[Wien] Strange lapw1 problem

Björn Hülsen huelsen at fhi-berlin.mpg.de
Thu Jul 28 11:03:23 CEST 2005


Hi,

I want to compute the bandstructures for fcc Indium (111) slabs with 
different numbers of layers. The lattice constant is obtained from the 
bulk volume optimization but the slabs are not relaxed. I have checked 
convergence of k-points and R-MT * K-max and the vacuum thickness with 
the 3 layers slab. I have succesfully finished the SCF cycles for the 6, 
7 and 8 layers slabs. The calculations for 9 and 15 layers are running 
and looking fine. The problem is in the 12 layers slab. There the SCF 
cycle crashes in the first iteration with
FORTRAN STOP LAPW0 END
FORTRAN STOP SECLR4 - Error
in the error file of the queing system.
The only non-empty error file in the case directory is lapw1.error that 
reads:
Cholesky INFO =   1722
'SECLR4' - POTRF (Scalapack/LAPACK) failed
I have tried the following things:
1. I started the case in a new directory. That lead to the same error.
2. I changed the mixing parameter in *.inm. That lead to the same error. 
Now I know that this could not work because at this point there is no 
mixing involved.
3. I changed the height (value of c) of my supercell to avoid possible 
linear dependencies in the matrix. I am not sure whether this was very 
sensible. That lead to the same error.
So I end up with 4 starts of the SCF cycle that all crashed after around 
15 minutes with the same error message. I don't know what to do. There 
are some questions with similar problems in the mailing list but none of 
them have got an answer.
I hope you can help me.

Regards,
Bjoern

-- 
-----------------------------------------------------------------------
Björn Hülsen              Phone:  (+49 30) 8413-4863
Fritz-Haber-Institut      
Faradayweg 4-6            WWW:    www.fhi-berlin.mpg.de
D-14195 Berlin (Germany)  E-Mail: huelsen at fhi-berlin.mpg.de
----------------------------------------------------------------------- 

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: slab_12L.struct
Url: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20050728/b850c5a5/slab_12L.ksh


More information about the Wien mailing list