[lammps-users] equilibriation

Hi all,

I have a couple of questions:

  1. I’m simulating a relatively small model of 5000 beads using the FENE model. I have run my system for over 10^7 timesteps in NPT, but the volume of the box keeps fluctuating and I’m lost on what I could do to the get the system to stabilize. Am I on the right track here? This is my eqilibriation input file:

dimension 2
boundary s s p
units lj
atom_style bond
special_bonds 0 1 1
thermo 10
pair_style lj/cut 2.5
pari_coeff * * 1.0 1.0 2.5
bond_style fene
bond_coeff 1 30.0 1.5 1.0 1.0
fix 1 all npt 5.0 5.0 10.0 xyz 0.007 0.007 100.0
#monitor rdf (and msd)
fix 2 all msd 10000 msd.dat
fix 3 all rdf 10000 rdf.dat
#2d fix
fix 4 all enforce2d
timestep 0.001
run 10000000
write_restart restart.dat

  1. Does lammps do non-orthogonal boundaries? If I wanted to simulate a circular domain (with free surface along the perimeter of the domain) in lammps, is it possible?

Thank you for any help you can give me.

Abilash.

Hi all,

I have a couple of questions:

abilash,

1. I'm simulating a relatively small model of 5000 beads using the
   FENE model. I have run my system for over 10^7 timesteps in NPT,
   but the volume of the box keeps fluctuating and I'm lost on what
   I could do to the get the system to stabilize. Am I on the right
   track here? This is my eqilibriation input file:

you may want to try out the "drag" option if the npt fix.

2. Does lammps do non-orthogonal boundaries? If I wanted to simulate
   a circular domain (with free surface along the perimeter of the
   domain) in lammps, is it possible?

you could just put it in a (large enough) quadratic box
and then turn off PBC.

cheers,
    axel.

Axel is right. A simulation only periodic in z and using
fix enforce2d makes no sense. All your atoms are in the xy
plane, so there is no volume in z to adjust.

Re: a circular domain in xy. You can use any geometry
of atoms you want and put them in a box in x,y, or z
with non-periodic boundaries (free surfaces). Whether they
stay circular is up to the problem. The atoms will go
where they want to go. If you want to constrain them
to be in a circle, you'd need to write some kind of wall
fix that kept them there. See the various fix wall commands
for ideas about how to write one of your own.

Steve

Yes, that was my mistake. I forgot about my 2d fix. I had switched from
an infinitely thick to infinitely thin (in z) model in z, I did'nt make
the change in the periodicity for z.

Thank you for your suggestions.

Abilash.

Quoting Steve Plimpton <[email protected]>:

Axel is right. A simulation only periodic in z and using
fix enforce2d makes no sense. All your atoms are in the xy
plane, so there is no volume in z to adjust.

Re: a circular domain in xy. You can use any geometry
of atoms you want and put them in a box in x,y, or z
with non-periodic boundaries (free surfaces). Whether they
stay circular is up to the problem. The atoms will go
where they want to go. If you want to constrain them
to be in a circle, you'd need to write some kind of wall
fix that kept them there. See the various fix wall commands
for ideas about how to write one of your own.

Steve

> Hi all,
>
> I have a couple of questions:
>
> 1. I'm simulating a relatively small model of 5000 beads using the
FENE
> model. I have run my system for over 10^7 timesteps in NPT, but the
volume
> of the box keeps fluctuating and I'm lost on what I could do to the
get the
> system to stabilize. Am I on the right track here? This is my
eqilibriation
> input file:
>
> dimension 2
> boundary s s p
> units lj
> atom_style bond
> special_bonds 0 1 1
> thermo 10
> pair_style lj/cut 2.5
> pari_coeff * * 1.0 1.0 2.5
> bond_style fene
> bond_coeff 1 30.0 1.5 1.0 1.0
> fix 1 all npt 5.0 5.0 10.0 xyz 0.007 0.007 100.0
> #monitor rdf (and msd)
> fix 2 all msd 10000 msd.dat
> fix 3 all rdf 10000 rdf.dat
> #2d fix
> fix 4 all enforce2d
> timestep 0.001
> run 10000000
> write_restart restart.dat
>
> 2. Does lammps do non-orthogonal boundaries? If I wanted to
simulate a
> circular domain (with free surface along the perimeter of the
domain) in
> lammps, is it possible?
>
> Thank you for any help you can give me.
>
> Abilash.
>
-------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
> Don't miss this year's exciting event. There's still time to save
$100.
> Use priority code J8TL2D2.
>

http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone