[CPMD-list] Bug?? Crash after 39000 steps..

[Axel Kohlmeyer]

On Mon, Mar 3, 2008 at 6:33 AM, Dave Sherman <dave.sherman at bristol.ac.uk> wrote:
>  My CPMD NOSE MD run suddenly exploded for no obvious reason:

dave,

please always provide more info about your calculation, otherwise
you force people to speculate instead of investigating your problem.

>      39061  0.02146   317.6    -701.37085    -700.89246    -700.87100
> 0.170E+02    5.01
>      39062  0.02144   317.4    -701.37079    -700.89244    -700.87100
> 0.170E+02    5.01
>      39063  0.02142   317.3    -701.37074    -700.89241    -700.87100
> 0.170E+02    5.05
>      39064  0.02140   317.2    -701.37071    -700.89239    -700.87100
> 0.169E+02    5.03
>      39065  0.02138   317.1    -701.37070    -700.89237    -700.87100
> 0.169E+02    5.04
>      39066  8.12319   322.1    -701.37071    -700.50728    -692.38410
> 0.169E+02    5.01
>      39067 49.30128   355.8    -697.35780    -686.55882    -637.25754
> 0.169E+02    5.01
>
>
>  PROGRAM STOPS IN SUBROUTINE DSPEVY| FAILED TO DIAGONALIZE [PROC=   0]
>
> The simulation is Cu + 2CL + 38 H2O's.  The funny thing is that if I repeat
> the exact same calculation but change the isotopic mass of the Cu
> from 65 to 63, the simulation works just fine and carries on for 60,000 +
> steps.   The crash isn't some computer chip/cosmic ray/power surge thing as
> it will
> occur again at timestep 39067 if I start the simulation over.  Notice that
> the cumulative displacement has decreased at timestep 39064!

the error you see is a typical sign of loss of adiabacity due to an
inadequate choice of timestep/fictitious mass. as indicated by the
displacement, you may have a fast movement in your system and
that requirese a smaller timestep/fictious mass combo. changing
the mass of the copper will give you a different trajetory and that
specific situation will not happen.

cheers,
    axel.

>
>
>
> David M. Sherman
>
> Professor of Geochemistry
>
> Department of Earth Sciences
>
> University of Bristol
>
> Bristol BS8 1RJ UNITED KINGDOM
>
> Phone: 44-(0)117-954-5446
>
> http://mineral.gly.bris.ac.uk
>
> _______________________________________________
>  CPMD-list mailing list
>  CPMD-list at cpmd.org
>  http://cpmd.org/mailman/listinfo/cpmd-list
>
>



-- 
=======================================================================
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.