write_dump command


write_dump group-ID style file dump-args modify dump_modify-args
  • group-ID = ID of the group of atoms to be dumped
  • style = any of the supported dump styles
  • file = name of file to write dump info to
  • dump-args = any additional args needed for a particular dump style
  • modify = all args after this keyword are passed to dump_modify (optional)
  • dump-modify-args = args for dump_modify (optional)


write_dump all atom dump.atom
write_dump subgroup atom dump.run.bin
write_dump all custom dump.myforce.* id type x y vx fx
write_dump flow custom dump.%.myforce id type c_myF[3] v_ke modify sort id
write_dump all xyz system.xyz modify sort id element O H
write_dump all image snap*.jpg type type size 960 960 modify backcolor white
write_dump all image snap*.jpg element element &
   bond atom 0.3 shiny 0.1 ssao yes 6345 0.2 size 1600 1600  &
   modify backcolor white element C C O H N C C C O H H S O H


Dump a single snapshot of atom quantities to one or more files for the current state of the system. This is a one-time immediate operation, in contrast to the dump command which will will set up a dump style to write out snapshots periodically during a running simulation.

The syntax for this command is mostly identical to that of the dump and dump_modify commands as if they were concatenated together, with the following exceptions: There is no need for a dump ID or dump frequency and the keyword modify is added. The latter is so that the full range of dump_modify options can be specified for the single snapshot, just as they can be for multiple snapshots. The modify keyword separates the arguments that would normally be passed to the dump command from those that would be given the dump_modify. Both support optional arguments and thus LAMMPS needs to be able to cleanly separate the two sets of args.

Note that if the specified filename uses wildcard characters “*” or “%”, as supported by the dump command, they will operate in the same fashion to create the new filename(s). Normally, dump image files require a filename with a “*” character for the timestep. That is not the case for the write_dump command; no wildcard “*” character is necessary.


All restrictions for the dump and dump_modify commands apply to this command as well, with the exception of the dump image filename not requiring a wildcard “*” character, as noted above.

Since dumps are normally written during a run or energy minimization, the simulation has to be ready to run before this command can be used. Similarly, if the dump requires information from a compute, fix, or variable, the information needs to have been calculated for the current timestep (e.g. by a prior run), else LAMMPS will generate an error message.

For example, it is not possible to dump per-atom energy with this command before a run has been performed, since no energies and forces have yet been calculated. See the variable doc page section on Variable Accuracy for more information on this topic.


The defaults are listed on the doc pages for the dump and dump image and dump_modify commands.