From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Mon Aug 31 2009 - 13:01:26 CDT
On Mon, 2009-08-31 at 11:56 -0500, jouko_at_uchicago.edu wrote:
jouko,
thanks for posting an interesting question, and for spending
some effort on researching your options before asking. we have
too few people doing this, so let me commend you for this and
suggest to others to follow your example.
> I am planing on running a md simulation of a membrane. I want
> different ion concentrations on different sides of the
> membrane. My concern is that if I apply periodic boundary
> conditions ions will be able to diffuse from the top of
> simulation box to the bottom of the simulation box and I will
> not have a difference in the concentration of the ions. I
right. this is a valid concern.
> thought that maybe I should create a wall with Tcl boundary
> conditions, or use two membranes, instead of one. I searched
using two double layers or membranes would be the clean solution,
but also a computationally more demanding. if you can afford to
run a reasonable size system at a proper simulation speed, this
would be the best way to proceed. you can average over the two
membranes in your analysis (after considering the symmetry) and
don't lose too much time, unless the k-space part of the PME
will be too time consuming.
using a regular wall is not such a good idea, as it will
create an additional potential drop at the surface that may
interfere with you calculation, so you'd have to use a very
large water slab. what might be worth considering, if you
are constrained in computational resources would be a
_reflecting_ wall (i.e. velocities get reversed if z is
larger than a cutoff value) that would only act on the ions.
but i would still be concerned of generating an artificial
potential drop across that wall, also because the waters
on the other side of the wall would still "see" the ions
and thus be affected by it. so if you'd go for this option,
it would be highly recommended to make some tests with
a (small) bulk water/electrolyte system comparing to pure
water/electrolyte and see how far the boundary effects
would reach.
> around and could not find anything that said periodic boundary
> conditions should be different for membrane simulations. I
> asked someone in my group and he said that I did not have to
> worry about this, and all I needed to do was turn wrap off. I
this is bad advice. that wrapping keyword only affects the
_output_ of coordinates, for the calculation of interactions,
coordinates are _always_ wrapped to comply to the minimum
image conventions.
> disagreed with this. Eventually we decided to ask the namd
> mailing list. I would appreciate any advice on how to set up
> the simulation. I would also like someone to confirm or deny
> whether molecules can move from the top of the simulation box
> to the bottom, if periodic boundary conditions are applied
> regardless of whether wrap is on or off.
see above. to reiterate the coordinate wrap option does
not affect the physics of your system at all.
cheers,
axel.
>
>
> Thanks,
>
> Jouko
>
-- Dr. Axel Kohlmeyer akohlmey_at_gmail.com Research Associate Professor Institute for Computational Molecular Science College of Science and Technology Temple University, Philadelphia PA, USA.
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:53:14 CST