search

OrderN / CONQUEST-release

Full public release of large scale and linear scaling DFT code CONQUEST
http://www.order-n.org/
MIT License
96 stars 25 forks source link

Hartree stress wrong for non-SCF calculations #214

Closed davidbowler closed 9 months ago

davidbowler commented 10 months ago

The Hartree stress found with non-SCF calculations is wrong, I think (it is often several hundred Ha while for the equivalent SCF calculation it is around 1Ha). More investigation required.

davidbowler commented 10 months ago

The problem is simply that this routine was not written to account for the neutral atom potential! Background physics will be needed followed by changes to evaluate Hartree stress on appropriate charge density.

lionelalexandre commented 10 months ago

Does this can also explain the problem with geometry optimisation with nonSCF?

davidbowler commented 10 months ago

Quite possibly: I have found a simple error for the nonSCF geometry optimisation (more on this soon) but possibly there is a problem with non-SCF forces as well…

davidbowler commented 10 months ago

Further information on the forces (which I will also put in a separate issue, but forces and stress are intimately linked) is that the presence of PCC seems to be the problem for non-SCF forces (atoms without PCC have perfect forces). I'm not sure why yet!

davidbowler commented 10 months ago

Part of the problem here has been fixed by #218 (the force error was also present in the stress) so non-neutral-atom calculations are now correct. But the neutral atom still needs a reformulation to find the stress correctly.

davidbowler commented 9 months ago

This is now fixed by #229