Output Data¶

Output is controlled by the frequency variables already introduced in Global Control, together with the writer operators below. This page only covers particle-level output (the species/positions/velocities you defined earlier) — see the tip at the end for grid-level output.

Thermodynamic state (screen & file)¶

Printing the thermodynamic state to the screen and writing it to a .csv file are both enabled by default; only their frequency and format need to be set from global:

1
2
3
4
5

global:
  simulation_thermostate_screen_frequency: 10   # print to screen every 10 steps
  simulation_thermostate_file_frequency: 10     # append to file every 10 steps
  thermostate_file: "thermodynamic_state.csv"
  log_mode: mechanical

log_mode selects which columns print_thermodynamic_state prints to screen (mechanical, default/thermo_basic, thermo/thermo_full, vol_fluct_ortho[_basic], vol_fluct_tricl[_basic], or a custom ;-separated list of item names). The file writer (dump_thermodynamic_state) always writes the same fixed set of columns (step, time, energies, temperature, stress tensor, box, volume, density) to thermostate_file.

Binary restart file¶

write_restart is nop by default; assign it one of the predefined aggregates from config_restart.msp (see Restarts) to activate restart writing, at the frequency set by simulation_restart_frequency:

1
2
3
4

global:
  simulation_restart_frequency: 1000

write_restart: write_restart_atoms

This writes an exaNBody-native, MPI-IO binary dump (particles, velocities, domain, species and timestep) through write_dump_atoms, which can later be re-read with read_dump_atoms (see Reading an atoms restart file). It also accepts compression_level (zlib level, default 6) and max_part_size (file partition size, default: system value).

1
2
3
4
5
6
7
8

write_restart_atoms:
  - timestep_file: "atoms_%09d.MpiIO"
  - message: { mesg: "Write restart " , endl: false }
  - print_restart_file:
      rebind: { mesg: filename }
      body:
        - message: { endl: true }
  - write_dump_atoms

Snapshot file¶

write_snapshot is likewise nop by default; assign it a predefined aggregate to write snapshots for visualization, at the frequency set by simulation_snapshot_frequency. Two common formats:

1
2
3
4

global:
  simulation_snapshot_frequency: 1000

write_snapshot: write_snapshot_xyz

write_snapshot_xyz writes a plain-text .xyz file (species name and position per atom) through write_xyz_file.

1
2
3
4
5
6
7
8

write_snapshot_xyz:
  - timestep_file: "exaStamp_%09d.xyz"
  - message: { mesg: "Write xyz file" , endl: false }
  - print_dump_file:
      rebind: { mesg: filename }
      body:
        - message: { endl: true }
  - write_xyz_file

1
2
3
4

global:
  simulation_snapshot_frequency: 1000

write_snapshot: write_snapshot_paraview

write_snapshot_paraview writes a Paraview/VTK file through write_paraview, including every available per-particle field (position, velocity, force, type, MPI rank, ...) by default — narrow it down with the fields: list of regular expressions if needed (e.g. fields: [ "type", "vx|vy|vz" ]).

1
2
3
4
5
6
7
8

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