diff --git a/README.md b/README.md index 07500e7..95a8ebd 100644 --- a/README.md +++ b/README.md @@ -3,20 +3,20 @@ *Geometric GNN Dojo* is a pedagogical resource for beginners and experts to explore the design space of **Graph Neural Networks for geometric graphs**. Check out the accompanying paper ['On the Expressive Power of Geometric Graph Neural Networks'](https://arxiv.org/abs/2301.09308), which studies the expressivity and theoretical limits of geometric GNNs. -> Chaitanya K. Joshi*, Cristian Bodnar*, Simon V. Mathis, Taco Cohen, and Pietro LiΓ². On the Expressive Power of Geometric Graph Neural Networks. *NeurIPS 2022 Workshop on Symmetry and Geometry in Neural Representations.* +> Chaitanya K. Joshi*, Cristian Bodnar*, Simon V. Mathis, Taco Cohen, and Pietro LiΓ². On the Expressive Power of Geometric Graph Neural Networks. *International Conference on Machine Learning*. > >[PDF](https://arxiv.org/pdf/2301.09308.pdf) | [Slides](https://www.chaitjo.com/publication/joshi-2023-expressive/Geometric_GNNs_Slides.pdf) | [Video](https://youtu.be/5ulJMtpiKGc) ❓**New to geometric GNNs:** try our practical notebook on [*Geometric GNNs 101*](geometric_gnn_101.ipynb), prepared for MPhil students at the University of Cambridge. - Open In Colab + Open In Colab (recommended!) ## Architectures -The `/src` directory provides unified implementations of several popular geometric GNN architectures: -- Invariant GNNs: [SchNet](https://arxiv.org/abs/1706.08566), [DimeNet](https://arxiv.org/abs/2003.03123) +The `/models` directory provides unified implementations of several popular geometric GNN architectures: +- Invariant GNNs: [SchNet](https://arxiv.org/abs/1706.08566), [DimeNet](https://arxiv.org/abs/2003.03123), [SphereNet](https://arxiv.org/abs/2102.05013) - Equivariant GNNs using cartesian vectors: [E(n) Equivariant GNN](https://proceedings.mlr.press/v139/satorras21a.html), [GVP-GNN](https://arxiv.org/abs/2009.01411) - Equivariant GNNs using spherical tensors: [Tensor Field Network](https://arxiv.org/abs/1802.08219), [MACE](http://arxiv.org/abs/2206.07697) - πŸ”₯ Your new geometric GNN architecture? @@ -76,17 +76,23 @@ pip install torch-geometric β”œβ”€β”€ geometric_gnn_101.ipynb # A gentle introduction to Geometric GNNs | β”œβ”€β”€ experiments # Synthetic experiments -β”‚ β”œβ”€β”€ incompleteness.ipynb # Experiment on counterexamples from Pozdnyakov et al. +| | β”‚ β”œβ”€β”€ kchains.ipynb # Experiment on k-chains -β”‚ └── rotsym.ipynb # Experiment on rotationally symmetric structures +β”‚ β”œβ”€β”€ rotsym.ipynb # Experiment on rotationally symmetric structures +β”‚ β”œβ”€β”€ incompleteness.ipynb # Experiment on counterexamples from Pozdnyakov et al. +| └── utils # Helper functions for training, plotting, etc. | -└── src # Geometric GNN models library - β”œβ”€β”€ models.py # Models built using layers - β”œβ”€β”€ gvp_layers.py # Layers for GVP-GNN - β”œβ”€β”€ egnn_layers.py # Layers for E(n) Equivariant GNN - β”œβ”€β”€ tfn_layers.py # Layers for Tensor Field Networks - β”œβ”€β”€ modules # Layers for MACE - └── utils # Helper functions for training, plotting, etc. +└── models # Geometric GNN models library + | + β”œβ”€β”€ schnet.py # SchNet model + β”œβ”€β”€ dimenet.py # DimeNet model + β”œβ”€β”€ spherenet.py # SphereNet model + β”œβ”€β”€ egnn.py # E(n) Equivariant GNN model + β”œβ”€β”€ gvpgnn.py # GVP-GNN model + β”œβ”€β”€ tfn.py # Tensor Field Network model + β”œβ”€β”€ mace.py # MACE model + β”œβ”€β”€ layers # Layers for each model + └── modules # Modules and layers for MACE ``` @@ -99,10 +105,10 @@ We welcome your questions and feedback via email or GitHub Issues. ## Citation ``` -@article{joshi2022expressive, +@inproceedings{joshi2023expressive, title={On the Expressive Power of Geometric Graph Neural Networks}, author={Joshi, Chaitanya K. and Bodnar, Cristian and Mathis, Simon V. and Cohen, Taco and LiΓ², Pietro}, - journal={NeurIPS Workshop on Symmetry and Geometry in Neural Representations}, - year={2022}, + booktitle={International Conference on Machine Learning}, + year={2023}, } -``` +``` \ No newline at end of file diff --git a/experiments/fig/axes-of-expressivity.png b/experiments/fig/axes-of-expressivity.png index 1c074f1..34ac868 100644 Binary files a/experiments/fig/axes-of-expressivity.png and b/experiments/fig/axes-of-expressivity.png differ diff --git a/experiments/fig/incompleteness.png b/experiments/fig/incompleteness.png index b9f21d6..ff2b192 100644 Binary files a/experiments/fig/incompleteness.png and b/experiments/fig/incompleteness.png differ diff --git a/experiments/incompleteness.ipynb b/experiments/incompleteness.ipynb index d293ce7..a079746 100644 --- a/experiments/incompleteness.ipynb +++ b/experiments/incompleteness.ipynb @@ -8,13 +8,14 @@ "# Identifying neighbourhood fingerprints: counterexamples from [Pozdnyakov et al., 2020](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.125.166001)\n", "\n", "*Background:*\n", - "Geometric GNNs identify local neighbourhoods around nodes via **'neighbourhood finderprints'** or scalarisations, where local geometric information from subsets of neighbours is aggregated to compute invariant scalars. The number of neighbours involved in computing the scalars is termed the **body order**.\n", + "Geometric GNNs identify local neighbourhoods around nodes via **'neighbourhood finderprints'**, where local geometric information from subsets of neighbours is aggregated to compute invariant scalars. \n", + "The number of neighbours involved in computing the scalars is termed the **body order**.\n", "The ideal neighbourhood fingerprint would perfectly identify neighbourhoods, which requires arbitrarily high body order.\n", "\n", "*Experiment:*\n", "To demonstrate the practical implications of scalarisation body order, we evaluate geometric GNN layers on their ability to discriminate counterexamples from [Pozdnyakov et al., 2020](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.125.166001).\n", "Each counterexample consists of a pair of local neighbourhoods that are **indistinguishable** when comparing their set of $k$-body scalars, i.e. geometric GNN layers with body order $k$ cannot distinguish the neighbourhoods.\n", - "The 3-body counterexample corresponds to Fig.1(b) in Pozdnyakov et al., 2020, 4-body chiral to Fig.2(e), and 4-body non-chiral to Fig.2(f); the 2-body counterexample is based on the two local neighbourhoods in our running example.\n", + "The 3-body counterexample corresponds to Fig.1(b) in Pozdnyakov et al., 2020, 4-body chiral to Fig.2(e), and 4-body non-chiral to Fig.2(f); the 2-body counterexample is based on the two local neighbourhoods in the running example from our paper.\n", "In this notebook, we train single layer geometric GNNs to distinguish the counterexamples using updated scalar features. \n", "\n", "![Counterexamples from Pozdnyakov et al., 2020](fig/incompleteness.png)" @@ -22,22 +23,9 @@ }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "The autoreload extension is already loaded. To reload it, use:\n", - " %reload_ext autoreload\n", - "PyTorch version 1.12.1\n", - "PyG version 2.1.0\n", - "e3nn version 0.4.4\n", - "Using device: cpu\n" - ] - } - ], + "outputs": [], "source": [ "%load_ext autoreload\n", "%autoreload 2\n", @@ -45,34 +33,24 @@ "import sys\n", "sys.path.append('../')\n", "\n", - "import random\n", - "import numpy as np\n", "import torch\n", - "from torch.nn import functional as F\n", "import torch_geometric\n", - "from torch_geometric.data import Data, Batch\n", + "from torch_geometric.data import Data\n", "from torch_geometric.loader import DataLoader\n", - "from torch_geometric.utils import is_undirected, to_undirected, remove_self_loops, to_dense_adj, dense_to_sparse\n", + "from torch_geometric.utils import to_undirected\n", "import e3nn\n", - "from e3nn import o3\n", "from functools import partial\n", "\n", "print(\"PyTorch version {}\".format(torch.