[QE-users] RE1: Pressure difference from the vc-relax and SCF

Lucas Nicolás Lodeiro Moraga lucas.lodeiro at ug.uchile.cl
Wed May 1 19:00:08 CEST 2019


Hi Amir.
This behaviour is normal. The final SCF is because when you start the
vc-relax, the program fix the G vectors for the first cell shape, and use
this for all vc-relax... but when the cell shape change, the G vectors must
change too. This is because the G vectors depends on the reciprocal
lattice, and reciprocal lattice depends on real space lattice --> cell
shape. Changing the vectors in each step is very expensive in time, for
that the program use the firsts, and when converge, recalculate the G
vectors for the final vc-relax cell and do a SCF calculation. As the G
vectors are not the same for the last vc-relax step and scf cycle the
results must not be the same, but must be similar. When the cell chape
change a lot the diference is bigger, but when the cell shape is little the
difference is very little and in cases produces the sames results.
Usually is necesary do a second vc-relax with the last cell and geometry,
to ensure a REAL FULL RELAXED cell.
In my case, I am very retailer, and i do in cases 3 vc-relax to ensure the
vc-relax G vectors and SCF G vectors are the same. By the way, the second
vc-relax it is a very short calculation.

Regards - Lucas Lodeiro
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190501/db1e31a8/attachment.html>


More information about the users mailing list