From: Axel Kohlmeyer (
Date: Wed Jan 04 2017 - 17:49:44 CST

On Wed, Jan 4, 2017 at 5:32 PM, matt wessel <> wrote:
> Hi all,
> I am seeing odd behavior when I use the topo readlammpsdata <file> full. As

this is not odd, it is intentional. data files contain image flag
information, so it is possible to properly reconstruct unwrapped
coordinates. this is preferable to throwing this information away (as
it is done when dumping .xyz files with wrapped coordinates). if you
absolutely want the wrapped coordinates, it is easy to obtain them via
pbctools. however, the opposite is much more difficult.

in general, i consider .xyz files a file format of last resort as they
drop useful information about the system. if you would dump your
trajectory in atom style dump files or custom dumps with either x y z
and ix iy iz properties included or xu yu zu, you can get the same,
plus you have box information and timestep etc.

in short, this is all well and useful.


> I run my simulations with LAMMPS (periodic boundary conditions), I am
> dumping both an xyz and LAMMPS data file at regular intervals. I find that
> when I read a given xyz file to VMD, I get my system as I would expect it,
> all atoms confined within the defined periodic box of the system(xlo, xhi,
> etc), but of course with bonds for some bonded atom pairs straddling the
> boundary so "wrapping" to the other side (dynamic bonds of course nicely
> takes care of that visualization situation) . However, when I use topo
> readlammpsdata <file> full, I get a different looking system, where it
> appears as if topotools has done some sort of detection of bonded atoms to
> ensure that each molecule is moved so that no bonds are ever spread across
> any periodic boundaries. In fact it's as if the periodic boundaries are
> being ignored or at least used to redefine the coordinates. I looked through
> the topotools documentation at Axel's topotools web pages, and did not find
> any reference to this behavior (I may have missed it), and I also looked
> through the VMD mailing lists and saw nothing about this (I may have missed
> it here as well - but I try to make sure I do both first before posting - so
> if I did miss it, politely point it out and I'll read).
> What I can confirm is that if I extract the XYZ coords from the LAMMPS data
> file and create a new xyz file, I get an identical looking system in VMD as
> compared to the xyz file LAMMPS generates on its own, so I know the XYZ
> coordinates and atom types between xyz and LAMMPS data files taken at
> identical intervals are the same.
> Interestingly, I also tried the command using Axel's HDPE example in his
> "Odds & Ends" section of the TopoTools page, and with his HDPE example data
> file, I do not see a "vanishing" boundary box, I get a cube with bonds
> straddling the periodic boundaries, and it looks just like the image on
> Axel's page. I'm just perplexed as to what is actually happening with my
> (admittedly much larger) system.
> Also, it's not as though the behavior is "wrong", I'm not at all inclined to
> call this a bug. It produces a rather nice looking result, and I have spot
> checked that the conformations of the individual molecules are not being
> altered, just their translational position in space with respect to each
> other has been altered relative to the original xyz file. I am happy to
> provide images/xyz/data files for others to better understand/try to
> reproduce, but they are rather large files (20-40 MB) and I 'm not sure what
> the VMD-list scrubbers might do to them.
> For reference, I am running VMD 1.9.3 with Topotools v 1.7 on a Mac running
> OS 10.11.6 (El Capitan)
> Thanks for any thought/advice/help
> Matt

Dr. Axel Kohlmeyer
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.