrixXc C(m,n);
c.real() = A;
c.imag() = B;

ggael wrote:In two steps:

MatrixXc C(m,n);
c.real() = A;
c.imag() = B;

The same procedure doesn't work if C is a sparse matrix.
I have:

Code: Select all

SparseMatrix<double> A(dim,dim);
SparseMatrix<double> B(dim,dim);
SparseMatrix<complex<double> > C(dim,dim);
C.real()=A;
C.imag()=B;


Why?

Indeed with a sparse representation such operations are not possible in general, thus not allowed. A workaround:

C = A.cast<complex<double> >() + B.cast<complex<double> >();

If A and B have the same sparcity pattern, this should be pretty fast (I hope the compiler will be able to remove the additions with 0).