Bug: Stress result is wrong in NSCF calculations

[WHUweiqingzhou]

Describe the bug

I find the NSCF calculation with a well-converged charge density gives a wrong stress.
b.t.w, I use LTS version.

[mohanchen]

Is it PW or LCAO calculations?

[WHUweiqingzhou]

Is it PW or LCAO calculations?

LCAO. I'm not sure if this bug also occurs in PW.

[dzzz2001]

In NSCF calculations, SCF iterations are not performed to solve for the charge density. However, the density matrix input for the grid integration part of the original force calculation depends on the dmr matrix input during the rho calculation. This results in the dmr matrix actually input to the grid integration part of the original force(and stress) calculation being zero, which may have caused computational errors here. This bug has been fixed in new gint module(PR #6296).

[mohanchen]

In NSCF calculations, SCF iterations are not performed to solve for the charge density. However, the density matrix input for the grid integration part of the original force calculation depends on the dmr matrix input during the rho calculation. This results in the dmr matrix actually input to the grid integration part of the original force(and stress) calculation being zero, which may have caused computational errors here. This bug has been fixed in new gint module(PR #6296).

Excellent job!