[CP2K-user] [CP2K:13916] Poisson solver for big-slab system

hut... at chem.uzh.ch hut... at chem.uzh.ch
Fri Sep 18 08:47:31 UTC 2020


Hi

a few remarks:

for a 2d system you can use either a 3d or 2d Poisson solver.
If your system is symmetric or only has a small dipole orthogonal
to the slab you can use 3d system with a large enough vacuum space.
For largeer dipoles there is also the dipole correction term.
Remember that in CP2K the overhead of using vacuum space is rather small.

For the 2d case there are several options. ANALYTIC, MT and Wavelet.
Wavelet requires the slab to be in the xz plane.
The convergence should not depend on the choice of solver. If it is
slow, you probably have another problem. I would choose ANALYTIC, and
WAVELET as the second choice.

Implicit solvers are for very specific applications. I don't see how
they relate to your system.

regards

Juerg Hutter

--------------------------------------------------------------
Juerg Hutter                         Phone : ++41 44 635 4491
Institut für Chemie C                FAX   : ++41 44 635 6838
Universität Zürich                   E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp... at googlegroups.com wrote: -----
To: cp... at googlegroups.com
From: "Lucas Lodeiro" 
Sent by: cp... at googlegroups.com
Date: 09/18/2020 05:57AM
Subject: [CP2K:13916] Poisson solver for big-slab system

Hi all,
I want to do an AIMD over a big slab (2D) system, which has 50 A of thickness. First of all I am performing SCF calculations to get the best and faster way to compute the SCF cycles, but the convergence is not yet fast enough as I need to run AIMD.
As my system is big, I try to not use a 3D periodic setup to avoid a lot of vacuum space... I tried with ANALYTIC solver, it is stable but requires a lot of iterations.
I see the IMPLICIT one, but I do not know what are specific conditions for each solver...
In the case of MT I know I need a double system-size cell, and that is not applicable  here. The WAVELET solver needs thinner vacuum space, what is wonderful, but it does not work with FFTW (somebody knows if FFTSG are slower?), and the most important thing, are not translational invariants which is important for the MD.

I need some insights over solvers.

Regards - Lucas Lodeiro

  
  -- 
 You received this message because you are subscribed to the Google Groups "cp2k" group.
 To unsubscribe from this group and stop receiving emails from it, send an email to cp... at googlegroups.com.
 To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAOFT4PJAC%2BhdVuGWagk%2BX9LmF-yxv0TGMNn2UE_No8akornnGw%40mail.gmail.com.



More information about the CP2K-user mailing list