Skip to content

Latest commit

 

History

History
81 lines (79 loc) · 8.36 KB

README_input_parameters.md

File metadata and controls

81 lines (79 loc) · 8.36 KB

pymatnext input TOML parameters

This reference documents every field emitted by the current pymatnext --defaults command. The itemized list mirrors the nesting of the TOML tables and the fields that can be set in each table. Each field includes its Pydantic type, TOML default, and description; fields without a default must be specified by the user. The TOML string "_NONE_" is used to set a Python None value.

  • [general] # type: GeneralParams; general parameters
    • output_filename_prefix = "NS" # type: str; prefix for all output files
    • output_filename_prefix_extra = "" # type: str; extra string to add to output_filename_prefix, designed for easy per-realization overriding
    • random_seed (no default; must be specified by the user) # type: PositiveInt; seed for random number generator
    • max_iter = "_NONE_" # type: PositiveInt | None; maximum NS iteration
    • stdout_report_interval_s = 60 # type: int; interval in seconds between reports to stdout, < 0 for no reports
    • sample_interval = 1 # type: int; interval in NS iterations between saved NS samples, <= 0 to disable
    • traj_interval = 100 # type: int; interval in NS iterations between saved NS configurations, <= 0 to disable
    • snapshot_interval = 10000 # type: int; interval in NS iterations between restart snapshots, <= 0 to disable
    • snapshot_save_old = 2 # type: int; how many old snapshots so save
    • clone_history = false # type: bool; save a full history of which config was cloned at each iteration
    • [general.step_size_tune] # type: StepSizeTuneParams; parameters for tuning step sizes
      • interval = 1000 # type: PositiveInt; NS iteration interval between step-size tuning
      • n_configs = 1 # type: PositiveInt; number of configs to run when computing step size related statistics
      • min_accept_rate = 0.25 # type: PositiveFloat; minimum accept rate for tuning step size
      • max_accept_rate = 0.5 # type: PositiveFloat; maximum accept rate for tuning step size
      • adjust_factor = 1.25 # type: PositiveFloat; factor by which step size is multiplied/divided at each tuning iteration
    • [general.walk_traj_info] # type: WalkTrajectoryInfoParams; parameters for saving walk trajectories
      • iter_min = "_NONE_" # type: NonNegativeInt | None; first iteration at which walk trajectory is saved
      • iter_max = "_NONE_" # type: int | None; last iteration (inclusive) at which walk trajectory is saved, negative for no maximum
      • interval = "_NONE_" # type: PositiveInt | None; interval at which walk trajectory is saved
      • avg_times = [] # type: List[PositiveInt] | None; save walk trajectory with time averaging over these time scales
  • [ns] # type: NSParams; parameters for nested-sampling iteration process
    • n_walkers (no default; must be specified by the user) # type: int; Number of NS walkers (live points)
    • walk_length (no default; must be specified by the user) # type: int; Length of NS walk to produce a new, decorrelated config, in number of energy/force evaluations
    • step_size_tune_walk_length = "_NONE_" # type: PositiveInt | None; NS walk length used to tune step size
    • configs_module (no default; must be specified by the user) # type: str; module that defines configurations
    • initial_config_file = "_NONE_" # type: str | None; file with initial configurations, if not generated randomly
    • initial_max_val = "_NONE_" # type: float | None; value of overriding NS energy initial maxmimum
    • [ns.exit_conditions] # type: NSLoopExitParams; parameters for optional exit conditions
      • module = "_NONE_" # type: str | None; module that defines an ExitLoop class that checks for an exit condition, with __call__ method that takes current iteration and max value and returns a bool
      • module_kwargs = {} # type: Dict[str, Any]; arbitrary kwargs for ExitLoop constructor in addition to the NS object itself
  • [configs] # type: ASEAtomsParams; config-type-specific parameters
    • full_composition = "" # type: Composition (str | List[str | int]); composition that spans all possible elements that could appear
