Hello Users,
I would like to understand a bit more about the output files from a parallel tempering run. I have attached an input file here (in.system). This is a toy file that I am using to figure out how the tempering is implemented etc…
- The output file “log.lammps” contains:
Step, T1, T2, …
Does this describe the replica-index at this instant in time present at the first
temperature, at the second temperature and so on?
-
I get the thermodynamic information in files: “log.lammps.0, log.lammps.1, …”
If I would like to look at the instantaneous potential energy of the system that is at 300K,
which file should I look at? Or, should this value be picked up from different files,
depending on which replica ID is currently at 300K?
-
My dump files are named: dump.300.system, dump.302.system, …
Is the file named dump.300.system, the one that has the configurations for 300K? It
appears obvious, but I thought I would get this confirmed.
Thanks for your help,
Cheers,
Bala
http://www.jncasr.ac.in/bala
in.system (782 Bytes)
Answers below:
Hello Users,
I would like to understand a bit more about the output files from a
parallel tempering run. I have attached an input file here (in.system). This
is a toy file that I am using to figure out how the tempering is implemented
etc..
1) The output file "log.lammps" contains:
Step, T1, T2, ...
Does this describe the replica-index at this instant in time present
at the first
temperature, at the second temperature and so on?
Yes, I think that's right.
2) I get the thermodynamic information in files: "log.lammps.0,
log.lammps.1, ....."
If I would like to look at the instantaneous potential energy of the
system that is at 300K,
which file should I look at? Or, should this value be picked up from
different files,
depending on which replica ID is currently at 300K?
The log and dump files are written by a processor (or processor group),
for the replica it owns. That replica changes T over time (due to exchanges).
In other words, which replica is at 300K is changing over time. So those
values are not in 1 file.
3) My dump files are named: dump.300.system, dump.302.system, ...
Is the file named dump.300.system, the one that has the configurations
for 300K? It
appears obvious, but I thought I would get this confirmed.
It depends on how you named your files, but the previous answer
should help. The dump files are also per-processor-group, not T.
See below:
Thanks Steve.
1) The line in my input for dump is:
====================
variable t world 300.0 302.0 305.0 309.0 314.0 319.0 324.0 329.0
dump 1 all custom 1000 dump.$t.system tag mol type xu yu zu
=====================
So, although the filename could be "dump.300.system", it could contain
configurations from other temperatures if that processor group had gotten
its configuration exchanged. Is this right?
yes, correct
2) Is there a way to flush the log.lammps file in a parallel tempering run?
You could put in a flush() statement after the file write. Search for
flush in the code and you'll see where this is done elsewhere.