__version__))\n", "print(\"PyG version {}\".format(torch_geometric.__version__))\n", "print(\"e3nn version {}\".format(e3nn.__version__))\n", "\n", - "from src.utils.plot_utils import plot_2d, plot_3d\n", - "from src.utils.train_utils import run_experiment\n", - "from src.models import MPNNModel, EGNNModel, GVPGNNModel, TFNModel, SchNetModel, DimeNetPPModel, MACEModel\n", - "\n", - "# Check PyTorch has access to MPS (Metal Performance Shader, Apple's GPU architecture)\n", - "# print(f\"Is MPS (Metal Performance Shader) built? {torch.backends.mps.is_built()}\")\n", - "# print(f\"Is MPS available? {torch.backends.mps.is_available()}\")\n", + "from experiments.utils.plot_utils import plot_3d\n", + "from experiments.utils.train_utils import run_experiment\n", + "from models import SchNetModel, DimeNetPPModel, SphereNetModel, EGNNModel, GVPGNNModel, TFNModel, MACEModel\n", "\n", "# Set the device\n", "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n", - "# device = torch.device(\"mps\" if torch.backends.mps.is_available() else \"cpu\")\n", - "# device = torch.device(\"cpu\")\n", "print(f\"Using device: {device}\")" ] }, @@ -88,7 +66,7 @@ }, { "cell_type": "code", - "execution_count": 3, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -96,7 +74,6 @@ " dataset = []\n", "\n", " # Environment 0\n", - " # atoms = torch.LongTensor([ 0, 1, 2 ])\n", " atoms = torch.LongTensor([ 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0], [1, 2] ])\n", " pos = torch.FloatTensor([ \n", @@ -110,7 +87,6 @@ " dataset.append(data1)\n", " \n", " # Environment 1\n", - " # atoms = torch.LongTensor([ 0, 1, 2 ])\n", " atoms = torch.LongTensor([ 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0], [1, 2] ])\n", " pos = torch.FloatTensor([ \n", @@ -123,93 +99,38 @@ " data2.edge_index = to_undirected(data2.edge_index)\n", " dataset.append(data2)\n", " \n", - " return dataset" - ] - }, - { - "cell_type": "code", - "execution_count": 5, - "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Running experiment for MACEModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [00:31<00:00, 3.15s/it]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 3.15s Β± 0.13. \n", - " - Best validation accuracy: 100.000 Β± 0.000. \n", - "- Test accuracy: 100.0 Β± 0.0. \n", - "\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - } - ], - "source": [ + " return dataset\n", + "\n", "# Create dataset\n", "dataset = create_two_body_envs()\n", "for data in dataset:\n", " plot_3d(data, lim=5)\n", "\n", - "# Set model\n", - "model_name = \"mace\"\n", - "\n", "# Create dataloaders\n", "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", "val_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", - "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", + "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# Set model\n", + "model_name = \"mace\"\n", "\n", - "num_layers = 1\n", "correlation = 2\n", "model = {\n", - " \"mpnn\": MPNNModel,\n", " \"schnet\": SchNetModel,\n", " \"dimenet\": DimeNetPPModel,\n", + " \"spherenet\": SphereNetModel,\n", " \"egnn\": EGNNModel,\n", " \"gvp\": GVPGNNModel,\n", " \"tfn\": TFNModel,\n", " \"mace\": partial(MACEModel, correlation=correlation),\n", - "}[model_name](num_layers=num_layers, in_dim=1, out_dim=2)\n", + "}[model_name](num_layers=1, in_dim=1, out_dim=2)\n", "\n", "best_val_acc, test_acc, train_time = run_experiment(\n", " model, \n", @@ -235,7 +156,7 @@ }, { "cell_type": "code", - "execution_count": 6, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -247,7 +168,6 @@ " c_x, c_y, c_z = 0, 5, 5\n", " \n", " # Environment 0\n", - " # atoms = torch.LongTensor([ 0, 1, 2, 3, 4 ])\n", " atoms = torch.LongTensor([ 0, 0, 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0, 0, 0], [1, 2, 3, 4] ])\n", " pos = torch.FloatTensor([ \n", @@ -263,7 +183,6 @@ " dataset.append(data1)\n", " \n", " # Environment 1\n", - " # atoms = torch.LongTensor([ 0, 1, 2, 3, 4 ])\n", " atoms = torch.LongTensor([ 0, 0, 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0, 0, 0], [1, 2, 3, 4] ])\n", " pos = torch.FloatTensor([ \n", @@ -278,93 +197,38 @@ " data2.edge_index = to_undirected(data2.edge_index)\n", " dataset.append(data2)\n", " \n", - " return dataset" - ] - }, - { - "cell_type": "code", - "execution_count": 7, - "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Running experiment for MACEModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [00:54<00:00, 5.45s/it]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 5.45s Β± 0.31. \n", - " - Best validation accuracy: 100.000 Β± 0.000. \n", - "- Test accuracy: 100.0 Β± 0.0. \n", - "\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - } - ], - "source": [ + " return dataset\n", + "\n", "# Create dataset\n", "dataset = create_three_body_envs()\n", "for data in dataset:\n", " plot_3d(data, lim=5)\n", "\n", - "# Set model\n", - "model_name = \"mace\"\n", - "\n", "# Create dataloaders\n", "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", "val_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", - "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", + "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# Set model\n", + "model_name = \"mace\"\n", "\n", - "num_layers = 1\n", "correlation = 3\n", "model = {\n", - " \"mpnn\": MPNNModel,\n", " \"schnet\": SchNetModel,\n", " \"dimenet\": DimeNetPPModel,\n", + " \"spherenet\": SphereNetModel,\n", " \"egnn\": EGNNModel,\n", " \"gvp\": GVPGNNModel,\n", " \"tfn\": TFNModel,\n", " \"mace\": partial(MACEModel, correlation=correlation),\n", - "}[model_name](num_layers=num_layers, in_dim=1, out_dim=2)\n", + "}[model_name](num_layers=1, in_dim=1, out_dim=2)\n", "\n", "best_val_acc, test_acc, train_time = run_experiment(\n", " model, \n", @@ -390,7 +254,7 @@ }, { "cell_type": "code", - "execution_count": 8, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -406,10 +270,9 @@ " c_x, c_y, c_z = 0, 5, 0\n", "\n", " angle = 2 * torch.pi / 10 # random angle\n", - " Q = o3.matrix_y(torch.tensor(angle)).numpy()\n", + " Q = e3nn.o3.matrix_y(torch.tensor(angle)).numpy()\n", "\n", " # Environment 0\n", - " # atoms = torch.LongTensor([ 0, 1, 1, 1, 1, 1, 1, 2 ])\n", " atoms = torch.LongTensor([ 0, 0, 0, 0, 0, 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0, 0, 0, 0, 0, 0], [1, 2, 3, 4, 5, 6, 7] ])\n", " pos = torch.FloatTensor([ \n", @@ -428,7 +291,6 @@ " dataset.append(data1)\n", " \n", " # Environment 1\n", - " # atoms = torch.LongTensor([ 0, 1, 1, 1, 1, 1, 1, 2 ])\n", " atoms = torch.LongTensor([ 0, 0, 0, 0, 0, 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0, 0, 0, 0, 0, 0], [1, 2, 3, 4, 5, 6, 7] ])\n", " pos = torch.FloatTensor([ \n", @@ -446,93 +308,38 @@ " data2.edge_index = to_undirected(data2.edge_index)\n", " dataset.append(data2)\n", " \n", - " return dataset" - ] - }, - { - "cell_type": "code", - "execution_count": 11, - "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Running experiment for MACEModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [02:49<00:00, 17.00s/it]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 16.99s Β± 0.36. \n", - " - Best validation accuracy: 50.000 Β± 0.000. \n", - "- Test accuracy: 50.0 Β± 0.0. \n", - "\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - } - ], - "source": [ + " return dataset\n", + "\n", "# Create dataset\n", "dataset = create_four_body_nonchiral_envs()\n", "for data in dataset:\n", " plot_3d(data, lim=5)\n", "\n", - "# Set model\n", - "model_name = \"mace\"\n", - "\n", "# Create dataloaders\n", "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", "val_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", - "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", + "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# Set