"wrap arround errors" and relaxation?

Message

lahaye · #1 Post by **lahaye** » Tue Sep 19, 2006 2:04 am

Hi all,

How bad are wrap around errors when doing a relaxation?
Do wrap around errors significantly affect the calculated forces?
Or do they have marginal effects on the final structure?

Thanks,
Rob.

#2 Post by **admin** » Tue Sep 19, 2006 7:55 am

please have a look at the vasp-workshop tutorial handout accuracy.pdf which you can download from http://cms.mpi.univie.ac.at/vasp-workshop
for a detailled discussion of that question

lahaye · #3 Post by **lahaye** » Tue Sep 19, 2006 11:08 am

Dear admin,

The vasp-workshop tutorial mentions on page 26:

for Accurate wrap around errors are avoided...

but when I use PREC=Accurate, then I still get:

WARNING: wrap around errors must be expected

Well, something is confusing me here.

The VASP Guide says in the chapter on "Energy cut-off ENCUT
and FFT-mesh" that the drift forces should be smaller than the
size of the forces interested in.
I'm relaxing a system till the forces are smaller than 0.01 eV/A
(EDIFFG = -0.01).
Should I therefore change ENCUT (or: NGX, NGY and NGZ ?)
so that the drift force is much less than 0.01?

Thanks,
Rob.

#4 Post by **admin** » Tue Sep 26, 2006 1:29 pm

rather increase ENCUT, it may be risky to fix NGX NGY and NGZ if your cell changes in shape and volume upon relaxation. If they are fixed, you can as well fix NGX NGY and NGZ

#5 Post by **admin** » Tue Jan 09, 2007 4:22 pm

download it from the VASP-workshop files:
http://cms.mpi.univie.ac.at/vasp-workshop
(tutorials and session talks)