multi-convergent run_lapw's ;-) , or :: RE: [Wien] Problems with PORT minimization!
Jorissen Kevin
Kevin.Jorissen at ua.ac.be
Fri Oct 22 18:38:43 CEST 2004
Laurence,
in case you are interested : I have a run_lapw which can take more than one convergence criterium (I have not put the same thing in runsp yet, but this is, as you know very well, trivial).
In fact adding this is very simple, and I believe it has been discussed rather extensively on the ML some time ago (actually, I believed it to be implemented in recent versions, and was a little surprised to see that it is not).
Of course this is not really superior to your script, it would just save you a little typing :-)
Kevin Jorissen
EMAT - Electron Microscopy for Materials Science (http://webhost.ua.ac.be/emat/)
Dept. of Physics
UA - Universiteit Antwerpen
Groenenborgerlaan 171
B-2020 Antwerpen
Belgium
tel +32 3 2653249
fax + 32 3 2653257
e-mail kevin.jorissen at ua.ac.be
________________________________
Van: wien-admin at zeus.theochem.tuwien.ac.at namens L. D. Marks
Verzonden: di 19-10-2004 14:55
Aan: wien at zeus.theochem.tuwien.ac.at
Onderwerp: Re: [Wien] Problems with PORT minimization!
What seems to have happened is that the calculations have not correctly
converged, even though you are using -cc 0.0001 first before moving to
forces. I've seen this myself in some cases. The culprit appears to be
the 4th step which gave you an energy of -80.215020. Please check:
a) That the energy is converged (testconv). Unfortunately you can
get apparently converged forces without having the energy converged.
b) That :PLA is small (0.00001 or better), use grep -e :PLA on
case.scf
c) That vox (exchange-correllation potential) is small at Z=0,
using grep -e VZERO on case.scf and/or case.scf0. If not, you probably
have a false convergence. (This is for the cell you've chosen; if it was
moved to about the origin it would be at Z=0.5 that you'd check.)
d) RKMAX & GMAX are OK. I commonly use GMAX 18 since this is
CPU-cheap. Your RKMAX of 3.9 looks small to me. (N.B., RKMAX also effects
vox.)
e) Your RMT's are OK. 1 for O looks small, but might be OK. I
assume you don't have ghost-bands.
f) Do a grep -e :NEC case.scf and check that the number on the
right is 1.00000 (or very close). If not, your RMT's are not right and you
can be losing core charge which can perturb the calculations.
g) That the physics of the calculation is correct. I've noticed
that better posed calculations (i.e. spin-polarized) often converge
better.
h) That you have a reasonable number of k-points.
For reference, a script I use to ensure convergence for a spin-polarized
calculation would be:
#!/bin/csh
runsp_lapw -I -i 80 -p -cc 0.01
runsp_lapw -I -i 80 -p -cc 0.0001 -in1new 1
runsp_lapw -I -i 80 -p -ec 0.000025 -in1new 1
runsp_lapw -I -fc 0.25 -p -i 80 -in1new 1
On Tue, 19 Oct 2004, Timo Jacob wrote:
> Attached are both files.
> After 11 geometry steps the forces converged to finite values, thus I removed
> the .minrestart and .min_hess and continued the run.
>
> In the *outputM file one of the GZ values is undefined.
>
> > Please attach your case.outputM file (and for reference your case.struct
> > file).
> >
> > On Mon, 18 Oct 2004, Timo Jacob wrote:
> > > > The "***false convergence***" means that the algorithm could not find a
> > > > location with a better energy consistent with the forces that you are
> > > > giving it. This may not be an issue, since the numerical accuracy of
> > > > the energies and forces in Wien is limited. Unfortunately you don't
> > > > provide all the information that is needed to diagnose your problem:
> > > > 1) Are your energies/charges etc well converged?
> > >
> > > Yes, they are well converged. Before running mini I always perform a full
> > > SCF run with -cc 0.0001. After that I calculate the forces with the
> > > propriate corrections.
> > >
> > > > 2) Is it a CXZ spacegroup (in which case there are other problems)
> > >
> > > It is a hexagonal surface unitcell, Pt(111).
> > >
> > > > Beyond this, there is a basic point. You have fixed the positions of
> > > > your central Pt atoms, so it is quite reasonable that the forces do not
> > > > go to zero at the (constrained) energy minimum.
> > >
> > > Since there is only a single atom in each layer (1x1 surface unit cell),
> > > this should not affect. Besides an avoided translation (in z-direction)
> > > of the entire slab, the system should be able to achieve his global
> > > minimum. The only real constraint is given by the symmetry, which fixes
> > > the x and y coordinates. Thus, I only checked the z-components of the
> > > force-vectors.
> > >
> > > Thanks,
> > > Timo.
> > >
> > > > On Mon, 18 Oct 2004, Timo Jacob wrote:
> > > > > Hello,
> > > > >
> > > > > I try to otimize the structure of a 5 layer slab (1x1) including 3
> > > > > central Pt layers and an oxygen layer at each surface.
> > > > > For my calculations I use an energy cutoff of 15 Ry (for an initial
> > > > > structure optimization) and the central Pt layer is fixed during the
> > > > > geometry optimization.