model\n", + "model_name = \"mace\"\n", "\n", - "num_layers = 1\n", "correlation = 4\n", "model = {\n", - " \"mpnn\": MPNNModel,\n", " \"schnet\": SchNetModel,\n", " \"dimenet\": DimeNetPPModel,\n", + " \"spherenet\": SphereNetModel,\n", " \"egnn\": EGNNModel,\n", " \"gvp\": GVPGNNModel,\n", " \"tfn\": TFNModel,\n", " \"mace\": partial(MACEModel, correlation=correlation),\n", - "}[model_name](num_layers=num_layers, in_dim=1, out_dim=2)\n", + "}[model_name](num_layers=1, in_dim=1, out_dim=2)\n", "\n", "best_val_acc, test_acc, train_time = run_experiment(\n", " model, \n", @@ -558,7 +365,7 @@ }, { "cell_type": "code", - "execution_count": 12, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -571,7 +378,6 @@ " c_x, c_y, c_z = 0, 5, 0\n", "\n", " # Environment 0\n", - " # atoms = torch.LongTensor([ 0, 1, 1, 1, 2 ])\n", " atoms = torch.LongTensor([ 0, 0, 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0, 0, 0], [1, 2, 3, 4] ])\n", " pos = torch.FloatTensor([ \n", @@ -587,7 +393,6 @@ " dataset.append(data1)\n", " \n", " # Environment 1\n", - " # atoms = torch.LongTensor([ 0, 1, 1, 1, 2 ])\n", " atoms = torch.LongTensor([ 0, 0, 0, 0, 0 ])\n", " edge_index = torch.LongTensor([ [0, 0, 0, 0], [1, 2, 3, 4] ])\n", " pos = torch.FloatTensor([ \n", @@ -602,93 +407,38 @@ " data2.edge_index = to_undirected(data2.edge_index)\n", " dataset.append(data2)\n", " \n", - " return dataset" - ] - }, - { - "cell_type": "code", - "execution_count": 13, - "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Running experiment for MACEModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [02:04<00:00, 12.41s/it]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 12.41s Β± 0.38. \n", - " - Best validation accuracy: 50.000 Β± 0.000. \n", - "- Test accuracy: 50.0 Β± 0.0. \n", - "\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - } - ], - "source": [ + " return dataset\n", + "\n", "# Create dataset\n", "dataset = create_four_body_chiral_envs()\n", "for data in dataset:\n", " plot_3d(data, lim=5)\n", "\n", - "# Set model\n", - "model_name = \"mace\"\n", - "\n", "# Create dataloaders\n", "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", "val_loader = DataLoader(dataset, batch_size=2, shuffle=False)\n", - "test_loader = DataLoader(dataset, batch_size=2, shuffle=False)\n", + "test_loader = DataLoader(dataset, batch_size=2, shuffle=False)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# Set model\n", + "model_name = \"tfn\"\n", "\n", - "num_layers = 1\n", "correlation = 4\n", "model = {\n", - " \"mpnn\": MPNNModel,\n", " \"schnet\": SchNetModel,\n", " \"dimenet\": DimeNetPPModel,\n", + " \"spherenet\": SphereNetModel,\n", " \"egnn\": EGNNModel,\n", " \"gvp\": GVPGNNModel,\n", - " \"tfn\": TFNModel,\n", - " \"mace\": partial(MACEModel, correlation=correlation),\n", - "}[model_name](num_layers=num_layers, in_dim=1, out_dim=2)\n", + " \"tfn\": partial(TFNModel, hidden_irreps=e3nn.o3.Irreps(f'64x0e + 64x0o + 64x1e + 64x1o + 64x2e + 64x2o')),\n", + " \"mace\": partial(MACEModel, correlation=correlation, hidden_irreps=e3nn.o3.Irreps(f'32x0e + 32x0o + 32x1e + 32x1o + 32x2e + 32x2o')),\n", + "}[model_name](num_layers=1, in_dim=1, out_dim=2)\n", "\n", "best_val_acc, test_acc, train_time = run_experiment(\n", " model, \n", @@ -719,7 +469,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.8.13" + "version": "3.8.16" }, "orig_nbformat": 4, "vscode": { diff --git a/experiments/kchains.ipynb b/experiments/kchains.ipynb index 0ec9332..da7ba0e 100644 --- a/experiments/kchains.ipynb +++ b/experiments/kchains.ipynb @@ -25,28 +25,15 @@ "- Notably, as the length of the chain gets larger than $k=4$, all equivariant GNNs tended to lose performance and required more than $(\\lfloor \\frac{k}{2} \\rfloor + 1)$ iterations to solve the task.\n", "- Invariant GNNs are **unable** to distinguish $k$-chains.\n", "\n", - "These results points to preliminary evidence of the **oversquashing** phenomenon for geometric GNNs.\n", + "These results point to preliminary evidence of the **oversquashing** phenomenon when geometric information is propagated across multiple layers using fixed dimensional feature spaces.\n", "These issues are most evident for E-GNN, which uses a single vector feature to aggregate and propagate geometric information." ] }, { "cell_type": "code", - "execution_count": 5, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "The autoreload extension is already loaded. To reload it, use:\n", - " %reload_ext autoreload\n", - "PyTorch version 1.12.1\n", - "PyG version 2.1.0\n", - "e3nn version 0.4.4\n", - "Using device: cpu\n" - ] - } - ], + "outputs": [], "source": [ "%load_ext autoreload\n", "%autoreload 2\n", @@ -54,40 +41,30 @@ "import sys\n", "sys.path.append('../')\n", "\n", - "import random\n", - "import numpy as np\n", "import torch\n", - "from torch.nn import functional as F\n", "import torch_geometric\n", - "from torch_geometric.data import Data, Batch\n", + "from torch_geometric.data import Data\n", "from torch_geometric.loader import DataLoader\n", - "from torch_geometric.utils import is_undirected, to_undirected, remove_self_loops, to_dense_adj, dense_to_sparse\n", + "from torch_geometric.utils import to_undirected\n", "import e3nn\n", - "from e3nn import o3\n", "from functools import partial\n", "\n", "print(\"PyTorch version {}\".format(torch.__version__))\n", "print(\"PyG version {}\".format(torch_geometric.__version__))\n", "print(\"e3nn version {}\".format(e3nn.__version__))\n", "\n", - "from src.utils.plot_utils import plot_2d, plot_3d\n", - "from src.utils.train_utils import run_experiment\n", - "from src.models import MPNNModel, EGNNModel, GVPGNNModel, TFNModel, SchNetModel, DimeNetPPModel, MACEModel\n", - "\n", - "# Check PyTorch has access to MPS (Metal Performance Shader, Apple's GPU architecture)\n", - "# print(f\"Is MPS (Metal Performance Shader) built? {torch.backends.mps.is_built()}\")\n", - "# print(f\"Is MPS available? {torch.backends.mps.is_available()}\")\n", + "from experiments.utils.plot_utils import plot_3d\n", + "from experiments.utils.train_utils import run_experiment\n", + "from models import SchNetModel, DimeNetPPModel, SphereNetModel, EGNNModel, GVPGNNModel, TFNModel, MACEModel\n", "\n", "# Set the device\n", "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n", - "# device = torch.device(\"mps\" if torch.backends.mps.is_available() else \"cpu\")\n", - "# device = torch.device(\"cpu\")\n", "print(f\"Using device: {device}\")" ] }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -131,190 +108,45 @@ }, { "cell_type": "code", - "execution_count": 3, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], + "outputs": [], "source": [ "k = 4\n", "\n", "# Create dataset\n", "dataset = create_kchains(k=k)\n", "for data in dataset:\n", - " # plot_2d(data, lim=5*k)\n", - " plot_3d(data, lim=5*k)" + " plot_3d(data, lim=5*k)\n", + "\n", + "# Create dataloaders\n", + "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", + "val_loader = DataLoader(dataset, batch_size=2, shuffle=False)\n", + "test_loader = DataLoader(dataset, batch_size=2, shuffle=False)" ] }, { "cell_type": "code", - "execution_count": 4, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Number of layers: 2\n", - "Running experiment for GVPGNNModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [00:34<00:00, 3.43s/it]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 3.43s Β± 0.39. \n", - " - Best validation accuracy: 50.000 Β± 0.000. \n", - "- Test accuracy: 50.0 Β± 0.0. \n", - "\n", - "\n", - "Number of layers: 3\n", - "Running experiment for GVPGNNModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [01:17<00:00, 7.72s/it]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 7.71s Β± 2.04. \n", - " - Best validation accuracy: 100.000 Β± 0.000. \n", - "- Test accuracy: 100.0 Β± 