VMD-L Mailing List

From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Fri Oct 29 2010 - 11:37:11 CDT

hi aric,

On Fri, Oct 29, 2010 at 10:16 AM, Aric Newton <agnewton_at_berkeley.edu> wrote:
> Hi Axel,
> Thank you for looking into the issues I was having with my input files.  I
> have replaced the mdfplugin and lammpsplugins.  Upon VMD start up, there is
> a message about these 2 new plugins (see below), but otherwise all runs
> smoothly.
> ...
> Found 2 plugins or data handlers in directory
> '/Users/agnewton/Applications/GeochemistrySoftware/VMD/VMD
> 1.8.7.app/Contents/vmd/plugins/MACOSXX86/molfile'.

excellent. the changes have been integrated into the VMD development
CVS tree and will be in the next alpha snapshots and - of course - the
next general release.

> I have followed your recommendation and used the *.psf file that I generated
> from the lammps data file and TopoTools and a *.dcd formatted trajectory for
>  visualization of my trajectories (I wrote redundant simulation trajectories
> (a binary and ASCII text) as insurance).  Your attachments at the bottom
> anticipate the other unspoken issue.  Namely, that my bonds are wrapped
> around the periodic boundaries when animating the trajectory.  I've been

for the scenario of dealing with the very long bonds, i have another
suggestion (outside of not using the regular bonds for visualization):
you can generate .psf file, where the initial set of those long bonds
is simply broken. the topotools plugin has a convenience function
that allows you to break bonds, but you can also do that interactively
using the mouse. since this psf file would only be used for visualization,
it doesn't matter that those bonds are missing, or whether they are
inconsistent with the defined angles, dihedrals and so on.

> able to unwrap the trajectories in the past with the PBC tools (pbc unwrap),
> but as you stated, it isn't very useful when trying to produce an animation
> of the trajectory because the water and ions just wanders out of the cell.

you can fix that by unwrapping first (or storing in unwrapped form from within
LAMMPS) and then wrap with -compound residue.

>  I've been reading the documentation for the PBC tools and thought that it
> would be possible (after unwrapping the trajectory data) to selectively wrap
> the Na ions and the water molecules in the image while avoiding bonds across
> the cell with something like  pbc wrap -sel (Na & H2O) -compound (H2O).  I
> thought that the -compound option would allow me to specify which atoms will
> not be wrapped if a compound (the water) would be split by the

the -compound flag can take one of three arguments: residue segid chain
(or abbreviations as indicated). that would mean that either residues stay
intact when wrapping, or segments or chains. in most cases -compound res
is the best choice.

> wrapping.  Your advice suggests that I have misunderstood the description (I
> haven't tried it)  and that I should try another avenue (maybe a LAMMPS
> script to "unwrap" the trajectory?  It's an option when writing the LAMMPS
> trajectory, maybe post-simulation, too?).

if you want to wrap by water molecule, it is probably most convenient
to already write an unwrapped trajectory in lammps. that will speed up
the process.

cheers,
    axel.

> Thank you again for looking into this.  I provided the above description as
> background.  I'll work on the issue myself and see if I come up with a
> solution before posting anything to the list.
> Thank you,
> Aric
>
> On Oct 27, 2010, at 11:43 AM, Axel Kohlmeyer wrote:
>
> On Sun, 2010-10-24 at 08:45 +0900, Aric Newton wrote:
>
> hi aric,
>
> a few updates.
>
> Approach 1
>
> Referring to archived posts (Wednesday, August 11 2010; Volume 01 :
>
> Number 1657), I attempted to load first the *.mdf (structure) and then
>
> *.car (coordinate) files from Materials Studio.  When loading the
>
> *.mdf file, I get an error due to the "namelist index value being too
>
> large" (message reproduced below) and the process is aborted.  Is
>
> there away to set the "larger index types" when compiling VMD?  The
>
> namelist index value at the time of overflow is 1851.  I am using the
>
> pre-packaged VMD for MACOSXX86, version 1.8.7 (August 1, 2009) and
>
> didn't recognize that option during installation.  My system isn't
>
> that large (6408 atoms), but I tried a smaller version of the model
>
> the first bug in the mdf plugin would be triggered when your
> system has more than 9 "molecules" defined in the .mdf file.
>
> without the hydration shell (no. atoms 1470).  Reading the *.mdf
>
> (structure) file, the mdfplugin "Could not find atom in hash table"
>
> and vmd reported an "Error reading bond information".  This wasn't
>
> completely fatal, as I could load the *.car (coordinate) file into the
>
> this was due to the fact, that you had sodium atoms without
> any bonds as part of the first "molecule". the parser in the
> mdfplugin would flag this as a parser error and do some other
> crazy stuff in other places. i've corrected the plugin to handle
> this situation more generously and consistently.
>
> molecule (car plugin reported no errors), but there were no bonds.  Is
>
> this an issue that I can solve by increasing the atom index value
>
> somehow?  Or is it an issue with the plugins/ files?  The import files
>
> were created with Materials Studio 5.0 (and not Insight II) which
>
> still uses an *.mdf file for structure but an alternative format
>
> looking at the change log for the mdfplugin, it looks as if its
> functionality has not been significantly changed since may 2004.
>
> there is still one problem interfering with a nice visualization
> and that is impossible to solve with any method: your system has
> bonds across periodic images. While that is treated correctly in
> LAMMPS, VMD will always display them as bonds within the same
> periodic image. thus it may be better to use some other representation
> for that part of the model.
>
> check out the attached pair of "good" and "bad" examples.
>
> [...]
>
> cheers,
>    axel.
>
>
> Thank you,
>
>
> Aric
>
> --
> Dr. Axel Kohlmeyer  akohlmey_at_gmail.com
> http://sites.google.com/site/akohlmey/
>
> Institute for Computational Molecular Science
> Temple University, Philadelphia PA, USA.
> <bad-viz.gif><nice-viz.gif>
> 

-- 
Dr. Axel Kohlmeyer
akohlmey_at_gmail.com  http://goo.gl/1wk0
Institute for Computational Molecular Science
Temple University, Philadelphia PA, USA.