GetTFDMatrices

nvmolkit.tfd.GetTFDMatrices(

mols: list,

useWeights: bool = True,

maxDev: str = 'equal',

symmRadius: int = 2,

ignoreColinearBonds: bool = True,

return_type: str = 'list',

) → list[list[float]] | list[ndarray] | list[Tensor]

Calculate TFD matrices for multiple molecules.

Parameters:

• mols – list of RDKit molecules, each with multiple conformers.

• useWeights – If True (default), use distance-based torsion weights.

• maxDev – Normalization mode (‘equal’ or ‘spec’).

• symmRadius – Radius for atom invariants (default: 2).

• ignoreColinearBonds – If True (default), ignore colinear bonds.

• return_type – Output format: ‘list’ (default): list of Python lists (RDKit-compatible). ‘numpy’: list of numpy float32 arrays. ‘tensor’: list of GPU torch.Tensors (no D2H copy).

Returns:

list of TFD matrices in the requested format.