Hi,

I have a simple sparse matrix class build out of 3 vectors
Val - non zero values.
Col - the column of each non zero value
R_ptr - row pointer that tells you when to go to the next row

All stored as std::vector

So basically standard CRS format to store sparse matrix
I can't figure out how to map in to eigen sparse matrix without having to loop through the values (which is very slow

Any help please?

See: http://eigen.tuxfamily.org/dox-devel/gr ... tml#title3 (last line). In short:

Map<SparseMatrix<double,RowMajor> > mat(rows,cols,nnz,R_ptr,Col,Val);

Thank you ggael!

However, I am still struggling making a simple task work.
Let say I just want to enter the complex diagonal matrix (with real entries at the moment) that i have built separately in CRS format
A = [ 1 0 0 ]
[ 0 2 0 ]
[ 0 0 3 ]

The following code doesn't work for me :/
Code:

#include <vector>
#include <complex>
#include <Eigen/Sparse>

std::vector<std::complex<double> > values({ 1,2,3 });
std::vector<int> columns({ 0,1,2 });
std::vector<int> row_pointer({ 0,1,2,3 });
Eigen::Map<Eigen::SparseMatrix<std::complex<double>, Eigen::RowMajor> > mat(3, 3, 3, row_pointer, columns, values);

code end.

I have tried several formatting similar to that. can't make it to work.
Have any idea why ?

Thanks again!

You need to pass pointers, so row_pointer .data(), columns.data(), etc.

Yea, I saw that in the Eigen manual. The problem that is still doesn't work.
Tried also to use .begin() of .front
with and with out & *
Nothing works

It gives me this error:

Error (active) no instance of constructor "Eigen::Map<PlainObjectType, MapOptions, StrideType>::Map [with PlainObjectType=Eigen::SparseMatrix<std::complex<double>, 1, int>, MapOptions=0, StrideType=Eigen::Stride<0, 0>]" matches the argument list


I can't find a way to resolve it.
Any clue what might cause it?
Maybe need to include more libraries?

oh, you must use Eigen 3.3 for that feature. In 3.2, there is MappedSparseMatrix which is equivalent but deprecated in 3.3.