[CP2K-user] [CP2K:14245] Re: Geometry optimisation of organic molecule on fcc Au(111) slab

[Anton S. Lytvynenko]

Dear Dmitrii,

I would recommend you to abstain from manual calculations of such kind.
Please employ any software able to generate slabs. At least at the
beginning -- you should play with it to realize how the slab generation
actually looks like. Use crystallographic data (you can find a plethora
of CIF files shipped with Avogadro or use Crystallography Open Database
to find the required CIF). Please remember that not only correct
supercell dimensions, but also correct and complete atom coordinates are
crucial for your task.

18.11.20 08:18, DMITRII Drugov пише:
> Also, do you think I need to set up different GPW cutoff and
> real_cutoff for each system individually? Or I can use one cutoff for
> these four different surfaces to calculate adsorption of organic
> molecule? 

AFAIK there are two points in that regard:

1. You should never compare electronic energies made with different cutoffs;

2. You should test the convergence of results with respect to cutoffs. I
understand that you could quickly run out of resources if you decide to
increase the cutoff to "real convergence" (where energy change on cutoff
change is really negligible), but at least you should make sure that the
cutoffs are enough to exclude drastic flaws able to provide
qualitatively wrong results.

However, it would be better if somebody more experienced comments this.

> However, I saw people specify it in this way to highlight GPW method
>     &QS
>       METHOD GPW
>       EXTRAPOLATION USE_GUESS
>     &END QS
> Do you think which one is correct for geometry optimisation?
No difference. "GPW" is the default value for METHOD in this section,
please refer to the manual:
https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/QS.html#list_METHOD
So, its explicit mention is correct but abundant.
> For calculation of Eab, I can use total energy of system for the last
> step of optimization? I don't need to take final structure and run
> energy calculation again, unless I need some charge density
> distribution etc?

If you check your output file, you should find a string "Reevaluating
energy at the minimum".

Yours,

Anton