Skip to content

Output

To write particles out to disk, generate a filename — usually with timestep_file, so each dump gets a name unique to the current iteration — then call one of the writer operators below.

Generating an output filename

Syntax
- timestep_file: <format_string>
Parameters
format:  string, required  # printf-style format string, e.g. "output_%09d".

timestep_file builds a filename from format and the current timestep, and outputs it as filename, which the writer operators below pick up automatically through same-name slot binding. format goes straight into a real snprintf call with timestep as its only argument — it must contain exactly one integer conversion (%d, %09d, …); a format string with more than one %-conversion isn't supported and reads past the supplied argument.

Usage example
- timestep_file: "paraview/output_%09d"
- write_paraview: { fields: [ mass, vz ] }

Logging what was written

exaStamp's own default snapshot configuration (data/config/config_snapshot.msp) additionally logs the filename to the console right before writing, using the generic message operator (mesg/endl) with its mesg slot rebound to the filename that timestep_file just produced:

- timestep_file: "paraview/output_%09d"
- message: { mesg: "Write paraview file", endl: false }
- print_dump_file:
    rebind: { mesg: filename }
    body:
      - message: { endl: true }
- write_paraview

print_dump_file isn't a distinct operator here — it's just a locally-chosen aggregate name wrapping message with a rebind, using the generic aggregate/rebind mechanism (see YAML extensions). The simple form above (just timestep_file + the writer) is enough for everyday use.

Writers

write_paraview

Syntax
write_paraview:
  filename: <string>
  fields: [<string>, ...]
  binary_mode: <bool>
  compression: <string>
  write_box: <bool>
  write_ghost: <bool>
Parameters
filename:     string, default "output"          # Output file name.
fields:       list of strings, default [".*"]   # Regular expressions selecting which fields to write (default = everything).
binary_mode:  bool, default true                # Write in binary mode (with compression) rather than ASCII.
compression:  string, default "default"         # Compression level for binary mode.
write_box:    bool, default true                # Also write the domain box.
write_ghost:  bool, default false               # Also write ghost particles.

Writes particle positions and attached fields in parallel, in ParaView-readable .vtp format.

Usage example
write_paraview:
  fields: [ type, vx, vy, vz, ep ]
  write_ghost: false

Note

filename names a directory, not a single file: the operator writes one .vtp piece per MPI rank under <filename>/, a master <filename>.pvtp tying them together, and (if write_box) a <filename>/box.vtp. Unlike write_xyz below, there's no units/field_alias equivalent here, and no simulation-time metadata is written into the file at all (confirmed unimplemented in the source, not just undocumented).

write_xyz

Syntax
write_xyz:
  filename: <string>
  fields: [<string>, ...]
  units: { <field>: <unit> }
  field_alias: { <field>: <alias> }
  ghost: <bool>
Parameters
filename:       string, default "output"                                 # Output file name.
fields:         list of strings, default [".*"]                          # Regular expressions selecting which fields to write.
units:          map, default {position: m, velocity: m/s, force: m/s/kg}  # Units to use for specific fields (yes, "m/s/kg" for force — verified verbatim in source).
field_alias:    map, default {position: pos, velocity: vel}               # Shorter output-header names for specific fields.
ghost:          bool, default false                                      # Also write ghost particles (note: named `ghost` here, `write_ghost` on write_paraview).
physical_time:  float, default 0.0                                       # Stamped into the file's `Time=` header field.

exaNBody's generic .xyz writer, in extended-XYZ format — despite the format's usual reputation, it isn't limited to positions/types/ids: any field can be selected the same way as write_paraview. Unlike write_paraview, it does write simulation-time metadata (Time=<physical_time>) into the file's header line, alongside a Properties= spec built from units/field_alias.

write_lmp

Syntax
write_lmp:
  filename: <string>
  triclinic: <bool>
  write_velocities: <bool>
Parameters
filename:          string, default "output.lmp"  # Output file name.
species:           required                      # Used for the Masses section (one line per species).
triclinic:         bool, default false            # false doesn't force orthogonal — see note below.
write_velocities:  bool, default true             # Include the Velocities section.

Writes a .data file with an Atoms section in atomic style (fixed, the style cannot be selected).

Note

triclinic: false isn't a hard override: the box is still auto-detected as triclinic (and written with xy/xz/yz tilt factors) if the domain's actual off-diagonal xform terms exceed a small tolerance. Set triclinic: true to always force triclinic output; false only means "don't force it beyond what auto-detection already decides."