0.0. \n", - "\n", - "\n", - "Number of layers: 4\n", - "Running experiment for GVPGNNModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [01:13<00:00, 7.36s/it]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 7.35s Β± 0.80. \n", - " - Best validation accuracy: 100.000 Β± 0.000. \n", - "- Test accuracy: 100.0 Β± 0.0. \n", - "\n", - "\n", - "Number of layers: 5\n", - "Running experiment for GVPGNNModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [01:16<00:00, 7.68s/it]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 7.67s Β± 0.54. \n", - " - Best validation accuracy: 100.000 Β± 0.000. \n", - "- Test accuracy: 100.0 Β± 0.0. \n", - "\n", - "\n", - "Number of layers: 6\n", - "Running experiment for GVPGNNModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [01:24<00:00, 8.40s/it]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 8.39s Β± 0.15. \n", - " - Best validation accuracy: 100.000 Β± 0.000. \n", - "- Test accuracy: 100.0 Β± 0.0. \n", - "\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - } - ], + "outputs": [], "source": [ "# Set model\n", "model_name = \"gvp\"\n", "\n", - "# Create dataloaders\n", - "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", - "val_loader = DataLoader(dataset, batch_size=2, shuffle=False)\n", - "test_loader = DataLoader(dataset, batch_size=2, shuffle=False)\n", - "\n", "for num_layers in range(k // 2 , k + 3):\n", "\n", " print(f\"\\nNumber of layers: {num_layers}\")\n", " \n", + " correlation = 2\n", " model = {\n", - " \"mpnn\": MPNNModel,\n", " \"schnet\": SchNetModel,\n", " \"dimenet\": DimeNetPPModel,\n", + " \"spherenet\": SphereNetModel,\n", " \"egnn\": EGNNModel,\n", - " \"gvp\": GVPGNNModel,\n", + " \"gvp\": partial(GVPGNNModel, s_dim=32, v_dim=1),\n", " \"tfn\": TFNModel,\n", - " \"mace\": partial(MACEModel, correlation=2),\n", + " \"mace\": partial(MACEModel, correlation=correlation),\n", " }[model_name](num_layers=num_layers, in_dim=1, out_dim=2)\n", " \n", " best_val_acc, test_acc, train_time = run_experiment(\n", @@ -346,7 +178,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.8.13" + "version": "3.8.16" }, "orig_nbformat": 4, "vscode": { diff --git a/experiments/rotsym.ipynb b/experiments/rotsym.ipynb index 172448e..b906444 100644 --- a/experiments/rotsym.ipynb +++ b/experiments/rotsym.ipynb @@ -21,28 +21,15 @@ "![Rotationally symmetric structures](fig/rotsym.png)\n", "\n", "*Result:*\n", - "- **We find that layers using order $L$ tensors are unable to identify the orientation of structures with rotation symmetry higher than $L$-fold.** This observation may be attributed to **spherical harmonics**, which are used as the underlying orthonormal basis and are rotationally symmetric themselves.\n", + "- **We find that layers using order $L$ tensors are unable to identify the orientation of structures with rotation symmetry higher than $L$-fold.** This observation may be attributed to **spherical harmonics**, which serve as an orthonormal basis for spherical tensor features and exhibit rotational symmetry themselves.\n", "- Layers such as E-GNN and GVP-GNN using **cartesian vectors** (corresponding to tensor order 1) are popular as working with higher order tensors can be computationally intractable for many applications. However, E-GNN and GVP-GNN are particularly poor at disciminating orientation of rotationally symmetric structures. " ] }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "The autoreload extension is already loaded. To reload it, use:\n", - " %reload_ext autoreload\n", - "PyTorch version 1.12.1\n", - "PyG version 2.1.0\n", - "e3nn version 0.4.4\n", - "Using device: cpu\n" - ] - } - ], + "outputs": [], "source": [ "%load_ext autoreload\n", "%autoreload 2\n", @@ -52,39 +39,30 @@ "\n", "import random\n", "import math\n", - "import numpy as np\n", "import torch\n", - "from torch.nn import functional as F\n", "import torch_geometric\n", - "from torch_geometric.data import Data, Batch\n", + "from torch_geometric.data import Data\n", "from torch_geometric.loader import DataLoader\n", - "from torch_geometric.utils import is_undirected, to_undirected, remove_self_loops, to_dense_adj, dense_to_sparse\n", + "from torch_geometric.utils import to_undirected\n", "import e3nn\n", - "from e3nn import o3\n", "from functools import partial\n", "\n", "print(\"PyTorch version {}\".format(torch.__version__))\n", "print(\"PyG version {}\".format(torch_geometric.__version__))\n", "print(\"e3nn version {}\".format(e3nn.__version__))\n", "\n", - "from src.utils.plot_utils import plot_2d, plot_3d\n", - "from src.utils.train_utils import run_experiment\n", - "from src.models import MPNNModel, EGNNModel, GVPGNNModel, TFNModel, SchNetModel, DimeNetPPModel, MACEModel\n", - "\n", - "# Check PyTorch has access to MPS (Metal Performance Shader, Apple's GPU architecture)\n", - "# print(f\"Is MPS (Metal Performance Shader) built? {torch.backends.mps.is_built()}\")\n", - "# print(f\"Is MPS available? {torch.backends.mps.is_available()}\")\n", + "from experiments.utils.plot_utils import plot_2d\n", + "from experiments.utils.train_utils import run_experiment\n", + "from models import SchNetModel, DimeNetPPModel, SphereNetModel, EGNNModel, GVPGNNModel, TFNModel, MACEModel\n", "\n", "# Set the device\n", "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n", - "# device = torch.device(\"mps\" if torch.backends.mps.is_available() else \"cpu\")\n", - "# device = torch.device(\"cpu\")\n", "print(f\"Using device: {device}\")" ] }, { "cell_type": "code", - "execution_count": 3, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -100,7 +78,7 @@ " x, # first spoke \n", " ]\n", " for count in range(1, fold):\n", - " R = o3.matrix_z(torch.Tensor([2*math.pi/fold * count])).squeeze(0)\n", + " R = e3nn.o3.matrix_z(torch.Tensor([2*math.pi/fold * count])).squeeze(0)\n", " pos.append(x @ R.T)\n", " pos = torch.stack(pos)\n", " y = torch.LongTensor([0]) # Label 0\n", @@ -111,7 +89,7 @@ " # Environment 1\n", " q = 2*math.pi/(fold + random.randint(1, fold))\n", " assert q < 2*math.pi/fold\n", - " Q = o3.matrix_z(torch.Tensor([q])).squeeze(0)\n", + " Q = e3nn.o3.matrix_z(torch.Tensor([q])).squeeze(0)\n", " pos = pos @ Q.T\n", " y = torch.LongTensor([1]) # Label 1\n", " data2 = Data(atoms=atoms, edge_index=edge_index, pos=pos, y=y)\n", @@ -123,68 +101,10 @@ }, { "cell_type": "code", - "execution_count": 5, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Running experiment for TFNModel (cpu).\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 10/10 [00:38<00:00, 3.86s/it]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "Done! Averaged over 10 runs: \n", - " - Training time: 3.85s Β± 0.12. \n", - " - Best validation accuracy: 50.000 Β± 0.000. \n", - "- Test accuracy: 50.0 Β± 0.0. \n", - "\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - } - ], + "outputs": [], "source": [ - "# Set parameters\n", - "model_name = \"tfn\"\n", - "correlation = 2\n", - "max_ell = 3\n", "fold = 5\n", "\n", "# Create dataset\n", @@ -195,18 +115,29 @@ "# Create dataloaders\n", "dataloader = DataLoader(dataset, batch_size=1, shuffle=True)\n", "val_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", - "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)\n", + "test_loader = DataLoader(dataset, batch_size=1, shuffle=False)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# Set parameters\n", + "model_name = \"tfn\"\n", + "correlation = 2\n", + "max_ell = 5\n", "\n", - "num_layers = 1\n", "model = {\n", - " \"mpnn\": MPNNModel,\n", " \"schnet\": SchNetModel,\n", " \"dimenet\": DimeNetPPModel,\n", - " \"egnn\": EGNNModel,\n", - " \"gvp\": GVPGNNModel,\n", - " \"tfn\": partial(TFNModel, max_ell=max_ell, scalar_pred=False),\n", - " \"mace\": partial(MACEModel, max_ell=max_ell, correlation=correlation, scalar_pred=False),\n", - "}[model_name](num_layers=num_layers, in_dim=1, out_dim=2)\n", + " \"spherenet\": SphereNetModel,\n", + " \"egnn\": partial(EGNNModel, equivariant_pred=True),\n", + " \"gvp\": partial(GVPGNNModel, equivariant_pred=True),\n", + " \"tfn\": partial(TFNModel, max_ell=max_ell, equivariant_pred=True),\n", + " \"mace\": partial(MACEModel, max_ell=max_ell, correlation=correlation, equivariant_pred=True),\n", + "}[model_name](num_layers=1, in_dim=1, out_dim=2)\n", "\n", "best_val_acc, test_acc, train_time = run_experiment(\n", " model, \n", @@ -219,6 +150,13 @@ " verbose=False\n", ")" ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] } ], "metadata": { @@ -237,7 +175,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.8.13" + "version": "3.8.16" }, "orig_nbformat": 4, "vscode": { diff --git a/src/__init__.py b/experiments/utils/__init__.py similarity index 100% rename from src/__init__.py rename to experiments/utils/__init__.py diff --git a/src/utils/plot_utils.py b/experiments/utils/plot_utils.py similarity index 100% rename from src/utils/plot_utils.py rename to experiments/utils/plot_utils.py diff --git a/src/utils/train_utils.py b/experiments/utils/train_utils.py similarity index 98% rename from src/utils/train_utils.py rename to experiments/utils/train_utils.py index 0630b6d..ad480f3 100644 --- a/src/utils/train_utils.py +++ b/experiments/utils/train_utils.py @@ -1,6 +1,6 @@ import time import random -from tqdm import tqdm +from tqdm.autonotebook import tqdm # from tqdm import tqdm import numpy as np from sklearn.metrics import accuracy_score diff --git a/models/__init__.py b/models/__init__.py new file mode 100644 index 0000000..262186e --- /dev/null +++ b/models/__init__.py @@ -0,0 +1,7 @@ +from models.schnet import SchNetModel +from models.dimenet import DimeNetPPModel +from models.spherenet import SphereNetModel +from models.egnn import EGNNModel +from models.gvpgnn import GVPGNNModel +from models.tfn import TFNModel +from models.mace import MACEModel diff --git a/models/dimenet.py b/models/dimenet.py new file mode 100644 index 0000000..b305001 --- /dev/null +++ b/models/dimenet.py @@ -0,0 +1,105 @@ +from typing import Callable, Union + +import torch +from torch.nn import functional as F +from torch_geometric.nn import DimeNetPlusPlus +from torch_scatter import scatter + + +class DimeNetPPModel(DimeNetPlusPlus): + """ + DimeNet model from "Directional message passing for molecular graphs". + + This class extends the DimeNetPlusPlus base class for PyG. + """ + def __init__( + self, + hidden_channels: int = 128, + in_dim: int = 1, + out_dim: int = 1, + num_layers: 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, + cutoff: float = 10, + max_num_neighbors: int = 32, + envelope_exponent: int = 5, + num_before_skip: int = 1, + num_after_skip: int = 2, + num_output_layers: int = 3, + act: Union[str, Callable] = 'swish' + ): + """ + Initializes an instance of the DimeNetPPModel class with the provided parameters. + + Parameters: + - hidden_channels (int): Number of channels in the hidden layers (default: 128) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - num_layers (int): Number of layers in the model (default: 4) + - int_emb_size (int): Embedding size for interaction features (default: 64) + - basis_emb_size (int): Embedding size for basis functions (default: 8) + - out_emb_channels (int): Number of channels in the output embeddings (default: 256) + - num_spherical (int): Number of spherical harmonics (default: 7) + - num_radial (int): Number of radial basis functions (default: 6) + - cutoff (float): Cutoff distance for interactions (default: 10) + - max_num_neighbors (int): Maximum number of neighboring atoms to consider (default: 32) + - envelope_exponent (int): Exponent of the envelope function (default: 5) + - num_before_skip (int): Number of layers before the skip connections (default: 1) + - num_after_skip (int): Number of layers after the skip connections (default: 2) + - num_output_layers (int): Number of output layers (default: 3) + - act (Union[str, Callable]): Activation function (default: 'swish' or callable) + + Note: + - The `act` parameter can be either a string representing a built-in activation function, + or a callable object that serves as a custom activation function. + """ + super().__init__( + hidden_channels, + out_dim, + num_layers, + int_emb_size, + basis_emb_size, + out_emb_channels, + num_spherical, + num_radial, + cutoff, + max_num_neighbors, + envelope_exponent, + num_before_skip, + num_after_skip, + num_output_layers, + act + ) + + def forward(self, batch): + + i, j, idx_i, idx_j, idx_k, idx_kj, idx_ji = self.triplets( + batch.edge_index, num_nodes=batch.atoms.size(0)) + + # Calculate distances. + dist = (batch.pos[i] - batch.pos[j]).pow(2).sum(dim=-1).sqrt() + + # Calculate angles. + pos_i = batch.pos[idx_i] + pos_ji, pos_ki = batch.pos[idx_j] - pos_i, batch.pos[idx_k] - pos_i + a = (pos_ji * pos_ki).sum(dim=-1) + b = torch.cross(pos_ji, pos_ki).norm(dim=-1) + angle = torch.atan2(b, a) + + rbf = self.rbf(dist) + sbf = self.sbf(dist, angle, idx_kj) + + # Embedding block. + x = self.emb(batch.atoms, rbf, i, j) + P = self.output_blocks[0](x, rbf, i, num_nodes=batch.pos.size(0)) + + # Interaction blocks. + for interaction_block, output_block in zip(self.interaction_blocks, + self.output_blocks[1:]): + x = interaction_block(x, rbf, sbf, idx_kj, idx_ji) + P += output_block(x, rbf, i) + + return P.sum(dim=0) if batch is None else scatter(P, batch.batch, dim=0) diff --git a/models/egnn.py b/models/egnn.py new file mode 100644 index 0000000..870ae62 --- /dev/null +++ b/models/egnn.py @@ -0,0 +1,87 @@ +import torch +from torch.nn import functional as F +from torch_geometric.nn import global_add_pool, global_mean_pool + +from models.layers.egnn_layer import EGNNLayer + + +class EGNNModel(torch.nn.Module): + """ + E-GNN model from "E(n) Equivariant Graph Neural Networks". + """ + def __init__( + self, + num_layers: int = 5, + emb_dim: int = 128, + in_dim: int = 1, + out_dim: int = 1, + activation: str = "relu", + norm: str = "layer", + aggr: str = "sum", + pool: str = "sum", + residual: bool = True, + equivariant_pred: bool = False + ): + """ + Initializes an instance of the EGNNModel class with the provided parameters. + + Parameters: + - num_layers (int): Number of layers in the model (default: 5) + - emb_dim (int): Dimension of the node embeddings (default: 128) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - activation (str): Activation function to be used (default: "relu") + - norm (str): Normalization method to be used (default: "layer") + - aggr (str): Aggregation method to be used (default: "sum") + - pool (str): Global pooling method to be used (default: "sum") + - residual (bool): Whether to use residual connections (default: True) + - equivariant_pred (bool): Whether it is an equivariant prediction task (default: False) + """ + super().