[gpaw-users] FD vs LCAO vs PW

Jens Jørgen Mortensen jensj at fysik.dtu.dk
Tue Jul 1 09:44:48 CEST 2014


On 07/01/2014 09:03 AM, Michael Walter wrote:
> Dear Ask,
>
> would it be possible to put your nice summary to the gpaw-web pages ? 
> This could be useful for others and such thinks are not easy to find 
> in email conversations.


After seeing Roberts email yesterday, I started to write something for 
the web-page:

###################################################
The following table can guide you in choosing the right mode, but you
better run some tests yourself also.

======================  =====  ===========  =========
                         FD     LCAO         PW
======================  =====  ===========  =========
memory consumption      large  small        medium
speed for small system  slow   fast         fast
speed for large system  fast   very fast    slow
Absolute convergence    easy   complicated  very easy
======================  =====  ===========  =========

With LCAO, it can be hard to reach the complete basis-set limit and get 
absolute
convergence of energies, whereas with FD and PW mode it is quite easy to do
by decreasing the grid spacing or increasing the plane-wave cutoff energy.
##################################################

But seeing Ask's brilliant reply, I also think that it would be valuable 
to have that on the web-page.

Ask: Should I commit my piece and then you can add to it?

Jens Jørgen

>
> Best,
> Michael
>
>
> 2014-06-30 21:53 GMT+02:00 Ask Hjorth Larsen <asklarsen at gmail.com 
> <mailto:asklarsen at gmail.com>>:
>
>     Hi Robert!
>
>     Generally:
>
>     LCAO has the advantage over FD/PW that it's fast for larger
>     systems since you have fewer degrees of freedom in the
>     diagonalization.
>
>     FD/PW have the advantage over LCAO that it's easy to crank up the
>     precision by setting grid spacing or planewave cutoff.
>
>     PW has the advantage over FD that planewaves are "nice" (they tend
>     to better represent wavefunctions) and you need less to have a
>     reasonable representation.  Also the planewaves don't suffer
>     egg-box effect like grid points do.  Apparently stuff like
>     response calculations are much nicer in PW mode.
>
>     FD has the advantage over PW that you can throw many CPUs at
>     domain decomposition.  In PW, you can only parallelize over
>     k-points/spins/bands.  If you want do do many hundreds of atoms,
>     PW would therefore be less useful.
>
>     PW is extremely efficient for small-ish periodic systems.
>
>     In GPAW specifically:
>
>     PW has the slight disadvantage that it's new-ish and thus more
>     prone to things not working.
>
>     I think PW is still the only mode that implements the stress tensor.
>
>     LCAO has the slight disadvantage that it's quite different from
>     the two other modes, and e.g. some advanced functionals have not
>     been implemented with LCAO.
>
>     LCAO is very inefficient for systems where the cell is smaller
>     than the basis functions.
>
>     Did I leave anything out?  Any questions?
>
>     Best regards
>     Ask
>
>
>
>     2014-06-30 17:48 GMT+02:00 Robert Warmbier
>     <Robert.Warmbier at wits.ac.za <mailto:Robert.Warmbier at wits.ac.za>>:
>
>         Dear GPAW users and developers,
>
>         I was wondering if someone could provide a comparative pro/con
>         list for the different modes GPAW supports. There is very
>         little information on the GPAW website, as far as I could find.
>         If there is already information and I missed it, I am sorry.
>
>         Best Robert
>
>         This communication is intended for the addressee only. It is
>         confidential. If you have received this communication in
>         error, please notify us immediately and destroy the original
>         message. You may not copy or disseminate this communication
>         without the permission of the University. Only authorised
>         signatories are competent to enter into agreements on behalf
>         of the University and recipients are thus advised that the
>         content of this message may not be legally binding on the
>         University and may contain the personal views and opinions of
>         the author, which are not necessarily the views and opinions
>         of The University of the Witwatersrand, Johannesburg. All
>         agreements between the University and outsiders are subject to
>         South African Law unless the University agrees in writing to
>         the contrary.
>
>
>         _______________________________________________
>         gpaw-users mailing list
>         gpaw-users at listserv.fysik.dtu.dk
>         <mailto:gpaw-users at listserv.fysik.dtu.dk>
>         https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>     _______________________________________________
>     gpaw-users mailing list
>     gpaw-users at listserv.fysik.dtu.dk
>     <mailto:gpaw-users at listserv.fysik.dtu.dk>
>     https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>
> -- 
> ------------------------------------------
> PD Dr Michael Walter
> Addresses:
> - Fraunhofer IWM, Wöhlerstrasse 11, D-79108 Freiburg i. Br., Germany
>   Tel.: +49 761 5142 493
> - Freiburg Materials Research Center, Stefan-meier-Str. 21, D-79104 
> Freiburg, Germany
>   Tel: +49 761 203 4758
> email: Michael.Walter at fmf.uni-freiburg.de 
> <mailto:Michael.Walter at fmf.uni-freiburg.de>
> www: http://omnibus.uni-freiburg.de/~mw767 
> <http://omnibus.uni-freiburg.de/%7Emw767>
> publications: http://scholar.google.com/citations?user=vlmryKEAAAAJ&hl=en
>
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20140701/14262298/attachment-0001.html>


More information about the gpaw-users mailing list