There is an inconsistency between the analytic and numerical stress when performing O(N) calculations. I am pretty confident that I have tracked this down to a missing term in the S-Pulay force and stress (the effect on forces is small, around 1%, while the effect on stress is very large for small L matrix cutoffs).
There is an inconsistency between the analytic and numerical stress when performing O(N) calculations. I am pretty confident that I have tracked this down to a missing term in the S-Pulay force and stress (the effect on forces is small, around 1%, while the effect on stress is very large for small L matrix cutoffs).