gt4sd.frameworks.cgcnn.model module

Model module.

Summary

Classes:

ConvLayer	Convolutional operation on graphs.
CrystalGraphConvNet	Create a crystal graph convolutional neural network for predicting total material properties.
Normalizer	Normalize a Tensor and restore it later.

Reference

class ConvLayer(atom_fea_len, nbr_fea_len)

Bases: Module

Convolutional operation on graphs.

__init__(atom_fea_len, nbr_fea_len)

Initialize ConvLayer.

Parameters

• atom_fea_len (int) – int Number of atom hidden features.

• nbr_fea_len (int) – int Number of bond features.

forward(atom_in_fea, nbr_fea, nbr_fea_idx)

Forward pass.

N: Total number of atoms in the batch. M: Max number of neighbors.

Parameters

• atom_in_fea (Tensor) – Variable(torch.Tensor) shape (N, atom_fea_len) Atom hidden features before convolution.

• nbr_fea (Tensor) – Variable(torch.Tensor) shape (N, M, nbr_fea_len) Bond features of each atom’s M neighbors.

• nbr_fea_idx (LongTensor) – torch.LongTensor shape (N, M) Indices of M neighbors of each atom.

Returns

nn.Variable shape (N, atom_fea_len)

Atom hidden features after convolution.

Return type

atom_out_fea

__annotations__ = {}

__doc__ = 'Convolutional operation on graphs.'

__module__ = 'gt4sd.frameworks.cgcnn.model'

class CrystalGraphConvNet(orig_atom_fea_len, nbr_fea_len, atom_fea_len=64, n_conv=3, h_fea_len=128, n_h=1, classification=False)

Bases: Module

Create a crystal graph convolutional neural network for predicting total material properties.

__init__(orig_atom_fea_len, nbr_fea_len, atom_fea_len=64, n_conv=3, h_fea_len=128, n_h=1, classification=False)

Initialize CrystalGraphConvNet.

Parameters

• orig_atom_fea_len (int) – int Number of atom features in the input.

• nbr_fea_len (int) – int Number of bond features.

• atom_fea_len (int) – int Number of hidden atom features in the convolutional layers.

• n_conv (int) – int Number of convolutional layers.

• h_fea_len (int) – int Number of hidden features after pooling.

• n_h (int) – int Number of hidden layers after pooling.

forward(atom_fea, nbr_fea, nbr_fea_idx, crystal_atom_idx)

Forward pass.

N: Total number of atoms in the batch. M: Max number of neighbors. N0: Total number of crystals in the batch.

Parameters

• atom_fea (Tensor) – Variable(torch.Tensor) shape (N, orig_atom_fea_len) Atom features from atom type.

• nbr_fea (Tensor) – Variable(torch.Tensor) shape (N, M, nbr_fea_len) Bond features of each atom’s M neighbors.

• nbr_fea_idx (LongTensor) – torch.LongTensor shape (N, M) Indices of M neighbors of each atom.

• crystal_atom_idx (LongTensor) – list of torch.LongTensor of length N0 Mapping from the crystal idx to atom idx.

Returns

nn.Variable shape (N, )

Atom hidden features after convolution.

Return type

prediction

pooling(atom_fea, crystal_atom_idx)

Pooling the atom features to crystal features.

N: Total number of atoms in the batch. N0: Total number of crystals in the batch.

Parameters

• atom_fea (Tensor) – Variable(torch.Tensor) shape (N, atom_fea_len) Atom feature vectors of the batch.

• crystal_atom_idx (LongTensor) – list of torch.LongTensor of length N0 Mapping from the crystal idx to atom idx.

Return type

Tensor

__annotations__ = {}

__doc__ = 'Create a crystal graph convolutional neural network for predicting total material properties.'

__module__ = 'gt4sd.frameworks.cgcnn.model'

class Normalizer(tensor)

Bases: object

Normalize a Tensor and restore it later.

__init__(tensor)

tensor is taken as a sample to calculate the mean and std.

norm(tensor)

Noramlize a tensor.

Parameters

tensor (Tensor) – tensor to be normalized.

Return type

Tensor

Returns

normalized tensor.

denorm(normed_tensor)

Denormalized tensor.

Parameters

tensor – tensor to be denormalized:

Return type

Tensor

Returns

denormalized tensor.

state_dict()

Return the state dict of normalizer.

Return type

Dict[str, Tensor]

Returns

dictionary including the used mean and std values.

__dict__ = mappingproxy({'__module__': 'gt4sd.frameworks.cgcnn.model', '__doc__': 'Normalize a Tensor and restore it later.', '__init__': <function Normalizer.__init__>, 'norm': <function Normalizer.norm>, 'denorm': <function Normalizer.denorm>, 'state_dict': <function Normalizer.state_dict>, 'load_state_dict': <function Normalizer.load_state_dict>, '__dict__': <attribute '__dict__' of 'Normalizer' objects>, '__weakref__': <attribute '__weakref__' of 'Normalizer' objects>, '__annotations__': {}})

__doc__ = 'Normalize a Tensor and restore it later.'

__module__ = 'gt4sd.frameworks.cgcnn.model'

__weakref__

list of weak references to the object (if defined)

load_state_dict(state_dict)

Return the state dict of normalizer.

Parameters

• mean – mean value to be used for the normalization.

• std – std value to be used for the normalization.

Return type

None