# [lammps-users] long range electrostatics

Hi all

I am trying to simulate fluid flow through Carbon Nanotubes using LAMMPS. the problem i am facing now is that, while using KSPACE solver there is a restriction - cannot use pppm/ewald in non periodic dimensions. but in my simulation only the flow direction can be made periodic, the other two must be non periodic. is there a method to overcome this restriction?
it would be really helpful if you could give your suggestions on this issue.
Thanking you in advance…

unni

Hi all

I am trying to simulate fluid flow through Carbon Nanotubes using
LAMMPS. the problem i am facing now is that, while using KSPACE solver
there is a restriction - cannot use pppm/ewald in non periodic
dimensions. but in my simulation only the flow direction can be made
periodic, the other two must be non periodic. is there a method to
overcome this restriction?

can you please explain,
why you need kspace at all
for that kind of system?

axel.

Dear Axel

I am trying to simulate water flow through carbon nanotubes. I beleive that inorder to study such a system we have to consider the long range electrostatics. Also in all the literatures related to this problem they have used pppm.

Unni

Dear Axel

I am trying to simulate water flow through carbon nanotubes. I beleive that
inorder to study such a system we have to consider the long range
electrostatics. Also in all the literatures related to this problem they
have used pppm.

i think in this case using long range electrostatics are overrated.
i would just use a long coulomb cutoff and one of the smooth
damping schemes for it.

your flow shoud add enough noise to your average water
density distribution structure that any residual effects from
the lack of long-range electrostatics will be negligible.

on the other hand you have to consider that with a 3d ewald/pppm,
you have unphysical periodic contributions to your forces.

what i would consider to implement is a cutoff scheme, that
considers only the one periodic direction for the cutoff.
that would reduce any cutoff artefacts significantly.

axel.

You can use periodic boundary conditions in all directions, and try
increasing the box size in the directions normal to the flow (hence
increasing the distance between the periodic images of your CNT), and
see if the properties you're trying to measure converge.

It is not possible to use long range electrostatic with only one
periodic dimension, and clearly all previous works on water flow
through CNT have in fact considered a periodic array of CNT, although
it is not always explicitely mentionned in the papers.

The standard procedure when only one dimension is not periodic is to
articificially increase the box size in that direction, typically a
factor of 3 with regards to the physical system is used (see the doc
page for kspace_modify slab option), so you can use that as a starting
point in the two directions normal to the flow...

Best regards,
Laurent

thank you Axel & Joly for your valuable suggestions…
unni