nvalchemi.enhanced_sampling.LowerWall#

class nvalchemi.enhanced_sampling.LowerWall(cv, threshold, stiffness=10.0, *, name='lower_wall', exponent=2.0, compute_stress=True)[source]#

Penalise the CV for falling below a threshold.

\[E_b = \frac{k}{p}\,\max(s_0 - s,\,0)^p\]
Parameters:
  • cv (Callable[[Batch], Tensor]) – Differentiable cv(batch) -> Tensor[B, D].

  • threshold (Tensor | float) – Lower bound s_0.

  • stiffness (Tensor | float) – Penalty prefactor k. Default 10.0.

  • name (str) – Unique bias identifier. Default "lower_wall".

  • exponent (float) – Power p. Default 2.0.

  • compute_stress (bool) – Passed through to ConservativeBias.

extra_repr()#

Format the model config for nn.Module.__repr__.

Parameters:

self (Any)

Return type:

str