[lammps-users] Press/berendsen

I already tried "Fix nve" instead of "Fix npt", but it didn't work either. It seems that "press/berendsen" cannot set the pressure of the system, as the system volume increase unboundly as simulation runs. 
I think that the order of magnitude of the pressure I'm trying to set is OK because I got it from the previous minimization of the structure. 
However, "temp/berendsen" works fine.  Any other suggestion?
 
 
 
Ahmed is correct - you are using 2 thermostats
and 2 barostats.

Steve

2010/5/4 Ahmed E. Ismail <[[email protected]](mailto:[email protected])>:
>
>
> 2010/5/4 Juan J. Meléndez Martí­nez <[[email protected]](mailto:[email protected])>
>>
>> Hi all:
>>
>> I am trying to run a NPT simulation on cubic ZrO2 containing Y2O3 as
>> dopants. My input file is:
>>
>>
>>
>> units           metal
>> atom_style      charge
>> atom_modify     map array sort 0 0.0
>> neighbor        0.3 bin
>> neigh_modify    delay 0 every 20 check no
>> boundary        p p p
>>
>> read_restart    restart.8YTZP.bulk
>>
>> #Potential
>> pair_style      buck/coul/long 4.0 10.0
>> pair_modify     shift yes
>> pair_coeff      1 1 0.0 10.0 0.0
>> pair_coeff      2 2 9547.96 0.2192 32.0   # Minervini
>> pair_coeff      3 3 0.0 10.0 0.0
>> pair_coeff      1 2 1502.11 0.3477 5.1    # Idem
>> pair_coeff      1 3 0.0 10.0 0.0
>> pair_coeff      2 3 1766.40 0.33849 19.43
>> kspace_style    ewald 1.0e-4
>>
>>
>> velocity        all create 300.0 82986 dist gaussian mom no
>>
>>
>> #Output settings
>> thermo_style    custom step temp etotal press vol
>> thermo          50
>> thermo_modify   lost warn norm yes
>>
>>
>> #Molecular Dynamics settings
>> fix             3 all press/berendsen iso 31300.0 31300.0 0.5
>> fix             4 all temp/berendsen 300.0 300.0 0.5
>> fix             5 all npt temp 300.0 300.0 0.1 iso 31300.0 31300.0 0.1
>> nreset 10 drag
>> 1.0
>> dump            1 all cfg 100 dump.cfg.* id type xs ys zs q
>> dump_modify     1 element Zr O Y
>> dump            2 all custom 1 dump.prueba id type q x y z
>>
>> run             10000
>>
>>
>> Data are read from a previous minimization carried out at 0 K. When I
>> launch the job I get the following output:
>>
>>
>> LAMMPS (26 Apr 2010)
>> Reading restart file ...
>>  orthogonal box = (-10.258 -10.258 -25.645) to (10.258 10.258 25.645)
>>  1 by 1 by 2 processor grid
>>  1873 atoms
>> Ewald initialization ...
>>  G vector = 0.273155
>>  vectors: actual 1d max = 845 13 9841
>> Setting up run ...
>> Memory usage per processor = 5.25492 Mbytes
>> Step Temp TotEng Press Volume
>>     431          300    -35.73188    34867.455    21588.282
>>     450          nan          nan          nan          nan
>>     500          nan          nan          nan          nan
>>
>>
>> and so on. This does not happen when I use temp/Berendsen + npt alone, but
>> only when I use press/Berendsen.
>>
>> Any idea about what is going on here?
>
> Well, for starters, you shouldn't be combining either temp/berendsen or
> press/berendsen with fix npt, which already includes a barostat and a
> thermostat. If you're using temp/berendsen or press/berendsen, you should
> probably be using something like fix nve instead. . . .
>
> --AEI
> =========================
> Ahmed E. Ismail
> [[email protected]](mailto:[email protected])
> [[email protected]](mailto:[email protected])
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> lammps-users mailing list
> [[email protected]](mailto:[email protected])
> [https://lists.sourceforge.net/lists/listinfo/lammps-users](https://lists.sourceforge.net/lists/listinfo/lammps-users)
>
>

2010/5/5 Juan José Meléndez Martínez <[email protected]>:

I already tried "Fix nve" instead of "Fix npt", but it didn't work either.
It seems that "press/berendsen" cannot set the pressure of the system, as
the system volume increase unboundly as simulation runs.

I think that the order of magnitude of the pressure I'm trying to set is OK
because I got it from the previous minimization of the structure.

However, "temp/berendsen" works fine. Any other suggestion?

juan,

the thermo output suggests that your system is far from equilibrium,
i would first run for a while with fix nve and fix langevin (using an
adequately high friction term) that will help to bring your system
closer to equilibrium and also help dissipating the kinetic energy
evenly throughout your system. once this has stabilized, and only
then, you should switch to a variable volume fix. this has been
discussed here several times already.

cheers,
    axel.

p.s.: you choice for neigh_modify of delay 0 and every 20 is a bit odd, too.