Re: PME and fixed atoms questions

From: Peter Freddolino (
Date: Mon Mar 09 2009 - 13:51:17 CDT

Hi Chris,

Christopher Hartshorn wrote:
> Hello all. I have two questions:
> 1) In one of my sims, after about 2ns I noticed that many of the Na and
> Cl ions were way off in space relative to the rest (e.g. their are about
> 30 ions way beyond the borders of the water box, as in 50 angstroms away
> in each direction). Now, I assume this happened because I "wrapped"
> only the water, but I am not sure whether this is a problem to my PME
> calculations. I have checked and seen that the .xsc file still is only
> registering the correct xyz-lengths for the water box only. Thus, my
> question is will these ions, since they are not in the cube that the PME
> is calculated from, have a detrimental effect upon the PME
> calculations/time step because the system is no longer electrostatically
> neutral (since there are 30 less ions in the actual PBC box then after
> the system was ionized) or are these still "seen" by the PME calculations?

It is important to understand exactly what you're calculating here.
You're simulating an infinite lattice of identical periodic cells, and
the equivalent atoms in different periodic cells undergo identical
motion. Turning off the wrapping of atoms affects only their coordinates
in the output, not the calculation itself. So, while your ions may have
diffused outside of the boundaries of the original unit cell, their
periodic images underwent the same diffusion path and have ended up
inside of what was the original unit cell. The coordinates and identity
of atoms in the unit cell in this scheme is exactly equivalent to what
would happen if you had a single unit cell and wrapped the coordinates
of any atoms that crossed its boundaries. The net effect is that the
composition of your unit cell when using PBC remains constant. You can
visualize the coordinates of your ions in the unit cell either by using
pbctools to wrap your coordinates or turning on periodic images in the
Graphical Representations menu. For large enough solutes you should even
be able to locate a void in the water box corresponding to the locations
of the ion images in your unit cell.

> 2) I have a subset of my atoms that I have kept as fixedatoms.
> Originally, I kept fixedatom forces=on because I was under the
> impression that doing so would allow everything surrounding the
> fixedatom set to feel the forces of the fixedatoms (e.g. nonbonded
> forces). But after reading the NAMD manual again, I am not sure what
> the fixedatom forces actually does. Does having this feature on make
> sure that all forces (bonded/nonbonded) remain on without allowing
> movement of the atoms or does having this on only turn on the bonded
> forces because the nonbonded are always active and accounted for? In
> simpler terms, do I need to have this feature on if all I want is for
> the rest of system to feel the nonbonded forces of this subset of
> fixedatoms?

fixedatomforces enables calculation of forces *between* pairs of fixed
atoms. This will have no effect on dynamics because the atoms are, well,
fixed, but does affect the virial, and may be needed for various types
of simulation or analysis. Even with fixedatomforces off the
interactions between fixed and non-fixed atoms will be calculated.


> As always Thank you for any and all help.
> Best,
> Chris

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:27 CST