Hi,
In the past and very recently, we investigated the discontinuity
in the laplacian and ELF function in particular. It is difficult
to give a general conclusion since we considered only two
cases (Li and diamond), but what we observed is that
reasons to have a big discontinuity are:
1) For large spheres (e.g., larger than 2.5 Bohr for Li), the
energy parameters in case.in1 can be crucial. In this
respect, the way the the energy parameters are currently chosen in
WIEN2k seems to be much better than 10 years ago (at least for Li).
2) For small spheres (e.g., 1.2 Bohr for Li or diamond), the numerical
derivatives of the plane-wave expansion of rho in the interstitial
region is very inaccurate just at the border with the sphere (we know
why). An increase of the FFT factor in case.in0 or the use of .lcore
do not help at all (at least not for diamond, I did not try for Li).
So, the discontinuity in the laplacian/ELF in diamond/Li for small
spheres is not yet solved. We have some ideas but this would require
some time to implement/test them.
Concerning the use of .lcore, we have also observed that it creates
some wiggles in the interstitial region.
FT
On Wednesday 2016-11-23 08:45, Georg Eickerling wrote:
Date: Wed, 23 Nov 2016 08:45:30
From: Georg Eickerling <georg.eickerl...@physik.uni-augsburg.de>
Reply-To: A Mailing list for WIEN2k users <wien@zeus.theochem.tuwien.ac.at>
To: A Mailing list for WIEN2k users <wien@zeus.theochem.tuwien.ac.at>
Subject: Re: [Wien] Poisson and clmsum
Dear Wien users,
now I have to join this thread, because this last piece of information sounds
interesting also to me.
I am doing topological analyses of electron densities/Laplacians via WIEN2k and
the
discontinuities at the MT radii spoil basically any nabla² rho(r) map
one tries to make with WIEN2k. I have tried many things (large RKMAX, lmax, APW
vs. LAPW)
but tiny steps at the MT boundary remain in rho(r) and therefore in all its
derivatives.
The information about generating a ".lcore" file is new to me - what does this
file actually do
if it exists and when should it be generated, already for the init or the scf
step?
best regards
Georg Eickerling
On 11/22/2016 07:51 PM, Laurence Marks wrote:
N.B., there can also be a discontinuity in the charge (small) due to the
tails of the core states which can be eliminated by doing "touch .lcore".
On Mon, Nov 21, 2016 at 8:36 AM, Laurence Marks <l-ma...@northwestern.edu>
wrote:
APW+lo methods have a step in the gradient of the density at the RMT. To
avoid this use a lapw basis set: to reduce it increase RKMAX.
---
Professor Laurence Marks
"Research is to see what everybody else has seen, and to think what nobody
else has thought", Albert Szent-Gyorgi
http://www.numis.northwestern.edu
Corrosion in 4D http://MURI4D.numis.northwestern.edu
Partner of the CFW 100% gender equity project, www.cfw.org/100-percent
Co-Editor, Acta Cryst A
On Nov 21, 2016 07:39, "John Rundgren" <j...@kth.se> wrote:
Dear Peter Blaha and Gavin Abo,
Non-overlapping muffin-tin spheres are used by WIEN2k and my LEED program
eeasisss (Elastic Electron-Atom Scattering in Solids and Surface Slabs).
But RMT(setrmt_lapw) is not automatically the best choice for
RMT(eeasisss). LEED touching radii of atoms depend on exchange-correlation
interaction between crystal electron gas (the WIEN2k electrons) and an
incident LEED electron (energy 20-500 eV).
This is a N+1 electron scattering situation, where "N" signifies the
WIEN2k electrons and "1" an alien LEED electron.
W2k can be reconciled with LEED using an atomic sphere approximation
(ASA) extending into the Fourier expansion realm of W2k. A while ago you
(P.B. and G.A.) suggested an ASA routine, in which I now use Poisson
differentiation of vcoul_ASA in order to obtain clmsum_ASA. I consider the
case LM=(0,0), sufficient for current LEED.
The considered structure is a supercell = a surface slab 15 layers thick,
where layers 1-2 and 14-15 are C-O and O-C, respectively. Mirror symmetry
about layer 8. At the C-O layers vcoul_ASA(0,0) is continuous across the
RMT radius, but clmsum_ASA(0,0) versus radius shows a step of the order of
10%.
Is the step k-point dependent? It does not seem so. With 16 and 48
k-points the clmsum_ASA(0,0) steps are preserved within 6 digits.
I shall be glad to supply the code. When the described numerical error is
fixed, WIEN2k and eeasisss can be re-run self-consistently within the model
of non-overlapping muffin-tin atoms.
Regards,
John Rundgren
KTH
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
--
============================
PD Dr. Georg Eickerling
Universität Augsburg
Institut für Physik
Lehrstuhl für Chemische Physik und Materialwissenschaften
Universitätsstr. 1
86159 Augsburg
E-Mail: georg.eickerl...@physik.uni-augsburg.de
Phone: +49-821-598-3362
FAX: +49-821-598-3227
WWW: http://www.physik.uni-augsburg.de/cpm/
=====================================================
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