__init__() + self.equivariant_pred = equivariant_pred + self.residual = residual + + # Embedding lookup for initial node features + self.emb_in = torch.nn.Embedding(in_dim, emb_dim) + + # Stack of GNN layers + self.convs = torch.nn.ModuleList() + for _ in range(num_layers): + self.convs.append(EGNNLayer(emb_dim, activation, norm, aggr)) + + # Global pooling/readout function + self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] + + if self.equivariant_pred: + # Linear predictor for equivariant tasks using geometric features + self.pred = torch.nn.Linear(emb_dim + 3, out_dim) + else: + # MLP predictor for invariant tasks using only scalar features + self.pred = torch.nn.Sequential( + torch.nn.Linear(emb_dim, emb_dim), + torch.nn.ReLU(), + torch.nn.Linear(emb_dim, out_dim) + ) + + def forward(self, batch): + + h = self.emb_in(batch.atoms) # (n,) -> (n, d) + pos = batch.pos # (n, 3) + + for conv in self.convs: + # Message passing layer + h_update, pos_update = conv(h, pos, batch.edge_index) + + # Update node features (n, d) -> (n, d) + h = h + h_update if self.residual else h_update + + # Update node coordinates (no residual) (n, 3) -> (n, 3) + pos = pos_update + + if not self.equivariant_pred: + # Select only scalars for invariant prediction + out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) + else: + out = self.pool(torch.cat([h, pos], dim=-1), batch.batch) + + return self.pred(out) # (batch_size, out_dim) diff --git a/models/gvpgnn.py b/models/gvpgnn.py new file mode 100644 index 0000000..96e4e20 --- /dev/null +++ b/models/gvpgnn.py @@ -0,0 +1,127 @@ +import torch +from torch.nn import functional as F +from torch_geometric.nn import global_add_pool, global_mean_pool + +from models.mace_modules.blocks import RadialEmbeddingBlock +import models.layers.gvp_layer as gvp + + +class GVPGNNModel(torch.nn.Module): + """ + GVP-GNN model from "Equivariant Graph Neural Networks for 3D Macromolecular Structure". + """ + def __init__( + self, + r_max: float = 10.0, + num_bessel: int = 8, + num_polynomial_cutoff: int = 5, + num_layers: int = 5, + in_dim=1, + out_dim=1, + s_dim: int = 128, + v_dim: int = 16, + s_dim_edge: int = 32, + v_dim_edge: int = 1, + pool: str = "sum", + residual: bool = True, + equivariant_pred: bool = False + ): + """ + Initializes an instance of the GVPGNNModel class with the provided parameters. + + Parameters: + - r_max (float): Maximum distance for Bessel basis functions (default: 10.0) + - num_bessel (int): Number of Bessel basis functions (default: 8) + - num_polynomial_cutoff (int): Number of polynomial cutoff basis functions (default: 5) + - num_layers (int): Number of layers in the model (default: 5) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - s_dim (int): Dimension of the node state embeddings (default: 128) + - v_dim (int): Dimension of the node vector embeddings (default: 16) + - s_dim_edge (int): Dimension of the edge state embeddings (default: 32) + - v_dim_edge (int): Dimension of the edge vector embeddings (default: 1) + - pool (str): Global pooling method to be used (default: "sum") + - residual (bool): Whether to use residual connections (default: True) + - equivariant_pred (bool): Whether it is an equivariant prediction task (default: False) + """ + super().__init__() + + self.r_max = r_max + self.num_layers = num_layers + self.equivariant_pred = equivariant_pred + self.s_dim = s_dim + self.v_dim = v_dim + + activations = (F.relu, None) + _DEFAULT_V_DIM = (s_dim, v_dim) + _DEFAULT_E_DIM = (s_dim_edge, v_dim_edge) + + # Node embedding + self.emb_in = torch.nn.Embedding(in_dim, s_dim) + self.W_v = torch.nn.Sequential( + gvp.LayerNorm((s_dim, 0)), + gvp.GVP((s_dim, 0), _DEFAULT_V_DIM, + activations=(None, None), vector_gate=True) + ) + + # Edge embedding + self.radial_embedding = RadialEmbeddingBlock( + r_max=r_max, + num_bessel=num_bessel, + num_polynomial_cutoff=num_polynomial_cutoff, + ) + self.W_e = torch.nn.Sequential( + gvp.LayerNorm((self.radial_embedding.out_dim, 1)), + gvp.GVP((self.radial_embedding.out_dim, 1), _DEFAULT_E_DIM, + activations=(None, None), vector_gate=True) + ) + + # Stack of GNN layers + self.layers = torch.nn.ModuleList( + gvp.GVPConvLayer( + _DEFAULT_V_DIM, _DEFAULT_E_DIM, + activations=activations, vector_gate=True, + residual=residual + ) + for _ in range(num_layers) + ) + + # Global pooling/readout function + self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] + + if self.equivariant_pred: + # Linear predictor for equivariant tasks using geometric features + self.pred = torch.nn.Linear(s_dim + v_dim * 3, out_dim) + else: + # MLP predictor for invariant tasks using only scalar features + self.pred = torch.nn.Sequential( + torch.nn.Linear(s_dim, s_dim), + torch.nn.ReLU(), + torch.nn.Linear(s_dim, out_dim) + ) + + def forward(self, batch): + + # Edge features + vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] + lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] + + h_V = self.emb_in(batch.atoms) # (n,) -> (n, d) + h_E = ( + self.radial_embedding(lengths), + torch.nan_to_num(torch.div(vectors, lengths)).unsqueeze_(-2) + ) + + h_V = self.W_v(h_V) + h_E = self.W_e(h_E) + + for layer in self.layers: + h_V = layer(h_V, batch.edge_index, h_E) + + out = self.pool(gvp._merge(*h_V), batch.batch) # (n, d) -> (batch_size, d) + + if not self.equivariant_pred: + # Select only scalars for invariant prediction + out = out[:,:self.s_dim] + + return self.pred(out) # (batch_size, out_dim) \ No newline at end of file diff --git a/src/utils/__init__.py b/models/layers/__init__.py similarity index 100% rename from src/utils/__init__.py rename to models/layers/__init__.py diff --git a/src/egnn_layers.py b/models/layers/egnn_layer.py similarity index 94% rename from src/egnn_layers.py rename to models/layers/egnn_layer.py index e98b051..74356ed 100644 --- a/src/egnn_layers.py +++ b/models/layers/egnn_layer.py @@ -5,11 +5,12 @@ class EGNNLayer(MessagePassing): - def __init__(self, emb_dim, activation="relu", norm="layer", aggr="add"): - """E(n) Equivariant GNN Layer + """E(n) Equivariant GNN Layer - Paper: E(n) Equivariant Graph Neural Networks, Satorras et al. - + Paper: E(n) Equivariant Graph Neural Networks, Satorras et al. + """ + def __init__(self, emb_dim, activation="relu", norm="layer", aggr="add"): + """ Args: emb_dim: (int) - hidden dimension `d` activation: (str) - non-linearity within MLPs (swish/relu) @@ -65,7 +66,9 @@ def message(self, h_i, h_j, pos_i, pos_j): msg = torch.cat([h_i, h_j, dists], dim=-1) msg = self.mlp_msg(msg) # Scale magnitude of displacement vector - pos_diff = pos_diff * self.mlp_pos(msg) # torch.clamp(updates, min=-100, max=100) + pos_diff = pos_diff * self.mlp_pos(msg) + # NOTE: some papers divide pos_diff by (dists + 1) to stabilise model. + # NOTE: lucidrains clamps pos_diff between some [-n, +n], also for stability. return msg, pos_diff def aggregate(self, inputs, index): diff --git a/src/gvp_layers.py b/models/layers/gvp_layer.py similarity index 74% rename from src/gvp_layers.py rename to models/layers/gvp_layer.py index e8f2916..8aefe83 100644 --- a/src/gvp_layers.py +++ b/models/layers/gvp_layer.py @@ -1,70 +1,78 @@ ########################################################################################### -# Implementation of Geometric Vector Perceptron layers +# Implementation of Geometric Vector Perceptron layers # -# Papers: -# (1) Learning from Protein Structure with Geometric Vector Perceptrons, +# Papers: +# (1) Learning from Protein Structure with Geometric Vector Perceptrons, # by B Jing, S Eismann, P Suriana, RJL Townshend, and RO Dror -# (2) Equivariant Graph Neural Networks for 3D Macromolecular Structure, +# (2) Equivariant Graph Neural Networks for 3D Macromolecular Structure, # by B Jing, S Eismann, P Soni, and RO Dror # # Orginal repository: https://github.com/drorlab/gvp-pytorch ########################################################################################### -import torch, functools -from torch import nn +import functools +import torch import torch.nn.functional as F +from torch import nn +import torch_scatter from torch_geometric.nn import MessagePassing -from torch_scatter import scatter_add + def tuple_sum(*args): - ''' + """ Sums any number of tuples (s, V) elementwise. - ''' + """ return tuple(map(sum, zip(*args))) + def tuple_cat(*args, dim=-1): - ''' + """ Concatenates any number of tuples (s, V) elementwise. - + :param dim: dimension along which to concatenate when viewed as the `dim` index for the scalar-channel tensors. This means that `dim=-1` will be applied as `dim=-2` for the vector-channel tensors. - ''' + """ dim %= len(args[0][0].shape) s_args, v_args = list(zip(*args)) return torch.cat(s_args, dim=dim), torch.cat(v_args, dim=dim) + def