回复: Control the pressure and exclude the effects ofother boxs at the same time

Dear Andrew,

thanks for your help and I am so sorry that I didn't express my problem
clearly.

according to your advice, I clarify my problem:
I want to simulate the problem "Pressure-induced structural transition of
double-walled carbon nanotubes"
http://prb.aps.org/abstract/PRB/v72/i3/e035454

Unfortunately I currently have no access to this, but I confess I'm
also too busy to read it as well.

1) In my box,there’s only one MWCNT. the "bundle" I mean that, with the NPT
ensemble, the volume get smaller with the increasing pressure and finally
the origin MWCNT in other box intrude this box. From a global view, many
period boxes form the MWCNT bundles.

okay. I think I understand:
  To equilibrate the pressure, LAMMPS will try to reduce the size of
the periodic boundary box until the neighboring MWCNTs (in different
boxes) touch each other.
This sounds like what should happen.
I don't see anything wrong with this.

(...unless you want to simulate hexagonal bundles, and your simulation
box is rectangular)

The neighboring MWCNTs should exert a force on each other at the
locations where they touch.

2) Just like the article I imitate, I didn't put any medium into the box, I
use "fix/npt" to add pressure to the box for the "hydrostatic pressure"
purpose. Is it OK? (I hope so)

okay.

3)I invoke the command like this:
"fix 1 all npt temp 1 1 0.1 x 1.03 10.3 1 y 1.03 10.3 1"
I didn't use the "dilate" keyword because I thougt that, with the geting
smaller volume , the MWCNT in this box will interact with other MWCNTs in
other boxex. I think this will go against "the isolated MWCNT" which I hope.

okay

4) In fact, I know my purpose is some kind contradictory. But I just want to
increase the pressure around the isolated MWCNT and exclude other MWCNTs
near it.

    In order to exert a well-defined pressure on the MWCNT you need to
have other objects pushing on it. As Axel said, those objects could
be solvent molecules, hard walls, soft walls, or other nanotubes (from
neighboring boxes, for example).

    If you just want to apply a force pushing inwards on the atoms in
your nanotube, you might try using "fix addforce" with some suitable
formula (and perhaps replace "fix npt" with something like "fix nvt")
http://lammps.sandia.gov/doc/fix_addforce.html
However this is unlikely to give the same exact result as simulating
under constant pressure. (...at least I don't really want to try make
sense of this kind of simulation. I suspect reviewers won't either,
unless you are very careful to justify exactly what you did).

    Again, I did not read the article you mentioned, but the authors
must have clarified the method they used to keep the simulation at
constant pressure. If not, it is a legitimate question to email them
and ask them how they did it.

    Good luck.

Andrew