Wavelet solver for slab

59 views
Skip to first unread message

MD Simulation

unread,
Dec 29, 2019, 10:30:53 AM12/29/19
to cp2k
Hello,

I have a slab periodic in the XY direction. Can the wavelet solver be used for this situation? The documentation says XZ for the wavelet slab.

Thanks
Stacey

Thomas Kühne

unread,
Dec 29, 2019, 10:43:47 AM12/29/19
to 'Dorothea Golze' via cp2k
Dear Stacey, 

other POISSON solvers within CP2K are, but using Wavelet XZ is indeed the 
only way to make 2D periodic calculations. Please do not ask why, but you 
have to rotate the system by hand to make it XZ periodic. 

Cheers, 
Thomas

--
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 cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/79dcf758-eafd-4c40-9e55-a583e8245471%40googlegroups.com.

==============================
Thomas D. Kühne
Dynamics of Condensed Matter
Chair of Theoretical Chemistry
University of Paderborn
Warburger Str. 100
D-33098 Paderborn
Germany

MD Simulation

unread,
Dec 29, 2019, 11:08:54 AM12/29/19
to cp2k
What other solvers can be used for non-periodic?

Thanks again!

Matt W

unread,
Dec 29, 2019, 11:53:16 AM12/29/19
to cp2k
Come on! You've only got to swap two columns of numbers!

I think you can use MT for this as well or there is a dipole correction that you can apply with a fully periodic set up.

Matt

MD Simulation

unread,
Dec 29, 2019, 12:36:31 PM12/29/19
to cp2k
What's the difference in the analytic solver and the wavelet solver?  The analytic does not need to be specifically XZ periodic.  Is there any major differences? 

Thomas Kühne

unread,
Dec 29, 2019, 3:13:04 PM12/29/19
to 'Dorothea Golze' via cp2k
Dear Stacey, 

the simplest would be to follow Matt’s advice and „just do it“. 
There is nothing wrong using the analytical Poisson solver, 
which to the best of my knowledge refers to Hockney’s method. 
It is simply not the most economic choice since it requires 
double the amount of memory and two additional FFTs on 
a cell of double the size. Also the molecule has to be at the 
center of the cell and as usual 3-5A has to be added such that 
the charge density vanishes at the cell boundaries. 

Cheers, 
Thomas

P.S. If you want to play it easy you can also simply do a supercell 
calculation ...

--
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 cp2k+uns...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages