[Wien] SECLR4 error in large supercells / high number of k points
Laurence Marks
L-marks at northwestern.edu
Tue Mar 30 15:52:07 CEST 2010
This problem is harder to pin down because (in my experience) the
problem is normally not in lapw1 but somewhere else.
What "Cholesky INFO" means is that the Hamiltonian is not positive
definite, so the LAPACK routine to factorize the Hamiltonian is
failing. It gives (from the documentation) when the leading minor of
order i (here 3166) is not positive definite. I am 95% certain that
the plane-waves are first in the Hamiltonian then the linearization
terms, so in your case it looks like one the the plane-wave terms has
a problem (not linearization) but this might be misleading.
Assuming that there is nothing wrong with your LAPACK (sometimes there
is), normally I run across this type of problem when something has
gone wrong with the potential due to a bad initial density (from
dstart), a ghost-band that is not quite big enough to crash lapw2 (but
gives a warning) or electrons are being lost. Some plausible things to
check:
-- In case.scf0 -- do the fits (at the end) and the VZER terms look
reasonable compared to what you have for a smaller case?
-- If it ran for a cycle or so, look in case.scfm (also in case.scf)
and see if there are very large differences in the QMT's, indicating
that you are far from a solution.
-- Similarly look in case.outputm at the core integrals -- they should
be close to an integer. If not you may need to change the lower energy
limit because you have semi-core states which should be in the valence
band but are in the core.
-- Similarly look in case.scf(m) at the :NEC terms to see if you are
losing electrons.
-- In general, look through the log files to try and find what is wrong.
On Tue, Mar 30, 2010 at 1:42 AM, Kurt Lejaeghere
<kurt.lejaeghere at ugent.be> wrote:
> Dear WIEN mailing list members
>
> while using WIEN2k version 9.2 for a system containing Fe and Mo I obtained
> the following notification in *lapw1*.error:
>
> Cholesky INFO = 3166
> 'SECLR4' - POTRF (Scalapack/LAPACK) failed.
>
> The studied case is a 64-atom supercell of Fe/Mo containing Mo/Fe
> impurities. This error occurs as soon as the size of the supercell or the
> number of k-points becomes too large. After browsing through the mailing
> list, I found this error can occur in a number of cases, with a variety of
> possible causes. The solution is not as easy as two overlapping atoms. The
> problem resembles
>
> http://zeus.theochem.tuwien.ac.at/pipermail/wien/2006-October/007971.html
>
>
> but the solution proposed in
>
> http://zeus.theochem.tuwien.ac.at/pipermail/wien/2007-September/009724.html
>
> does not work in my case, since the problem does not occur at the LO
> coefficients, but further. For example in case.output1dn_10 (where the error
> occurs at the very end) I obtain:
>
> K= 0.89474 0.47368 -0.15789 208
> MATRIX SIZE 3705 WEIGHT=48.00 PGR:
> EIGENVALUES ARE:
> -5.2553427 -3.4536147 -3.4519203 -3.4508302 -3.4491568
> -3.4455076 -3.4450189 -3.4434554 -3.4433002 -3.4430847
> -3.4423217 -3.4406247 -3.4399921 -3.4393338 -3.4392715
> -3.4385077 -3.4381873 -3.4380193 -3.4378560 -3.4375117
> -3.4360709 -3.4359494 -3.4358333 -3.4354174 -3.4350120
> -3.4343914 -3.4328851 -3.4325970 -3.4307198 -3.4294717
> -3.4278116 -3.4276248 -2.7639594 -2.7639589 -2.7639584
> (...)
> 1.7762388 1.8232372 1.8316494 1.8410636 1.8474444
> 1.8539221 1.8655600 1.8731463 1.8814693 1.8866394
> 1.8955957 1.9015362 1.9056197 1.9286860 1.9344532
> 1.9411832 1.9447140 1.9454413 1.9509334 1.9535638
> 1.9605156 1.9634609 1.9652118 1.9798407 1.9821146
> 1.9896013 1.9966855
> 0 EIGENVALUES BELOW THE ENERGY -10.00000
> ********************************************************
>
> Matrix size 3697
> allocate HS 104.3 MB
> allocate Z 52.1 MB
> allocate spanel 3.6 MB dimensions 3697 128
> allocate hpanel 3.6 MB dimensions 3697 128
> allocate spanelus 3.6 MB dimensions 3697 128
> allocate slen 3.6 MB dimensions 3697 128
> allocate x2 3.6 MB dimensions 3697 128
> allocate legendre 46.9 MB dimensions 3697 13 128
> allocate al,bl (row) 1.2 MB dimensions 3697 11
> allocate al,bl (col) 0.0 MB dimensions 128 11
> allocate YL 7.2 MB dimensions 15 3697 8
> Time for al,bl (hamilt) : 0.3
> Time for legendre (hamilt) : 0.4
> Time for phase (hamilt) : 3.7
> Time for us (hamilt) : 1.2
> Time for overlaps (hamilt) : 1.8
> Time for distrib (hamilt) : 0.1
> Time for iouter (hamilt) : 7.6
> number of local orbitals, nlo (hamilt) 416
> allocate YL 7.2 MB dimensions 15 3697 8
> allocate phsc 0.1 MB dimensions 3697
> Time for los (hamilt) : 1.7
> Time for alm (hns) : 0.1
> Time for vector (hns) : 0.7
> Time for vector2 (hns) : 0.6
> Time for VxV (hns) : 7.5
> Wall Time for VxV (hns) : 0.2
> Cholesky INFO = 3166
>
> For case.in1 I used the following linearization energies (-in1ef used):
>
> .84535 5 0 (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
> APW/LAPW)
> 0 .84535 0.000 CONT 1
> 0 -6.50 0.001 STOP 1
> 1 .84535 0.000 CONT 1
> 1 -3.97 0.001 STOP 1
> 2 .84535 0.005 CONT 1
> .84535 5 0 (GLOBAL E-PARAMETER WITH n OTHER CHOICES, global
> APW/LAPW)
> 0 .84535 0.000 CONT 1
> 0 -4.62 0.001 STOP 1
> 1 -2.65 0.002 CONT 1
> 1 .84535 0.000 CONT 1
> 2 .84535 0.005 CONT 1
>
> Thank you
>
> With kind regards
>
> Kurt Lejaeghere
> Ghent University
>
> _______________________________________________
> 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
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
Chair, Commission on Electron Crystallography of IUCR
www.numis.northwestern.edu/
Electron crystallography is the branch of science that uses electron
scattering and imaging to study the structure of matter.
More information about the Wien
mailing list