import keras as ks
from kgcnn.layers.modules import Input
from kgcnn.models.utils import update_model_kwargs
from keras.backend import backend as backend_to_use
from kgcnn.layers.scale import get as get_scaler
from kgcnn.models.casting import (template_cast_output, template_cast_list_input,
template_cast_list_input_docs, template_cast_output_docs)
from ._model import model_disjoint_crystal
# To be updated if model is changed in a significant way.
__model_version__ = "2023-11-28"
# Supported backends
__kgcnn_model_backend_supported__ = ["tensorflow", "torch", "jax"]
if backend_to_use() not in __kgcnn_model_backend_supported__:
raise NotImplementedError("Backend '%s' for model 'PAiNN' is not supported." % backend_to_use())
# Implementation of CGCNN in `keras` from paper:
# Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties
# Tian Xie and Jeffrey C. Grossman
# https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.145301
model_crystal_default = {
'name': 'CGCNN',
'inputs': [
{'shape': (None,), 'name': 'node_number', 'dtype': 'int64'},
{'shape': (None, 3), 'name': 'node_frac_coordinates', 'dtype': 'float64'},
{'shape': (None, 2), 'name': 'edge_indices', 'dtype': 'int64'},
{'shape': (None, 3), 'name': 'cell_translations', 'dtype': 'float32'},
{'shape': (3, 3), 'name': 'lattice_matrix', 'dtype': 'float64'},
# {'shape': (None, 1), 'name': 'multiplicities', 'dtype': 'float32'}, # For asu"
# {'shape': (None, 4, 4), 'name': 'symmops', 'dtype': 'float64'},
{"shape": (), "name": "total_nodes", "dtype": "int64"},
{"shape": (), "name": "total_edges", "dtype": "int64"}
],
"input_tensor_type": "padded",
"input_embedding": None, # deprecated
"cast_disjoint_kwargs": {},
"input_node_embedding": {"input_dim": 95, "output_dim": 64},
'representation': 'unit', # None, 'asu' or 'unit'
'expand_distance': True,
'make_distances': True,
'gauss_args': {'bins': 40, 'distance': 8, 'offset': 0.0, 'sigma': 0.4},
'depth': 3,
"verbose": 10,
'conv_layer_args': {
'units': 64,
'activation_s': 'softplus',
'activation_out': 'softplus',
'batch_normalization': True,
},
'node_pooling_args': {'pooling_method': 'scatter_mean'},
"output_embedding": "graph",
"output_to_tensor": None, # deprecated
"output_scaling": None,
"output_tensor_type": "padded",
'output_mlp': {'use_bias': [True, False], 'units': [64, 1],
'activation': ['softplus', 'linear']},
}
[docs]@update_model_kwargs(model_crystal_default, update_recursive=0, deprecated=["input_embedding", "output_to_tensor"])
def make_crystal_model(inputs: list = None,
input_tensor_type: str = None,
input_embedding: dict = None, # noqa
input_node_embedding: dict = None,
cast_disjoint_kwargs: dict = None,
representation: str = None,
make_distances: bool = None,
conv_layer_args: dict = None,
expand_distance: bool = None,
depth: int = None,
name: str = None,
verbose: int = None, # noqa
gauss_args: dict = None,
node_pooling_args: dict = None,
output_to_tensor: dict = None, # noqa
output_mlp: dict = None,
output_embedding: str = None,
output_scaling: dict = None,
output_tensor_type: str = None
):
r"""Make `CGCNN <https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.145301>`__ graph network
via functional API.
Default parameters can be found in :obj:`kgcnn.literature.CGCNN.model_crystal_default`.
**Model inputs**:
Model uses the list template of inputs and standard output template.
Model supports :obj:`[node_attributes, node_frac_coordinates, bond_indices, lattice, cell_translations, ...]`
if representation='unit'` and `make_distances=True` or
:obj:`[node_attributes, node_frac_coords, bond_indices, lattice, cell_translations, multiplicities, symmops, ...]`
if `representation='asu'` and `make_distances=True`
or :obj:`[node_attributes, edge_distance, bond_indices, ...]`
if `make_distances=False` .
The optional tensor :obj:`multiplicities` is a node-like feature tensor with a single value that gives
the multiplicity for each node.
The optional tensor :obj:`symmops` is an edge-like feature tensor with a matrix of shape `(4, 4)` for each edge
that defines the symmetry operation.
%s
**Model outputs**:
The standard output template:
%s
Args:
inputs (list): List of dictionaries unpacked in :obj:`tf.keras.layers.Input`. Order must match model definition.
input_tensor_type (str): Input type of graph tensor. Default is "padded".
input_embedding (dict): Deprecated in favour of input_node_embedding etc.
cast_disjoint_kwargs (dict): Dictionary of arguments for casting layers.
input_node_embedding (dict): Dictionary of embedding arguments for nodes unpacked in :obj:`Embedding` layers.
make_distances (bool): Whether input is distance or coordinates at in place of edges.
expand_distance (bool): If the edge input are actual edges or node coordinates instead that are expanded to
form edges with a gauss distance basis given edge indices. Expansion uses `gauss_args`.
representation (str): The representation of unit cell. Can be either `None`, 'asu' or 'unit'. Default is 'unit'.
conv_layer_args (dict):
depth (int): Number of graph embedding units or depth of the network.
verbose (int): Level of verbosity.
name (str): Name of the model.
gauss_args (dict): Dictionary of layer arguments unpacked in :obj:`GaussBasisLayer` layer.
node_pooling_args (dict): Dictionary of layer arguments unpacked in :obj:`PoolingNodes` layers.
output_embedding (str): Main embedding task for graph network. Either "node", "edge" or "graph".
output_mlp (dict): Dictionary of layer arguments unpacked in the final classification :obj:`MLP` layer block.
Defines number of model outputs and activation.
output_scaling (dict): Dictionary of layer arguments unpacked in scaling layers. Default is None.
output_tensor_type (str): Output type of graph tensors such as nodes or edges. Default is "padded".
output_to_tensor (bool): Deprecated in favour of `output_tensor_type` .
Returns:
:obj:`keras.models.Model`
"""
# Make input
model_inputs = [Input(**x) for x in inputs]
d_in = template_cast_list_input(
model_inputs,
input_tensor_type=input_tensor_type,
cast_disjoint_kwargs=cast_disjoint_kwargs,
mask_assignment=[0, 0 if make_distances else 1, 1, 1, None] + ([0, 1] if representation == "asu" else []),
index_assignment=[None, None, 0, None, None] + ([None, None] if representation == "asu" else [])
)
if representation == "asu":
n, x, djx, img, lattice, m, sym, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges = d_in
else:
n, x, djx, img, lattice, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges = d_in
m, sym = None, None
# Wrapp disjoint model
out = model_disjoint_crystal(
[n, m, x, sym, djx, img, lattice, batch_id_node, batch_id_edge, count_nodes, count_edges],
use_node_embedding=("int" in inputs[0]['dtype']) if input_node_embedding is not None else False,
representation=representation,
output_embedding=output_embedding,
input_node_embedding=input_node_embedding,
expand_distance=expand_distance,
conv_layer_args=conv_layer_args,
make_distances=make_distances,
depth=depth,
gauss_args=gauss_args,
node_pooling_args=node_pooling_args,
output_mlp=output_mlp
)
if output_scaling is not None:
scaler = get_scaler(output_scaling["name"])(**output_scaling)
if scaler.extensive:
# Node information must be numbers, or we need an additional input.
out = scaler([out, n, batch_id_node])
else:
out = scaler(out)
# Output embedding choice
out = template_cast_output(
[out, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges],
output_embedding=output_embedding,
output_tensor_type=output_tensor_type,
input_tensor_type=input_tensor_type,
cast_disjoint_kwargs=cast_disjoint_kwargs
)
model = ks.models.Model(inputs=model_inputs, outputs=out, name=name)
model.__kgcnn_model_version__ = __model_version__
if output_scaling is not None:
def set_scale(*args, **kwargs):
scaler.set_scale(*args, **kwargs)
setattr(model, "set_scale", set_scale)
return model
make_crystal_model.__doc__ = make_crystal_model.__doc__ % (template_cast_list_input_docs, template_cast_output_docs)