Dear all,
I am using quasicontinuum method to do some simulations. For post processing, I have to interpolate all the atoms and the model is too large for normal software to handle. But I actually only need to extract the dislocation inside the model. Is there any software parallelized that I can use to extract the dislocation inside the simulations? Thanks very much!
Best!
Hao
Dear all,
I am using quasicontinuum method to do some simulations. For post
processing, I have to interpolate all the atoms and the model is too large
for normal software to handle. But I actually only need to extract the
dislocation inside the model. Is there any software parallelized that I can
use to extract the dislocation inside the simulations? Thanks very much!
can you explain a bit more in detail what you understand by "extract
the dislocation" and in what kind of (file) format you would want to
access/read/output it?
axel.
If a LAMMPS compute can identify what
you define as a dislocation, then you could
read in one or more snapshots (in a LAMMPS dump
file format) via the rerun command, invoke the
compute, and do another dump that only outputs
the atoms around the dislocation (based on
thresholdng the numeric
value returned by the compute).
Possible computes are compute centro/atom,
cna/atom, coord/atom, etc. Or you could
write your own compute to take an atoms
and its neighbors and flag it as dislocation
or not.
Steve
Steve
Thanks very much, Steve and Axel. I am trying to do that.
Dear All
How to dump that only outputs the atoms around the dislocation (based on thresholdng the numeric value returned by the compute). Use the group command or use the if command? Thanks very much!
Best!
Hao
Dear All
How to dump that only outputs the atoms around the dislocation (based on
thresholdng the numeric value returned by the compute). Use the group
command or use the if command? Thanks very much!
neither.
the "if" command is for branching in input scripts and totally
inapplicable for this.
and the dump command doesn't handle dynamic groups, if i remember correctly.
instead use "dump_modify thresh" where any per-atom property that can
be output from a custom dump can be used to select atoms to output.
since this includes computes, fixes and atom style variables, those
selections can be fairly complex, if needed.
axel.
Also since the number of atoms in the quasicontinuum method can be as large as 20 billion or more, can I make several data files and read them by several read_data commands? Thanks very much!
Also since the number of atoms in the quasicontinuum method can be as large
as 20 billion or more, can I make several data files and read them by
several read_data commands? Thanks very much!
please consult the documentation on the restrictions of the read_data
command and how you can invoke it multiple times. reading and parsing
large text mode files is particularly time consuming and data files
need to be read/processed multiple times.
thus consider building your system with the create_atoms command
instead and using (much faster to read and write) binary restart
files.
axel.
Thanks very much. Let me try to do that.