Dear Axel and Ray,
My comment and questions are below. Thanks very much for the previous
advices!
ouch. that is a rather old version of LAMMPS. [...]
Well, I'm aware of that but I'm working on universisty and this is the
Lammp's version that is currently installed. I don't have permission to
change anything since I'm only student but I'll talk to the administrator
about upgrading lammps when he will come back from holidays. For this moment
I'm forced to work on this version of Lammps.
this is complete nonsense. lammps doesn't need anything that
requires administrator privileges. such software can be compiled
by yourself and you can place the executable where you can
store your data. i am doing it all the time and on many different
machines (including some of the largest supercomputers available
to researchers) and never needed an administrator to do anything
about that.
[...]
I still have problems with visualization. I tried with this line:
"dump snap all cfg 2 pdmsmixdump.txt.*cfg id type xs ys zs"
but I find it annoying that there are several dozens of files. Is there a
way to make a one file which will be understandable for AtomEye? And in
addition in AtomEye all my atoms are "C" type and i don't know why.
this is how atomeye works. you would have to use dump_modify
element to set the element names as needed.
I tried also with this:
"dump 1 all custom 10 xxx.txt id type xs ys zs"
but VMD still have problems with understanding this.
that is creating essentially the same file
as the atom style dump. the issue is that
there is no information about the elements.
type is just a number. VMD has no psychic
powers and knows what "type" is supposed
to stand for.
*however*, what you *can* do is to create/use
a psf file that matches your element names.
your data file, since it was created with VMD
has almost all the necessary information.
running the following commands on the VMD
console creates a .psf file that can be used
in combination with the dump files to convey
the necessary information to VMD:
topo readlammpsdata pdmsmix.data charge
set sel [atomselect top all]
$sel set segname PDMS
animate write psf pdmsmix.psf
quit
furthermore, the dump file also does not contain any
bonding information (for reax there isn't anything to dump in the first
place). [...]
Well, I undestand that in the begining there are no bonds in the system but
after running a simulation, ReaxFF should compute bond orders and bonds
should appeare. Maybe I don't attach the information about bonds into the
dump file? If that's true, how to include this type of information? Now when
yes, reax computes those properties. but those are not "bonds" in the
LAMMPS sense, i.e. defined as explicit force field interactions.
you can output the bond information that is computed by reax/c using this fix.
http://lammps.sandia.gov/doc/fix_reaxc_bonds.html
I load the date to VMD I have bonds from the 1st second of simulation, which
is weird for me becous I thought the bonds will appeare later. Correct me if
I'm wrong. I would be grateful for any kind of advices connected with
"dumping" the correct (and understandable for visualization software) data.
you have to spend some time with the VMD tutorials.
a visualization tool shows bond not because of any knowledge
of the bonding situation, but based on some rather simple
heuristics. also, in the case of VMD, bonding is considered
as in forcefield calculations, i.e. it usually imports bonds
from a .psf file. short of that, it uses a heuristic guess and
then *keeps* those bonds. the entire topology information
is only stored once, as for a regular force field calculation,
this doesn't change and hence a lot of storage can be
saved to allow for storing longer trajectories.
in short, there is no simple way. you can use the simple
heuristics of a viz program (but you have to feed it the
missing information, e.g. what types of atoms you have,
and work around cases where it gets it wrong) or use
your own heuristics or take the data stored by lammps
and do some scripting. you will have to spend some time
on this and develop some tools. i can provide some
suggestions for VMD, but there exists no perfect solution,
since reax is a new concept compared to traditional
force field calculations and there are only a small number
of people that use this. just do a little research on
the problem that people have with visualization data
from ab initio simulations. i had an entire section
of my home page dedicated to it, when i was still
working on/with some of those codes. you can find
the remnants here:
https://sites.google.com/site/akohlmey/redirect/cpmd-vmd.pdf?attredirects=0&d=1
because you "created" over 4000K worth of kinetic energy.
*everything decomposes" at that temperature. in your case
the cause is obvious. you didn't choose your simulation
parameters like the time step properly. running a reax based
simulation with hydrogens in them with the default timestep
for real units will result in a totally bogus simulation. you
need to use a much shorter timestep. what is a good time
step depends on many factors. please refer to your favorite
text book on MD for a detailed explanation.
Thanks for the advice. I've changed the timestep to 0.2 and now everything
looks good.
Step Temp E_pair E_mol TotEng Press
3 0 -228485.11 0 -228485.11 139.05638
203 299.18188 -231017.52 0 -228507.09 -235.05011
As you see the temperature is now circa 26C (almost room temperature). I
think that would be correct.
well, i guess that is by chance, since you were using NVE integration.
and what you see is just the result of the potential energy from
changing to reax
cheers,
axel.
