[gmx-developers] MPI crash with ghost molecules

Jason de Joannis jdejoan at emory.edu
Thu Nov 2 20:25:03 CET 2006


Naveen, I suppose the extra degrees of freedom affect the dynamics in a 
small way. For example if you are using the Berendsen thermostat, the 
velocity scaling factor depends upon the most recent kinetic 
temperature which is proportional to the kinetic energy of the coupling 
group. Therefore the ghost molecules will contribute to the scaling 
factor - but anyways the temperature coupling should be preserved (and 
in my simulations it is). It would probably be better to set the masses 
to zero, but I was afraid of creating singularities in integration of 
the bonded forces. i.e. dv/dt = f/m.

/Jason

Quoting nmichaud at jhu.edu:

>
> Hello Jason,
>
> Won't you have problems with temperature and pressure coupling and such
> from the extra degrees of freedom contributed by the turned off D8PC/DPPC?
> I am trying to implement semi-grand canonical hybrid MC/MD in another MD
> code (LAMMPS) and initially I was going to add code to insert and delete
> molecules but I got stuck on how to introduce bonded terms during parallel
> runs.
>
> I like your idea of having all molecules already existing and then turning
> them on or off during the MC stage, but I'm not sure how to deal with the
> extra degrees of freedom from the off molecules. How do you deal with it?
>
> Naveen
>
> ---------------------------------------------------------------------
> Naveen Michaud-Agrawal
> Program in Molecular Biophysics
> Johns Hopkins University
> (410) 614 4435
>
>
> On Thu, 2 Nov 2006, Jason de Joannis wrote:
>
>> I start with a 128 DPPC bilayer. I also have 128 D8PC which I turn off
>> at the beginning of the simulation. When I say 'turn off' I mean that
>> the atom types are set to 13 (same as HW) and the charges are zeroed.
>> Therefore these molecules will only experience their own intramolecular
>> forces and will not exert any influence on the system. They are
>> probably absent from the neighbor lists too right?
>>
>> Now. During the simulation (in between move_x and ns) I pick a DPPC
>> molecule and turn it off while turning on a corresponding D8PC
>> molecule. The D8PC coordinates and velocities are mapped onto those of
>> the outgoing DPPC.
>>
>> This is successful in single-thread runs but fails in MPI np=2 mode.
>> For example after such a conversion on step 50, the LINCS constraints
>> diverge.
>>
>> Any thoughts?
>>
>> --
>> \ Jason de Joannis  \
>> \ Emory University  \
>>   \ jdejoan at emory.edu \
>>    \ 404-402-1332      \
>>
>>





More information about the gromacs.org_gmx-developers mailing list