The accuracy with which the self-consistent potential is computed can be set with the variable potential_convergence_criterion.
Example:
potential_convergence_criterion 1e-4
Requires that the largest difference between the current and the previous potential (in G-space) be 1e-4.
Default: potential_convergence_criterion 1e-6
(this default is good enough for an energy accuracy of at least 1e-6, since the error enters the potential linearly, but the energy quadratically, so the energy converges faster than the potential) Nothing smaller than 1e-6 is suggested.
If no convergence is achieved within max_iter_scfloop, the self-consistent iteration terminates, and goes on to compute the stress and forces.
Example:
max_iter_scfloop 50
Default: max_iter_scfloop 40