tuple_index(x, idx): - ''' + """ Indexes into a tuple (s, V) along the first dimension. - + :param idx: any object which can be used to index into a `torch.Tensor` - ''' + """ return x[0][idx], x[1][idx] + def randn(n, dims, device="cpu"): - ''' + """ Returns random tuples (s, V) drawn elementwise from a normal distribution. - + :param n: number of data points :param dims: tuple of dimensions (n_scalar, n_vector) - + :return: (s, V) with s.shape = (n, n_scalar) and V.shape = (n, n_vector, 3) - ''' - return torch.randn(n, dims[0], device=device), \ - torch.randn(n, dims[1], 3, device=device) + """ + return torch.randn(n, dims[0], device=device), torch.randn( + n, dims[1], 3, device=device + ) + def _norm_no_nan(x, axis=-1, keepdims=False, eps=1e-8, sqrt=True): - ''' + """ L2 norm of tensor clamped above a minimum value `eps`. - + :param sqrt: if `False`, returns the square of the L2 norm - ''' + """ out = torch.clamp(torch.sum(torch.square(x), axis, keepdims), min=eps) return torch.sqrt(out) if sqrt else out + def _split(x, nv): ''' Splits a merged representation of (s, V) back into a tuple. @@ -74,10 +82,11 @@ def _split(x, nv): :param x: the `torch.Tensor` returned from `_merge` :param nv: the number of vector channels in the input to `_merge` ''' - v = torch.reshape(x[..., -3*nv:], x.shape[:-1] + (nv, 3)) - s = x[..., :-3*nv] + s = x[..., :-3 * nv] + v = x[..., -3 * nv:].contiguous().view(x.shape[0], nv, 3) return s, v + def _merge(s, v): ''' Merges a tuple (s, V) into a single `torch.Tensor`, where the @@ -85,89 +94,97 @@ def _merge(s, v): Should be used only if the tuple representation cannot be used. Use `_split(x, nv)` to reverse. ''' - v = torch.reshape(v, v.shape[:-2] + (3*v.shape[-2],)) + v = v.contiguous().view(v.shape[0], v.shape[1] * 3) return torch.cat([s, v], -1) + class GVP(nn.Module): - ''' + """ Geometric Vector Perceptron. See manuscript and README.md for more details. - + :param in_dims: tuple (n_scalar, n_vector) :param out_dims: tuple (n_scalar, n_vector) :param h_dim: intermediate number of vector channels, optional :param activations: tuple of functions (scalar_act, vector_act) :param vector_gate: whether to use vector gating. (vector_act will be used as sigma^+ in vector gating if `True`) - ''' - def __init__(self, in_dims, out_dims, h_dim=None, - activations=(F.relu, torch.sigmoid), vector_gate=False): + """ + + def __init__( + self, + in_dims, + out_dims, + h_dim=None, + activations=(F.relu, torch.sigmoid), + vector_gate=True, + ): super(GVP, self).__init__() self.si, self.vi = in_dims self.so, self.vo = out_dims self.vector_gate = vector_gate - if self.vi: - self.h_dim = h_dim or max(self.vi, self.vo) + if self.vi: + self.h_dim = h_dim or max(self.vi, self.vo) self.wh = nn.Linear(self.vi, self.h_dim, bias=False) self.ws = nn.Linear(self.h_dim + self.si, self.so) if self.vo: self.wv = nn.Linear(self.h_dim, self.vo, bias=False) - if self.vector_gate: self.wsv = nn.Linear(self.so, self.vo) + if self.vector_gate: + self.wsv = nn.Linear(self.so, self.vo) else: self.ws = nn.Linear(self.si, self.so) - + self.scalar_act, self.vector_act = activations self.dummy_param = nn.Parameter(torch.empty(0)) - + def forward(self, x): - ''' - :param x: tuple (s, V) of `torch.Tensor`, + """ + :param x: tuple (s, V) of `torch.Tensor`, or (if vectors_in is 0), a single `torch.Tensor` :return: tuple (s, V) of `torch.Tensor`, or (if vectors_out is 0), a single `torch.Tensor` - ''' + """ if self.vi: s, v = x v = torch.transpose(v, -1, -2) - vh = self.wh(v) + vh = self.wh(v) vn = _norm_no_nan(vh, axis=-2) s = self.ws(torch.cat([s, vn], -1)) - if self.vo: - v = self.wv(vh) + if self.vo: + v = self.wv(vh) v = torch.transpose(v, -1, -2) - if self.vector_gate: - if self.vector_act: - gate = self.wsv(self.vector_act(s)) - else: - gate = self.wsv(s) + if self.vector_gate: + gate = ( + self.wsv(self.vector_act(s)) if self.vector_act else self.wsv(s) + ) v = v * torch.sigmoid(gate).unsqueeze(-1) elif self.vector_act: - v = v * self.vector_act( - _norm_no_nan(v, axis=-1, keepdims=True)) + v = v * self.vector_act(_norm_no_nan(v, axis=-1, keepdims=True)) else: s = self.ws(x) if self.vo: - v = torch.zeros(s.shape[0], self.vo, 3, - device=self.dummy_param.device) + v = torch.zeros(s.shape[0], self.vo, 3, device=self.dummy_param.device) if self.scalar_act: s = self.scalar_act(s) - + return (s, v) if self.vo else s + class _VDropout(nn.Module): - ''' + """ Vector channel dropout where the elements of each vector channel are dropped together. - ''' + """ + def __init__(self, drop_rate): super(_VDropout, self).__init__() self.drop_rate = drop_rate self.dummy_param = nn.Parameter(torch.empty(0)) def forward(self, x): - ''' + """ :param x: `torch.Tensor` corresponding to vector channels - ''' + """ device = self.dummy_param.device if not self.training: return x @@ -177,43 +194,47 @@ def forward(self, x): x = mask * x / (1 - self.drop_rate) return x + class Dropout(nn.Module): - ''' + """ Combined dropout for tuples (s, V). Takes tuples (s, V) as input and as output. - ''' + """ + def __init__(self, drop_rate): super(Dropout, self).__init__() self.sdropout = nn.Dropout(drop_rate) self.vdropout = _VDropout(drop_rate) def forward(self, x): - ''' + """ :param x: tuple (s, V) of `torch.Tensor`, - or single `torch.Tensor` + or single `torch.Tensor` (will be assumed to be scalar channels) - ''' + """ if type(x) is torch.Tensor: return self.sdropout(x) s, v = x return self.sdropout(s), self.vdropout(v) + class LayerNorm(nn.Module): - ''' + """ Combined LayerNorm for tuples (s, V). Takes tuples (s, V) as input and as output. - ''' + """ + def __init__(self, dims): super(LayerNorm, self).__init__() self.s, self.v = dims self.scalar_norm = nn.LayerNorm(self.s) - + def forward(self, x): - ''' + """ :param x: tuple (s, V) of `torch.Tensor`, - or single `torch.Tensor` + or single `torch.Tensor` (will be assumed to be scalar channels) - ''' + """ if not self.v: return self.scalar_norm(x) s, v = x @@ -221,15 +242,16 @@ def forward(self, x): vn = torch.sqrt(torch.mean(vn, dim=-2, keepdim=True)) return self.scalar_norm(s), v / vn + class GVPConv(MessagePassing): - ''' + """ Graph convolution / message passing with Geometric Vector Perceptrons. Takes in a graph with node and edge embeddings, and returns new node embeddings. - + This does NOT do residual updates and pointwise feedforward layers ---see `GVPConvLayer`. - + :param in_dims: input node embedding dimensions (n_scalar, n_vector) :param out_dims: output node embedding dimensions (n_scalar, n_vector) :param edge_dims: input edge embedding dimensions (n_scalar, n_vector) @@ -240,63 +262,77 @@ class GVPConv(MessagePassing): :param activations: tuple of functions (scalar_act, vector_act) to use in GVPs :param vector_gate: whether to use vector gating. (vector_act will be used as sigma^+ in vector gating if `True`) - ''' - def __init__(self, in_dims, out_dims, edge_dims, - n_layers=3, module_list=None, aggr="mean", - activations=(F.relu, torch.sigmoid), vector_gate=False): + """ + + def __init__( + self, + in_dims, + out_dims, + edge_dims, + n_layers=3, + module_list=None, + aggr="mean", + activations=(F.relu, torch.sigmoid), + vector_gate=True, + ): super(GVPConv, self).