[CPMD-list] k-points & SIGSEGV

[Juerg Hutter]

Hi Martijn

when you want to use k-points you have to use
the diagonalization schemes for optimization.
Currently, the direct optimizations (ODIIS, PCG)
do not work with k-points.
Unfortunately, the default is ODIIS and the
program does not check for the correct
combination DIAGONALIZATION/KPOINTS.

Juerg

----------------------------------------------------------
Juerg Hutter                   Phone : ++41 1 635 4491
Physical Chemistry Institute   FAX   : ++41 1 635 6838
University of Zurich           E-mail: hutter at pci.unizh.ch
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
----------------------------------------------------------


On Mon, 23 Sep 2002, Martijn Zwijnenburg wrote:

> Hi,
>
> We are having a problem with CPMD v3.5.1 (we haven't come
> round to installing the latest release yet). When we do the simple
> wavefunction optimization of a eight silicon periodic system
> (example 1 on page 49 of the manual) everything proceeds
> perfectly. However, when we include k-points by means of adding
> the following lines:
> 	POINT GROUP
> 	AUTO
> 	K-POINTS MONKHORST PACK
> 	8 8 8
> The calculation crashes with a SIGSEGV error when it does the
> ewald sumation in real space. The last line printed in the output is
> "EWALD SUMATION IN REAL SPACE OVER 1* 1* 1 CELLS". No
> other error messages apear in the output and lowering the amount
> of k-points to 1 1 1 doesn't change what happens.
> Anybody any clue what might be the problem?? CPMD runs on a
> linux beowulf (both tested on multiple (MPI/mpich) and single
> processor).
>
> cheers,
>
> Martijn
> -------------------------------------------------------------------------
> Martijn Zwijnenburg
> Lab. of Applied Organic Chemistry and Catalysis
> Delft University of Technology
> Julianalaan 136
> 2628 BL Delft
> The Netherlands
> Tel: 0031-(0)152782691
> Fax: 0031-(0)152784700
> e-mail: M.A.Zwijnenburg at tnw.tudelft.nl
> web page: http://come.to/tock
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://www.cpmd.org/mailman/listinfo/cpmd-list
>