nvt_vrescale_step

torch_sim.integrators.nvt.nvt_vrescale_step(model, state, *, dt, kT, tau=None)[source]

Perform one complete V-Rescale dynamics integration step.

This function implements the canonical sampling through velocity rescaling (V-Rescale) thermostat combined with velocity Verlet integration. The V-Rescale thermostat samples the canonical distribution by rescaling velocities with a properly chosen random factor that ensures correct canonical sampling.

Parameters:

  • model (ModelInterface) – Neural network model that computes energies and forces. Must return a dict with ‘energy’ and ‘forces’ keys.

  • state (NVTVRescaleState) – Current system state containing positions, momenta, forces

  • dt (float | Tensor) – Integration timestep, either scalar or shape [n_systems]

  • kT (float | Tensor) – Target temperature in energy units, either scalar or with shape [n_systems]

  • tau (float | Tensor | None) – Thermostat relaxation time controlling the coupling strength, either scalar or with shape [n_systems]. Defaults to 100*dt.

  • seed – Random seed for reproducibility

Returns:

Updated state after one complete V-Rescale step with new positions,

momenta, forces, and energy

Return type:

MDState

Notes

  • Uses V-Rescale thermostat for proper canonical ensemble sampling

  • Unlike Berendsen thermostat, V-Rescale samples the true canonical distribution

  • Integration sequence: V-Rescale rescaling + Velocity Verlet step

  • The rescaling factor follows the distribution derived in Bussi et al.

References

Bussi G, Donadio D, Parrinello M. “Canonical sampling through velocity rescaling.” The Journal of chemical physics, 126(1), 014101 (2007).