[Wien] Convergence problem
Fecher, Gerhard
fecher at uni-mainz.de
Fri Jan 2 09:46:14 CET 2015
Maybe this paper also helps
K. Lejaeghere, V. Van Speybroeck G. Van Oost, and S. Cottenier
Error Estimates for Solid-State Density-Functional
Theory Predictions: An Overview by Means of the
Ground-State Elemental Crystals
Critical Reviews in Solid State and Materials Sciences, 39:1–24, 2014
DOI: 10.1080/10408436.2013.772503
Ciao
Gerhard
DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy:
"I think the problem, to be quite honest with you,
is that you have never actually known what the question is."
====================================
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
and
Max Planck Institute for Chemical Physics of Solids
01187 Dresden
________________________________________
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at zeus.theochem.tuwien.ac.at] im Auftrag von tran at theochem.tuwien.ac.at [tran at theochem.tuwien.ac.at]
Gesendet: Donnerstag, 1. Januar 2015 21:36
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] Convergence problem
>From the total energies it's difficult to say. It's better to
compare the cohesive energies. My cohesive energies for Fe and Ni
with the WC functional are 5.54 and 5.38 eV/atom, respectively, with
an error bar of 0.05 eV/atom maximum. In this paper you can find
PBE cohesive energies calculated with the VASP code (my PBE values
agree very well with them):
http://journals.aps.org/prb/abstract/10.1103/PhysRevB.87.214102
You have to know which error is acceptable for your purposes.
On Thu, 1 Jan 2015, Muhammad Sajjad wrote:
> Dear Tran Thank you for your further help. I am also attaching here with the structure file. Please let know the E0 values you calculated for Ni and Fe. With me
> these values are -3040.43215615 and -2544.39535597 (in Ry) respectively.
>
> On Tue, Dec 30, 2014 at 4:52 AM, <tran at theochem.tuwien.ac.at> wrote:
> Hi,
>
> The struct file that I used is attached. For my purpose, the size of
> the unit cell was large enough to avoid spurious interactions between
> neighbouring cells. Note that a, b and c are different in order to avoid
> to high symmetry.
>
> F. Tran
>
> On Mon, 29 Dec 2014, Muhammad Sajjad wrote:
>
> Dear Prof. Marks and F. TranThank you so much for your helpful suggestions. I was already doing the spin polarization calculations. I
> have got the convergence
> by using
> mixing factor 0.1, starting calculation with PRATT and then switched to MSR1 after 7 cycles, and the command runsp_lapw -cc 0.00001
> -in1ef -i 150
> The obtained magnetic moment was 2.00009.
>
> Dear Tran I would really appreciate if you share some more details about case.struct file. It will definitely be helpful for me as well
> as for others.
>
> Kind Regards
> Dr. Sajjad
>
> On Mon, Dec 29, 2014 at 3:50 AM, <tran at theochem.tuwien.ac.at> wrote:
> The calculation for an isolated atom with a code which uses periodic
> boundary conditions (like WIEN2k) is not trivial. This is what I have done
> recently for most atoms of the periodic table (excluding f-systems), and
> for some of the transition-metal atoms this was extremely difficult to
> achieve convergence.
>
> A few hints (that I used):
>
> 1) It is important (and necessary depending on the atom) to reduce the
> symmetry from cubic to, e.g., orthorhombic to be able to access the
> electronic configuration with the lowest energy (this is the procedure
> followed by several research groups like VASP for instance).
> If necessary I can give more details about the case.struct that I used.
>
> 2) lapw0 requires a lot of memory, while lapw1 requires both
> memory and time. To reduce computer time for lapw1, I was using
> iterative digonalization (this was my command for all atoms):
> runsp_lapw -ec 0.0001 -cc 0.0001 -it -i 1000 -NI
>
> 3) For the Ni atom the magnetic moment should be 2:
> :MMTOT: SPIN MAGNETIC MOMENT IN CELL = 2.00238
>
> 4) I was using the default setting for mixer.
>
> F. Tran
>
> On Sun, 28 Dec 2014, Laurence Marks wrote:
>
>
> Also, only 1 k-point (Gamma), an RMT and RKMax similar to what you use for
> bulk Ni, particularly as I assume you are doing the calculation to get an
> enthalpy of formation. You may have to use the mpi versions as it is
> probably too large for a non-mpi run.
>
> If you are doing WC+U (or -ineece) the U (or on-site hybrid) removes the
> phase transition so convergence should be simple.
>
> Straight WC is not a simple calculation because the physics for an isolated
> Ni atom is wrong. With wrong physics there is in fact no guarantee that the
> calculation will ever converge!
>
> ___________________________
> Professor Laurence Marks
> Department of Materials Science and Engineering
> Northwestern University
> www.numis.northwestern.edu
> MURI4D.numis.northwestern.edu
> Co-Editor, Acta Cryst A
> "Research is to see what everybody else has seen, and to think what nobody
> else has thought"
> Albert Szent-Gyorgi
>
> On Dec 28, 2014 11:47 AM, "Laurence Marks" <L-marks at northwestern.edu> wrote:
>
> You almost certainly need to run spin polarized, probably MSR1
> (GREED 0.1), TEMPS (room temp). The convergence is complicated
> for WC (& simple GGAs) due to an electronic phase transition
> between sp & d occupation near the fixed point.
>
> ___________________________
> Professor Laurence Marks
> Department of Materials Science and Engineering
> Northwestern University
> www.numis.northwestern.edu
> MURI4D.numis.northwestern.edu
> Co-Editor, Acta Cryst A
> "Research is to see what everybody else has seen, and to think
> what nobody else has thought"
> Albert Szent-Gyorgi
>
> On Dec 28, 2014 10:47 AM, "Muhammad Sajjad" <sajjadpu at gmail.com>
> wrote:
> Dear Users I am running SCF calculation for Ni with
> lattice constant of 30 Bohr and using WC-GGA. The
> calculation is not converging even upto 100
> iterations and more. To solve the problem I have
> performed following steps
>
> switched to TEMPS = 0.005 from TETRA
> Changed MSR1 to PRATT
> changed mixing factor (increase and decrease from 0.2)
>
> Thanking in advance.
>
> Dr. Sajjad
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at 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 at 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
>
>
>
>
More information about the Wien
mailing list