[CPMD-list] problem with Wannier

[Axel Kohlmeyer]

On Sun, 10 Dec 2006, Yong Zhang wrote:

YZ> Hi all,

dear yong,

YZ> I am using CPMD 3.11.1, I got a problem with "WANNIER" option.
YZ> 
YZ> 
YZ> I am running CPMD on a small modecule, say N-methylacetamide, with PBE
YZ> functional and MT pp, time step=4 a.u.. Everything was fine until I added
YZ> the WANNIER option like this:
YZ> =========
YZ>   DIPOLE DYNAMICS SAMPLE WANNIER
YZ>   2
YZ> =========


yep. this is a known bug in version 3.11.1 which has been reported
before (see mailing list archives of october 2006). alessandro 
curioni already identified a workaround: either use WANNER SERIAL
or change the two calls to SET_ORBDIST in the file jrotation.F so 
that the second argument is 1 and not 0 and then recompile. a more
complete and better parallelizing (especially on BG/L) solution is
in the works.

[...]

YZ> I am running CPMD on BlueGene/L using 128 CPUs. I also tried IBM p690
YZ> cluster with 16 cpus, got the same error. Both CP-MD and BO-MD had the same
YZ> error.
YZ> But the Linux PC with one processor did not make any problem.
YZ> Is this a bug in the CPMD code or just caused by improper compilation or any
YZ> other reason?

this is a bug in the parallelization of the localization code, 
so it does not show up in the serial execution.

cheers,
   axel.

YZ> 
YZ> I appreciate your comments.
YZ> 
YZ> Best wishes,
YZ> Yong
YZ> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.