aenet.torch_featurize.RadialBasis

class aenet.torch_featurize.RadialBasis(*args: Any, **kwargs: Any)

Radial basis functions combining Chebyshev polynomials with cutoff.

Implements: G_rad = T_n(r) * fc(r)

Parameters:

• rad_order (int) – Maximum order for radial Chebyshev polynomials

• rad_cutoff (float) – Radial cutoff radius in Angstroms

• min_cutoff (float, optional) – Minimum cutoff (inner radius) in Angstroms (default: 0.55)

• dtype (torch.dtype, optional) – Data type for computations (default: torch.float64)

Examples

>>> rad_basis = RadialBasis(rad_order=10, rad_cutoff=4.0)
>>> distances = torch.tensor([1.0, 2.0, 3.0], dtype=torch.float64)
>>> G_rad = rad_basis(distances)  # Shape: (3, 11) for orders 0-10

__init__(rad_order: int, rad_cutoff: float, min_cutoff: float = 0.55, dtype: torch.dtype = torch.float64)

Methods

__init__(rad_order, rad_cutoff[, ...])
forward(distances)	Evaluate radial symmetry functions.
forward_with_derivatives(distances)	Evaluate radial symmetry functions with derivatives.