Configuration Guide

Default Values for Input Files

The amaceing_toolkit provides a comprehensive set of default configurations for various molecular simulation engines. These configurations can be customized to suit your specific needs.

Configuration files are located in the src/amaceing_toolkit/default_configs/ directory:

  • CP2K: cp2k_configs.py

  • MACE: mace_configs.py

  • MatterSim: mattersim_configs.py

  • SevenNet: sevennet_configs.py

  • ORB: orb_configs.py

  • Grace: grace_configs.py

Each configuration file contains a dictionary of settings organized by simulation type (MD, GEO_OPT, etc.) and other parameters relevant to each engine.

Additional Configuration Files

  • CP2K basis sets and XC functionals: cp2k_kind_data.py contains basis set information and exchange-correlation functional definitions for CP2K.

  • MACE atomic energies: mace_e0s.py contains reference atomic energies used by MACE models.

Example Configuration Structure

CP2K configuration example:

'default': {
  'cp2k_newer_than_2023x': 'y',
  'coord_file': 'coord.xyz',
  'box_cubic': 'y',
  'run_type': 'MD',
  'use_default_input': 'y',
  'MD': {
    'ensemble': 'NVT',
    'nsteps': 200000,
    'timestep': 0.5,
    'temperature': 300,
    'print_forces': 'y',
    'print_velocities': 'n',
    'xc_functional': 'BLYP',
    'equilibration_steps': 2000,
    'pressure_b': 1.0,
    'equilibration_run': 'y'
  },
  'GEO_OPT': {...}, ...
}

MACE configuration example:

'default': {
  'coord_file': 'coord.xyz',
  'box_cubic': 'pbc',
  'run_type': 'MD',
  'use_default_input': 'y',
  'MD': {
    'simulation_environment': 'ase',
    'foundation_model': 'mace_mp',
    'model_size': 'small',
    'dispersion_via_simenv': 'n',
    'temperature': '300',
    'pressure': '1.0',
    'thermostat': 'Langevin',
    'nsteps': 2000000,
    'write_interval': 10,
    'timestep': 0.5,
    'log_interval': 100,
    'print_ext_traj': 'y'
  },
  'GEO_OPT': {...}, ...
}

HPC Runscript Creator

The toolkit provides templates for creating HPC job submission scripts for different computing environments. These templates support LSF and SLURM workload managers.

The path to the runscript templates is defined in:

src/amaceing_toolkit/default_configs/hpc_setup_path.txt

The toolkit includes templates for various simulation engines and computing platforms:

  • CP2K: cp2k, cp2k_local

  • MACE: mace_py_cpu, mace_py_gpu, mace_lmp_cpu, mace_lmp_gpu

  • MatterSim: mattersim_py_cpu, mattersim_py_gpu, mattersim_ft_gpu

  • SevenNet: sevennet_py_cpu, sevennet_py_gpu, sevennet_lmp_cpu, sevennet_lmp_gpu

  • ORB: orb_py_cpu, orb_py_gpu

  • Grace: grace_py_cpu, grace_py_gpu, grace_lmp_cpu, grace_lmp_gpu, grace_ft_gpu

These templates can be customized to match your specific HPC environment requirements.

Logger

The toolkit includes built-in logging capabilities for tracking runs and models:

  • Run logger: Keeps track of simulation runs in run_logger.log

  • Model logger: Records information about machine learning models in finetuned_models.log

The logger files are located in:

src/amaceing_toolkit/runs/

To view logs, use the utility commands:

amaceing_utils --logger=run
amaceing_toolkit --logger=runexport
amaceing_utils --logger=model

Custom Configurations

Users can create their own configuration profiles by modifying the default configuration files or by creating new configuration entries. These custom configurations can be specified when creating new simulation inputs using the toolkit’s API or command-line interface.

These modified configurations have to be hard-coded into the input wrapper.