restart2data

Hi,
It used to be possible to take a binary restart file that produced a “Did not assign all atoms correctly” error, run it through the restart2data utility and have a usable data file. I’ve noticed that the restart2data.cpp file is no longer distributed in the tools directory (at least as of 1 Feb, although the tools/Makefile still references it). This is a problem now that the restart-> data conversion is built in to lammps, since lammps doesn’t accept the restart file, even for the conversion process.

The context for this is the respa restart problem which surfaced recently, using a “restart 10000 A.rst B.rst” line. It seems like two potential workarounds are a command that writes periodic data files like the “restart” command or including a restart2data program that would again allow for working around the strict pbc check. Are there any other simple fixes that I could use to get around this issue?

Thanks,
Mike

Hi,
It used to be possible to take a binary restart file that produced a "Did
not assign all atoms correctly" error, run it through the restart2data
utility and have a usable data file. I've noticed that the restart2data.cpp
file is no longer distributed in the tools directory (at least as of 1 Feb,
although the tools/Makefile still references it). This is a problem now
that the restart-> data conversion is built in to lammps, since lammps
doesn't accept the restart file, even for the conversion process.

The context for this is the respa restart problem which surfaced recently,
using a "restart 10000 A.rst B.rst" line. It seems like two potential

this issue is not on my radar and considering how busy steve and
others have been over the last months, it is likely that this is on
somebody's todo list. if you want to report a problem that is not
"lost", please use the issue tracker that i have set up here:

and remember to attach a minimal input deck that allows to reproduce
the issue and debug it.

workarounds are a command that writes periodic data files like the "restart"
command or including a restart2data program that would again allow for
working around the strict pbc check. Are there any other simple fixes that
I could use to get around this issue?

the proper approach would be to resolve the inconsistencies in the
restart files. i don't think that reactivating restart2data is a
proper solution.

axel.

The solution is to fix rRESPA so that it writes correct restart
files. This was a bug introduced into rRESPA recently in
the process of fixing another rRESPA issue to do with per-atom
energies. Rolf has been looking into it and sent some updated
files. So we’ll probably have a fix shortly.

Steve

Another thought. We could add an option to
the read_restart command to allow it to read
files with atoms that are out-of-bounds, like rRESPA
is currently producing. There are other reasons
not to do this generally, but it might be useful with
the -r command-line switch, which is meant to take
the place of the old restart2data tool. I.e. it just
reads a restart file and writes a data file without
running the rest of LAMMPS. Currently, I imagine it
will give the same error with the restart file you
are currently getting.

Steve