[Wien] Querie on the -i switch for iterations_Spin Polarized Cases

[Gerhard Fecher]

Check which criterion was not met charge, energy, or both.
Before trying to sharpen the convergency criteria, check if the number of k-points and plane waves are sufficient
to meet that criteria. 
This is for magnetic systems more critical than for non-magnetic ones.

In systems with more than one kind of magnetic atoms, it also might appear that you have two different magnetic states close together,
sometimes it helps already if one monitors the behaviour of the individual moments to see if large fluctuations appear.
Check if you have ferrimagnetic solutions with some moments being antiparallel to the others.
I had also some rare cases were the charge extrapolation was leading to instable magnetic states and the "from scratch"
optimization with dstart (-up, -dn) improved the situation, in such cases you may also need to check and change the initial polarisation
of the atoms in the case.inst.

In many cases calculations for magnetic states are a matter of experience rather than a straight, fixed recipe.

Ciao
Gerhard

====================================
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
________________________________________
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at zeus.theochem.tuwien.ac.at]" im Auftrag von "Ghosh SUDDHASATTWA [ssghosh at igcar.gov.in]
Gesendet: Mittwoch, 27. Oktober 2010 10:01
Bis: 'A Mailing list for WIEN2k users'
Betreff: Re: [Wien] Querie on the -i switch for iterations_Spin Polarized       Cases

Thank you Gerhard,
My idea was to calculate enthalpy of formation of a typical Laves phase AB2
And for all the individual components
I used
Run_lapw -cc 0.0001 -ec 0.0001 -so -i 90 -p
I thought it might work for AB2 as well. But I did not get a smooth
variation of the volume with energy.
I also used
Run_lapw -cc 0.0001 -ec 0.00001 -in1ef -p which work beautifully for Fe_bcc

The idea therefore is two fold; one to get the desired accuracy and the
second that nobody should question the result !!

SG


-----Original Message-----
From: wien-bounces at zeus.theochem.tuwien.ac.at
[mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Gerhard Fecher
Sent: Wednesday, October 27, 2010 12:34 PM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] Querie on the -i switch for iterations_Spin Polarized
Cases

Do you realy need the charge convergency  -cc 0.0001 for the volume
optimization ?
What for ?

For a rough estimate -ec 0.0001 may be fine, you can sharpen the convergency
criteria afterwards.
Then try for example -cc 0.001 -ec 0.00001 and  increase the k-points (copy
the clm files don't use charge extrapolation in the second step)
check if the relaxed volume is different.

Sometimes a lot of 0's don't help a lot.

Ciao
Gerhard

====================================
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
________________________________________
Von: wien-bounces at zeus.theochem.tuwien.ac.at
[wien-bounces at zeus.theochem.tuwien.ac.at]" im Auftrag von "Ghosh
SUDDHASATTWA [ssghosh at igcar.gov.in]
Gesendet: Mittwoch, 27. Oktober 2010 06:45
Bis: 'A Mailing list for WIEN2k users'
Betreff: [Wien] Querie on the -i switch for iterations_Spin Polarized Cases

Dear Wien2k users,
I was trying to volume optimize (some 5 structure changes, 4 atoms per
lattice) a cubic case with the command in the optimize.job
Runsp_lapw -cc 0.0001 -ec 0.0001 -i 90 -so -p
However, in one or two cases, the SCF did not converge. What changes could
be made keeping the same accuracy of charge and energy convergence.

SUDDHASATTWA GHOSH
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien