torch_featurize API Reference

The aenet.torch_featurize module provides PyTorch-based implementations of atomic environment descriptors and neighbor lists.

Module Overview

Main Classes

ChebyshevDescriptor(*args, **kwargs)	Complete featurization pipeline using Chebyshev descriptors.
BatchedFeaturizer(*args, **kwargs)	Batched featurization for multiple structures.
TorchNeighborList(cutoff[, atom_types, ...])	PyTorch-based neighbor list for atomic structures.

Basis Functions

ChebyshevPolynomials(*args, **kwargs)	Vectorized Chebyshev polynomial evaluation using cosine form.
RadialBasis(*args, **kwargs)	Radial basis functions combining Chebyshev polynomials with cutoff.
AngularBasis(*args, **kwargs)	Angular basis functions using Chebyshev polynomials.