core.models.dimenet_plus_plus

Copyright (c) Meta, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree.

—

This code borrows heavily from the DimeNet implementation as part of pytorch-geometric: rusty1s/pytorch_geometric. License:

—

Copyright (c) 2020 Matthias Fey <matthias.fey@tu-dortmund.de>

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Attributes

sym

Classes

InteractionPPBlock	Base class for all neural network modules.
OutputPPBlock	Base class for all neural network modules.
DimeNetPlusPlus	DimeNet++ implementation based on klicperajo/dimenet.
DimeNetPlusPlusWrapEnergyAndForceHead	Base class for all neural network modules.
DimeNetPlusPlusWrap	DimeNet++ implementation based on klicperajo/dimenet.
DimeNetPlusPlusWrapBackbone	DimeNet++ implementation based on klicperajo/dimenet.

Module Contents

core.models.dimenet_plus_plus.sym = None

class core.models.dimenet_plus_plus.InteractionPPBlock(hidden_channels: int, int_emb_size: int, basis_emb_size: int, num_spherical: int, num_radial: int, num_before_skip: int, num_after_skip: int, act='silu')

Bases: torch.nn.Module

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

act

lin_rbf1

lin_rbf2

lin_sbf1

lin_sbf2

lin_kj

lin_ji

lin_down

lin_up

layers_before_skip

lin

layers_after_skip

reset_parameters() → None

forward(x, rbf, sbf, idx_kj, idx_ji)

class core.models.dimenet_plus_plus.OutputPPBlock(num_radial: int, hidden_channels: int, out_emb_channels: int, out_channels: int, num_layers: int, act: str = 'silu')

Bases: torch.nn.Module

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

act

lin_rbf

lin_up

lins

lin

reset_parameters() → None

forward(x, rbf, i, num_nodes: int | None = None)

class core.models.dimenet_plus_plus.DimeNetPlusPlus(hidden_channels: int, out_channels: int, num_blocks: int, int_emb_size: int, basis_emb_size: int, out_emb_channels: int, num_spherical: int, num_radial: int, cutoff: float = 5.0, envelope_exponent: int = 5, num_before_skip: int = 1, num_after_skip: int = 2, num_output_layers: int = 3, act: str = 'silu')

Bases: torch.nn.Module

DimeNet++ implementation based on klicperajo/dimenet.

Parameters:

• hidden_channels (int) – Hidden embedding size.

• out_channels (int) – Size of each output sample.

• num_blocks (int) – Number of building blocks.

• int_emb_size (int) – Embedding size used for interaction triplets

• basis_emb_size (int) – Embedding size used in the basis transformation

• out_emb_channels (int) – Embedding size used for atoms in the output block

• num_spherical (int) – Number of spherical harmonics.

• num_radial (int) – Number of radial basis functions.

• cutoff – (float, optional): Cutoff distance for interatomic interactions. (default: 5.0)

• envelope_exponent (int, optional) – Shape of the smooth cutoff. (default: 5)

• num_before_skip – (int, optional): Number of residual layers in the interaction blocks before the skip connection. (default: 1)

• num_after_skip – (int, optional): Number of residual layers in the interaction blocks after the skip connection. (default: 2)

• num_output_layers – (int, optional): Number of linear layers for the output blocks. (default: 3)

• act – (function, optional): The activation funtion. (default: silu)

url = 'https://github.com/klicperajo/dimenet/raw/master/pretrained'

cutoff

num_blocks

rbf

sbf

emb

output_blocks

interaction_blocks

reset_parameters() → None

triplets(edge_index, cell_offsets, num_nodes: int)

abstract forward(z, pos, batch=None)

class core.models.dimenet_plus_plus.DimeNetPlusPlusWrapEnergyAndForceHead(backbone)

Bases: torch.nn.Module, fairchem.core.models.base.HeadInterface

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

regress_forces

forward(data: torch_geometric.data.batch.Batch, emb: dict[str, torch.Tensor]) → dict[str, torch.Tensor]

Head forward.