> > > > > Using the PORT option with the *.inM input attached below,
> > > > > the total energy mostly decreases but the forces seem to converge to
> > > > > finite numbers
> > > > >
> > > > > :FOR002 -38.535
> > > > > :FOR003 16.344
> > > > >
> > > > > and after 3-4 geometry iterations with same forces the geometry
> > > > > optimization stops with stop_forces_small. However, *.outputM
> > > > > contains the line ****false convergence*****.
> > > > >
> > > > > Any suggestions?
> > > > >
> > > > > Thanks in advance.
> > > > >
> > > > > ------------- *.inM -------------
> > > > > PORT 2.0
> > > > > 0 0 0 1.0
> > > > > 1.0 1.0 1.0 1.0
> > > > > 1.0 1.0 1.0 1.0
> > > > > ----------------------------------
> > > > >
> > > > > ------------ *.in1 ---------------
> > > > > WFFIL (WFPRI, SUPWF)
> > > > > 3.900000 10 4 (R-MT*K-MAX; MAX L IN WF, V-NMT
> > > > > -.22515 6 0 global e-param with N other choices, napw
> > > > > 0 -0.407 0.000 CONT 1
> > > > > 1 -0.226 0.000 CONT 1
> > > > > 1 -3.803 0.000 CONT 1
> > > > > 2 -0.167 0.000 CONT 1
> > > > > 3 -0.176 0.000 CONT 1
> > > > > 3 -4.858 0.000 CONT 1
> > > > > -.22515 7 0 global e-param with N other choices, napw
> > > > > 0 -0.452 0.000 CONT 1
> > > > > 1 -0.283 0.000 CONT 1
> > > > > 1 -3.872 0.000 CONT 1
> > > > > 2 -0.240 0.000 CONT 1
> > > > > 2 -1.471 0.000 CONT 1
> > > > > 3 -0.212 0.000 CONT 1
> > > > > 3 -4.930 0.000 CONT 1
> > > > > -.22515 3 0 global e-param with N other choices, napw
> > > > > 0 -0.367 0.000 CONT 1
> > > > > 0 -1.468 0.000 CONT 1
> > > > > 1 -0.292 0.000 CONT 1
> > > > > K-VECTORS FROM UNIT:4 -9.0 2.5 emin/emax window
> > > > > ----------------------------------
> > > > >
> > > > >
> > > > > --
> > > > > Dr. Timo Jacob Phone : ++49-30-8413 4816
> > > > > Fritz-Haber-Institut Fax : ++49-30-8413 4701
> > > > > der Max-Planck-Gesellschaft
> > > > > Faradayweg 4-6 e-mail: jacob at fhi-berlin.mpg.de
> > > > > D-14 195 Berlin-Dahlem Germany
> > > > > ---------------------------------------------------------------------
> > > > >- WWW: http://www.fhi-berlin.mpg.de/th/th.html
> > > > >
> > > > > _______________________________________________
> > > > > Wien mailing list
> > > > > Wien at zeus.theochem.tuwien.ac.at
> > > > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > > >
> > > > -----------------------------------------------
> > > > Laurence Marks
> > > > Department of Materials Science and Engineering
> > > > MSE Rm 2036 Cook Hall
> > > > 2225 N Campus Drive
> > > > Northwestern University
> > > > Evanston, IL 60201, USA
> > > > Tel: (847) 491-3996 Fax: (847) 491-7820
> > > > mailto:L-marks at northwestern.edu
> > > > http://www.numis.northwestern.edu
> > > > -----------------------------------------------
> > > >
> > > > _______________________________________________
> > > > Wien mailing list
> > > > Wien at zeus.theochem.tuwien.ac.at
> > > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > >
> > > --
> > > Dr. Timo Jacob Phone : ++49-30-8413 4816
> > > Fritz-Haber-Institut Fax : ++49-30-8413 4701
> > > der Max-Planck-Gesellschaft
> > > Faradayweg 4-6 e-mail: jacob at fhi-berlin.mpg.de
> > > D-14 195 Berlin-Dahlem Germany
> > > ----------------------------------------------------------------------
> > > WWW: http://www.fhi-berlin.mpg.de/th/th.html
> > >
> > > _______________________________________________
> > > Wien mailing list
> > > Wien at zeus.theochem.tuwien.ac.at
> > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> > -----------------------------------------------
> > Laurence Marks
> > Department of Materials Science and Engineering
> > MSE Rm 2036 Cook Hall
> > 2225 N Campus Drive
> > Northwestern University
> > Evanston, IL 60201, USA
> > Tel: (847) 491-3996 Fax: (847) 491-7820
> > mailto:L-marks at northwestern.edu
> > http://www.numis.northwestern.edu
> > -----------------------------------------------
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
> --
> Dr. Timo Jacob Phone : ++49-30-8413 4816
> Fritz-Haber-Institut Fax : ++49-30-8413 4701
> der Max-Planck-Gesellschaft
> Faradayweg 4-6 e-mail: jacob at fhi-berlin.mpg.de
> D-14 195 Berlin-Dahlem Germany
> ----------------------------------------------------------------------
> WWW: http://www.fhi-berlin.mpg.de/th/th.html
>
>
-----------------------------------------------
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2225 N Campus Drive
Northwestern University
Evanston, IL 60201, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
mailto:L-marks at northwestern.edu
http://www.numis.northwestern.edu
-----------------------------------------------
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 18640 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20041022/0a176aac/attachment.bin
More information about the Wien
mailing list