Molecular Dynamics

Overview

SPARC run MD simulation leveraging the ASE MD engine framework. It supports both ab initio MD (AIMD) and machine-learning MD (ML-MD) using DeepPotential models. This module integrates thermostats, logging, checkpointing, and trajectory handling into a unified workflow.

It currently supports the NVT ensemble with Langevin thermostat or the Nosé-Hoover chain thermostat. Parameters such as damping time (for Nose) or friction coefficient (Langevin) needs to defined.

Each ML/MD iteration is saved in the corresponding iter_0000xx/02.dpmd directory, which includes the simulation log (log_file), the trajectory of atomic positions (extxyz or traj). It also write a checkpoint file to restart long simulations.

Simulation Outputs

>>> cat Iter1_dpmd.log
Time[ps]      Etot[eV]     Epot[eV]     Ekin[eV]    T[K]
0.0000        -112.0807    -112.8950       0.8143   300.0
0.0700        -111.6322    -112.7149       1.0828   398.9
0.1400        -112.4215    -113.3518       0.9303   342.7
0.2100        -112.9996    -113.6775       0.6779   249.8
0.2800        -112.6910    -113.7220       1.0310   379.8
0.3500        -112.8007    -113.2903       0.4896   180.4

Usage Examples

  • Nose-Hoover NVT Simulation:

sparc.src.ase_md.NoseNVT(atoms, timestep=1, temperature=300, tdamp=10, restart=False)[source]

Set up a Nose-Hoover chain NVT thermostat for MD simulation.

Parameters:

  • atoms (ase.Atoms) – The ASE Atoms object representing the system.

  • timestep (float, optional) – The simulation time step in femtoseconds (default is 1 fs).

  • temperature (float, optional) – The target temperature in Kelvin (default is 300 K).

  • tdamp (float, optional) – The damping time for the thermostat in femtoseconds (default is 10 fs).

  • restart (bool, optional) – If True, the simulation will be restarted from a checkpoint.

Returns:

dynamics – The initialized dynamics object using the Nose-Hoover chain thermostat.

Return type:

NoseHooverChainNVT

from ase import Atoms
from sparc.src.ase_md import NoseNVT

atoms = Atoms("H2O")
dyn = NoseNVT(atoms, temperature=300)
dyn.run(1000)

  • Langevin NVT Simulation:

sparc.src.ase_md.LangevinNVT(atoms, timestep=1, temperature=300, friction=0.01, restart=False)[source]

Set up a Langevin thermostat for NVT MD simulation.

Parameters:

  • atoms (ase.Atoms) – The ASE Atoms object representing the system.

  • timestep (float, optional) – The simulation time step in femtoseconds (default is 1 fs).

  • temperature (float, optional) – The target temperature in Kelvin (default is 300 K).

  • friction (float, optional) – The friction coefficient for the Langevin thermostat (default is 0.01 fs^-1).

  • restart (bool, optional) – If True, the simulation will be restarted from a checkpoint.

Returns:

dynamics – The initialized dynamics object using the Langevin thermostat.

Return type:

Langevin

from sparc.src.ase_md import LangevinNVT

dyn = LangevinNVT(atoms, temperature=300, friction=0.01)
dyn.run(1000)

  • Ab-initio Molecular Dynamics:

sparc.src.ase_md.ExecuteAbInitioDynamics(system, dyn, steps, pace, log_filename, trajfile, dir_name, name)[source]

Run an ab initio MD simulation.

Parameters:

  • system (ase.Atoms) – The ASE Atoms object representing the system.

  • dyn (dynamics object) – The initialized MD dynamics object.

  • steps (int) – The number of MD steps to run.

  • pace (int) – The interval (in steps) at which to log data and save checkpoints.

  • log_filename (str) – The filename for the MD log file.

  • trajfile (str) – The filename for the trajectory file.

  • dir_name (str) – The directory where log and trajectory files will be saved.

  • name (str) – A label for the simulation (e.g., the thermostat type).

Return type:

None

from sparc.src.ase_md import ExecuteAbInitioDynamics

ExecuteAbInitioDynamics(system=atoms, dyn=dyn, steps=500, pace=10,
                        log_filename="aimd.log", trajfile="aimd.traj",
                        dir_name="aimd_results", name="DFT_AIMD")

References

For more information on ASE, visit: https://wiki.fysik.dtu.dk/ase/