Hi Hadi, That certainly is possible. Check out this example for using MUMPS with sparse diagonalization: https://gitlab.kwant-project.org/qt/cookbook/blob/master/mumps_sparse_diag.p... It should hopefully get you going, but is something is unclear—let me know. Also on a related topic Joe Weston and another Kwant developer Kelvin Loh were recently working on a standalone MUMPS Python wrapper (not yet packaged though): https://gitlab.kwant-project.org/kwant/mumpy Best, Anton On Sat, Jan 26, 2019 at 9:26 AM sazeh_9999@yahoo.com <sazeh_9999@yahoo.com> wrote:

Dear Kwant developers,

I created a general lattice with 10 million orbitals in Kwant and I obtained the Hamiltonian(H). At some point, I need to solve the inhomogeneous matrix equation H u = v, where u is the vector of unknowns and v is a constant vector. My life will be easier if I could use MUMPS solver provided by Kwant to obtain u. Is there any chance?

Best regards, Hadi