    • composition (no default; must be specified by the user) # type: Composition (str | List[str | int]); initial composition of configurations
    • n_atoms (no default; must be specified by the user) # type: int; number of atoms in each configuration
    • dims = 3 # type: int; number of dimensions (2 or 3)
    • pbc = [true, true, true] # type: List[bool]; periodicity of system along each cell vector
    • initial_rand_vol_per_atom (no default; must be specified by the user) # type: float; random initial configuration volume per atom
    • initial_rand_min_dist (no default; must be specified by the user) # type: float; random initial configuration minimum distance between atoms
    • initial_rand_n_tries = 10 # type: int; number of tries to get initial configuration that obeys minimum distances
    • [configs.calculator] # type: CalculatorParams; parameters for calculator of interatomic interactions
      • type (no default; must be specified by the user) # type: Literal["ASE", "LAMMPS"]; calculator type
      • args = {} # type: Dict[str, Any]; arbitrary args for calculator constructor. If 'ASE', 'module' with name of importable module defining a calc Calculator object. If 'LAMMPS', 'cmds': list of pair_style ... etc. lammps commands, 'types': dict with atomic numbers or chemical symbols as keys and lammps types as values, 'header': optional list of header commands, 'cmd_args': optional command args, 'log_file': optional file for lammps log, 'name': optional lammps shared lib name, 'activate_mliappy_kokkos': optional bool for mliappy kokkos, 'boundary': optional args for lammps boundary command
    • [configs.walk] # type: ASEAtomsWalkParams; parameters for NS walks
      • gmc_traj_len = 8 # type: int; length of GMC walks
      • cell_traj_len = 8 # type: int; length of cell move walks
      • type_traj_len = 8 # type: int; length of type (sGC) move walks
      • gmc_proportion = 0.0 # type: float; proportion of steps to do GMC walks with
      • cell_proportion = 0.0 # type: float; proportion of steps to do cell walks with
      • type_proportion = 0.0 # type: float; proportion of steps to do type (sGC) walks with
      • combined = false # type: bool; use NS walk function that combines all move types
      • [configs.walk.max_step_size] # type: MaxStepSizeParams; maximum step size for each move type
        • pos_gmc_each_atom = -0.1 # type: float; maximum GMC position move (for each atom), if negative multiplied by cube root of atomic volume
        • cell_volume_per_atom = -0.05 # type: float; maximum cell volume step (per atom), if negative multiplied by atomic volume
        • cell_shear_per_rt3_atom = -1.0 # type: float; maximum cell shear step (per natoms^1/3), if negative multiplied by cube root of atomic volume
        • cell_stretch = 0.2 # type: float; maximum cell stretch (unitless strain)
      • [configs.walk.step_size] # type: StepSizeParams; initial step size for each move type
        • pos_gmc_each_atom = -1.0 # type: float; initial GMC positions move (for each atom)
        • cell_volume_per_atom = -1.0 # type: float; initial cell volume step (per atom)
        • cell_shear_per_rt3_atom = -1.0 # type: float; initial cell shear step (per natoms^1/3)
        • cell_stretch = -1.0 # type: float; initial cell stretch step
      • [configs.walk.cell] # type: CellWalkParams; parameters controlling cell walks
        • min_aspect_ratio = 0.8 # type: float; minimum cell aspect ratio to accept
        • flat_V_prior = true # type: bool; use a prior independent of volume, rather than ensemble-correct V^natoms
        • pressure = 0.0 # type: float; applied pressure in eV/A^3 (or pressure_GPa in GPa)
        • [configs.walk.cell.submove_probabilities] # type: SubmoveProbabilitiesParams; parameters controlling probabilities of volume, shear, and stretch move types
          • volume = 0.7 # type: float; probability to try a cell volume move
          • shear = 0.15 # type: float; probability to try a cell shear move
          • stretch = 0.15 # type: float; probability to do a cell stretch move
      • [configs.walk.type] # type: TypeWalkParams; parameters controlling type (sGC) walks
        • sGC = false # type: bool; semi-grand-canonical (species change) moves.
        • mu = {} # type: Dict[PositiveInt, float]; dict with atomic numbers as keys and chemical potentials as values for semi-grand-canonical moves