nvalchemi.dynamics.mep.NEBMethod#

class nvalchemi.dynamics.mep.NEBMethod(tangent_weights_fn=<Function improved_tangent_weights(energy_prev: Any, energy_curr: Any, energy_next: Any)>, effective_force_fn=<Function neb_effective_force(physical_force: Any, tangent: Any, force_dot_tangent: Any, k_plus: Any, k_minus: Any, norm_d_plus: Any, norm_d_minus: Any, energy_prev: Any, energy_curr: Any, energy_next: Any, path_energy_ref: Any, path_energy_max: Any)>, climbing_force_fn=<Function climbing_image_effective_force(physical_force: Any, tangent: Any, force_dot_tangent: Any)>)[source]#

Configure the device equations used to construct NEB forces.

Equation functions must be importable module-level functions decorated with warp.func(). The hook prepares their kernel key during setup.

Parameters:
  • tangent_weights_fn (wp.Function, optional) – Return the forward and backward weights used to construct the path tangent from adjacent image displacements.

  • effective_force_fn (wp.Function, optional) – Construct the effective force for an active, non-climbing interior image using the stored-tangent or Gram-statistics signature.

  • climbing_force_fn (wp.Function, optional) – Construct the effective force for a climbing image.

Examples

Override only the tangent weighting while retaining regular NEB and climbing-image force laws:

import warp as wp

from nvalchemi.dynamics.mep import NEBMethod

@wp.func
def central_tangent_weights(
    energy_prev: float,
    energy_curr: float,
    energy_next: float,
):
    return 1.0, 1.0

method = NEBMethod(tangent_weights_fn=central_tangent_weights)
classmethod from_key(key)[source]#

Restore the equation functions from a prepared kernel key.

Parameters:

key (str) – Key produced by to_key().

Returns:

Method using the imported equation functions.

Return type:

NEBMethod

to_key()[source]#

Return the prepared kernel key containing importable equation paths.

Returns:

Kernel kind and dotted paths for the three equation functions.

Return type:

str