boundary problems

Hi! Recent I am using lammps and meet with some problems and get confused. 3×3×30(x×y×z)nm Copper model was set up. Firstly I used p boundary for the x and y directions and s boundary for the z direction. Fix NVT and then NVE, however, oscillation would appear in the NVE even if I use drag. Then for the same model, I use p boundaries for the all three directions. Oscillation disappeared! Finally I set m boundary for z direction and found the oscillation appear again. I really do not know why and how this can happen. And my question is:
(1)Why does S boundary result in oscillation while P not?
(2)I set a very large simulation box and can see the atom did not go out through VMD. Then the m boundary should be same as P boundary condition. Why would m boundary result in oscillation?
(3)Last but not least, if it’s necessary for me to use free boundary condition in z direction, how can I solve the above problems and reach thermal equilibrium?

Thanks in advance!

2011/9/28 李冲 <[email protected]>:

Hi! Recent I am using lammps and meet with some problems and get confused.
3×3×30(x×y×z)nm Copper model was set up. Firstly I used p boundary for the x
and y directions and s boundary for the z direction. Fix NVT and then NVE,
however, oscillation would appear in the NVE even if I use drag. Then for

fix nve has no drag option!

the same model, I use p boundaries for the all three directions. Oscillation
disappeared! Finally I set m boundary for z direction and found the
oscillation appear again. I really do not know why and how this can happen.
And my question is:
(1)Why does S boundary result in oscillation while P not?

when you cut through your system by switching from a
periodic system to a non-periodic system, you induce a
shock due to the suddenly disappearing forces. try using
fix langevin. if you still have problem, please post an
example that reproduces what you see, so people can
comment on this properly. right now all advice has to be
very much a speculation. if what you describe was commonplace,
it would be documented, so you must be doing something
that not many people do.

(2)I set a very large simulation box and can see the atom did not go out
through VMD. Then the m boundary should be same as P boundary condition. Why
would m boundary result in oscillation?

no. 'm' is not the same as 'p'. it is like 's' with a minimum value.
it is still not clear what this "oscillation" represents and whether
it is a real effect or just a result of a mistake on our side.

(3)Last but not least, if it's necessary for me to use free boundary
condition in z direction, how can I solve the above problems and reach
thermal equilibrium?

hard to say without actually seeing what happens
and knowing what you are trying to do.

axel.