Hello.

I know, there are already two posts concerning this topic, but unfortunately they doesn't help on.
ALthough it is not recommended, I want to inverse a sparse matrix. In particular, I want to calculate
the covariance matrix from a given sparse normal equations matrix. For solving my least squares system
I use a cholesky decomposition and solve it for the unknowns. This works quite well.

Eigen::SparseMatrix A(n,n); //n=17495
Eigen::VectorXd b(n);
Ax = b;
Eigen::SimplicialCholesky<Eigen::SparseMatrix<double> > chol(A);
x = chol.solve(b);

But how to compute A^(-1)

Thank you in advance!
Marco

call solve with a sparse matrix representing the identity matrix.

Thank you! My approach was to use the Lower triangular matrix from the Cholesky LLT. But you are completely right, using the Identity matrix works fine!

Again, thanks!