Parameters:

• data (DataBatch) – Atomic systems as input

• emb (dict[str->torch.Tensor]) – Embeddings of the input as generated by the backbone

Returns:

outputs – Return one or more targets generated by this head

Return type:

dict[str->torch.Tensor]

class core.models.dimenet_plus_plus.DimeNetPlusPlusWrap(use_pbc: bool = True, use_pbc_single: bool = False, regress_forces: bool = True, hidden_channels: int = 128, num_blocks: int = 4, int_emb_size: int = 64, basis_emb_size: int = 8, out_emb_channels: int = 256, num_spherical: int = 7, num_radial: int = 6, otf_graph: bool = False, cutoff: float = 10.0, envelope_exponent: int = 5, num_before_skip: int = 1, num_after_skip: int = 2, num_output_layers: int = 3)

Bases: DimeNetPlusPlus, fairchem.core.models.base.GraphModelMixin

DimeNet++ implementation based on klicperajo/dimenet.

Parameters:

• hidden_channels (int) – Hidden embedding size.

• out_channels (int) – Size of each output sample.

• num_blocks (int) – Number of building blocks.

• int_emb_size (int) – Embedding size used for interaction triplets

• basis_emb_size (int) – Embedding size used in the basis transformation

• out_emb_channels (int) – Embedding size used for atoms in the output block

• num_spherical (int) – Number of spherical harmonics.

• num_radial (int) – Number of radial basis functions.

• cutoff – (float, optional): Cutoff distance for interatomic interactions. (default: 5.0)

• envelope_exponent (int, optional) – Shape of the smooth cutoff. (default: 5)

• num_before_skip – (int, optional): Number of residual layers in the interaction blocks before the skip connection. (default: 1)

• num_after_skip – (int, optional): Number of residual layers in the interaction blocks after the skip connection. (default: 2)

• num_output_layers – (int, optional): Number of linear layers for the output blocks. (default: 3)

• act – (function, optional): The activation funtion. (default: silu)

regress_forces

use_pbc

use_pbc_single

cutoff

otf_graph

max_neighbors = 50

_forward(data)

forward(data)

property num_params: int

class core.models.dimenet_plus_plus.DimeNetPlusPlusWrapBackbone(use_pbc: bool = True, use_pbc_single: bool = False, regress_forces: bool = True, hidden_channels: int = 128, num_blocks: int = 4, int_emb_size: int = 64, basis_emb_size: int = 8, out_emb_channels: int = 256, num_spherical: int = 7, num_radial: int = 6, otf_graph: bool = False, cutoff: float = 10.0, envelope_exponent: int = 5, num_before_skip: int = 1, num_after_skip: int = 2, num_output_layers: int = 3)

Bases: DimeNetPlusPlusWrap, fairchem.core.models.base.BackboneInterface

DimeNet++ implementation based on klicperajo/dimenet.

Parameters:

• hidden_channels (int) – Hidden embedding size.

• out_channels (int) – Size of each output sample.

• num_blocks (int) – Number of building blocks.

• int_emb_size (int) – Embedding size used for interaction triplets

• basis_emb_size (int) – Embedding size used in the basis transformation

• out_emb_channels (int) – Embedding size used for atoms in the output block

• num_spherical (int) – Number of spherical harmonics.

• num_radial (int) – Number of radial basis functions.

• cutoff – (float, optional): Cutoff distance for interatomic interactions. (default: 5.0)

• envelope_exponent (int, optional) – Shape of the smooth cutoff. (default: 5)

• num_before_skip – (int, optional): Number of residual layers in the interaction blocks before the skip connection. (default: 1)

• num_after_skip – (int, optional): Number of residual layers in the interaction blocks after the skip connection. (default: 2)

• num_output_layers – (int, optional): Number of linear layers for the output blocks. (default: 3)

• act – (function, optional): The activation funtion. (default: silu)

forward(data: torch_geometric.data.batch.Batch) → dict[str, torch.Tensor]

Backbone forward.

Parameters:

data (DataBatch) – Atomic systems as input

Returns:

embedding – Return backbone embeddings for the given input

Return type:

dict[str->torch.Tensor]