__init__(aggr=aggr) self.si, self.vi = in_dims self.so, self.vo = out_dims self.se, self.ve = edge_dims - - GVP_ = functools.partial(GVP, - activations=activations, vector_gate=vector_gate) - + + GVP_ = functools.partial(GVP, activations=activations, vector_gate=vector_gate) + module_list = module_list or [] if not module_list: if n_layers == 1: module_list.append( - GVP_((2*self.si + self.se, 2*self.vi + self.ve), - (self.so, self.vo), activations=(None, None))) + GVP_( + (2 * self.si + self.se, 2 * self.vi + self.ve), + (self.so, self.vo), + activations=(None, None), + ) + ) else: module_list.append( - GVP_((2*self.si + self.se, 2*self.vi + self.ve), out_dims) + GVP_((2 * self.si + self.se, 2 * self.vi + self.ve), out_dims) ) for i in range(n_layers - 2): module_list.append(GVP_(out_dims, out_dims)) - module_list.append(GVP_(out_dims, out_dims, - activations=(None, None))) + module_list.append(GVP_(out_dims, out_dims, activations=(None, None))) self.message_func = nn.Sequential(*module_list) def forward(self, x, edge_index, edge_attr): - ''' + """ :param x: tuple (s, V) of `torch.Tensor` :param edge_index: array of shape [2, n_edges] :param edge_attr: tuple (s, V) of `torch.Tensor` - ''' + """ x_s, x_v = x - message = self.propagate(edge_index, - s=x_s, v=x_v.reshape(x_v.shape[0], 3*x_v.shape[1]), - edge_attr=edge_attr) - return _split(message, self.vo) + message = self.propagate( + edge_index, + s=x_s, + v=x_v.contiguous().view(x_v.shape[0], x_v.shape[1] * 3), + edge_attr=edge_attr, + ) + return _split(message, self.vo) def message(self, s_i, v_i, s_j, v_j, edge_attr): - v_j = v_j.view(v_j.shape[0], v_j.shape[1]//3, 3) - v_i = v_i.view(v_i.shape[0], v_i.shape[1]//3, 3) + v_j = v_j.view(v_j.shape[0], v_j.shape[1] // 3, 3) + v_i = v_i.view(v_i.shape[0], v_i.shape[1] // 3, 3) message = tuple_cat((s_j, v_j), edge_attr, (s_i, v_i)) message = self.message_func(message) return _merge(*message) class GVPConvLayer(nn.Module): - ''' - Full graph convolution / message passing layer with + """ + Full graph convolution / message passing layer with Geometric Vector Perceptrons. Residually updates node embeddings with - aggregated incoming messages, applies a pointwise feedforward + aggregated incoming messages, applies a pointwise feedforward network to node embeddings, and returns updated node embeddings. - + To only compute the aggregated messages, see `GVPConv`. - + :param node_dims: node embedding dimensions (n_scalar, n_vector) :param edge_dims: input edge embedding dimensions (n_scalar, n_vector) :param n_message: number of GVPs to use in message function @@ -308,19 +344,31 @@ class GVPConvLayer(nn.Module): :param activations: tuple of functions (scalar_act, vector_act) to use in GVPs :param vector_gate: whether to use vector gating. (vector_act will be used as sigma^+ in vector gating if `True`) - ''' - def __init__(self, node_dims, edge_dims, - n_message=3, n_feedforward=2, drop_rate=.1, - autoregressive=False, - activations=(F.relu, torch.sigmoid), vector_gate=False, - residual=True): - + """ + + def __init__( + self, + node_dims, + edge_dims, + n_message=3, + n_feedforward=2, + drop_rate=0.1, + autoregressive=False, + activations=(F.relu, torch.sigmoid), + vector_gate=True, + residual=True, + ): super(GVPConvLayer, self).__init__() - self.conv = GVPConv(node_dims, node_dims, edge_dims, n_message, - aggr="add" if autoregressive else "mean", - activations=activations, vector_gate=vector_gate) - GVP_ = functools.partial(GVP, - activations=activations, vector_gate=vector_gate) + self.conv = GVPConv( + node_dims, + node_dims, + edge_dims, + n_message, + aggr="add" if autoregressive else "mean", + activations=activations, + vector_gate=vector_gate, + ) + GVP_ = functools.partial(GVP, activations=activations, vector_gate=vector_gate) self.norm = nn.ModuleList([LayerNorm(node_dims) for _ in range(2)]) self.dropout = nn.ModuleList([Dropout(drop_rate) for _ in range(2)]) @@ -328,30 +376,28 @@ def __init__(self, node_dims, edge_dims, if n_feedforward == 1: ff_func.append(GVP_(node_dims, node_dims, activations=(None, None))) else: - hid_dims = 4*node_dims[0], 2*node_dims[1] + hid_dims = 4 * node_dims[0], 2 * node_dims[1] ff_func.append(GVP_(node_dims, hid_dims)) - for i in range(n_feedforward-2): - ff_func.append(GVP_(hid_dims, hid_dims)) + ff_func.extend(GVP_(hid_dims, hid_dims) for _ in range(n_feedforward - 2)) ff_func.append(GVP_(hid_dims, node_dims, activations=(None, None))) self.ff_func = nn.Sequential(*ff_func) self.residual = residual - def forward(self, x, edge_index, edge_attr, - autoregressive_x=None, node_mask=None): - ''' + def forward(self, x, edge_index, edge_attr, autoregressive_x=None, node_mask=None): + """ :param x: tuple (s, V) of `torch.Tensor` :param edge_index: array of shape [2, n_edges] :param edge_attr: tuple (s, V) of `torch.Tensor` - :param autoregressive_x: tuple (s, V) of `torch.Tensor`. + :param autoregressive_x: tuple (s, V) of `torch.Tensor`. If not `None`, will be used as src node embeddings - for forming messages where src >= dst. The corrent node - embeddings `x` will still be the base of the update and the + for forming messages where src >= dst. The corrent node + embeddings `x` will still be the base of the update and the pointwise feedforward. :param node_mask: array of type `bool` to index into the first dim of node embeddings (s, V). If not `None`, only these nodes will be updated. - ''' - + """ + if autoregressive_x is not None: src, dst = edge_index mask = src < dst @@ -359,29 +405,34 @@ def forward(self, x, edge_index, edge_attr, edge_index_backward = edge_index[:, ~mask] edge_attr_forward = tuple_index(edge_attr, mask) edge_attr_backward = tuple_index(edge_attr, ~mask) - + dh = tuple_sum( self.conv(x, edge_index_forward, edge_attr_forward), - self.conv(autoregressive_x, edge_index_backward, edge_attr_backward) + self.conv(autoregressive_x, edge_index_backward, edge_attr_backward), + ) + + count = ( + torch_scatter.scatter_add( + torch.ones_like(dst), dst, dim_size=dh[0].size(0) + ) + .clamp(min=1) + .unsqueeze(-1) ) - - count = scatter_add(torch.ones_like(dst), dst, - dim_size=dh[0].size(0)).clamp(min=1).unsqueeze(-1) - + dh = dh[0] / count, dh[1] / count.unsqueeze(-1) else: dh = self.conv(x, edge_index, edge_attr) - + if node_mask is not None: x_ = x x, dh = tuple_index(x, node_mask), tuple_index(dh, node_mask) - + x = self.norm[0](tuple_sum(x, self.dropout[0](dh))) if self.residual else dh - + dh = self.ff_func(x) x = self.norm[1](tuple_sum(x, self.dropout[1](dh))) if self.residual else dh - + if node_mask is not None: x_[0][node_mask], x_[1][node_mask] = x[0], x[1] x = x_ diff --git a/models/layers/spherenet_layer.py b/models/layers/spherenet_layer.py new file mode 100644 index 0000000..c46be2e --- /dev/null +++ b/models/layers/spherenet_layer.py @@ -0,0 +1,564 @@ +################################################################ +# Implementation of SphereNet layers +# +# Paper: Spherical Message Passing for 3D Graph Networks +# by Y Liu, L Wang, M Liu, X Zhang, B Oztekin, and S Ji +# +# Orginal repository: https://github.com/divelab/DIG +################################################################ + +import torch +from torch import nn +from torch.nn import Linear, Embedding +from torch_geometric.nn.inits import glorot_orthogonal +from torch_scatter import scatter +from math import sqrt + +import numpy as np +from scipy.optimize import brentq +from scipy import special as sp +import torch +from math import pi as PI + +import sympy as sym + +import torch +from torch_scatter import scatter +from torch_sparse import SparseTensor +from math import pi as PI + +def swish(x): + return x * torch.sigmoid(x) + +class emb(torch.nn.Module): + def __init__(self, num_spherical, num_radial, cutoff, envelope_exponent): + super(emb, self).__init__() + self.dist_emb = dist_emb(num_radial, cutoff, envelope_exponent) + self.angle_emb = angle_emb(num_spherical, num_radial, cutoff, envelope_exponent) + self.torsion_emb = torsion_emb(num_spherical, num_radial, cutoff, envelope_exponent) + self.reset_parameters() + + def reset_parameters(self): + self.dist_emb.reset_parameters() + + def forward(self, dist, angle, torsion, idx_kj): + dist_emb = self.dist_emb(dist) + angle_emb = self.angle_emb(dist, angle, idx_kj) + torsion_emb = self.torsion_emb(dist, angle, torsion, idx_kj) + return dist_emb, angle_emb, torsion_emb + +class ResidualLayer(torch.nn.Module): + def __init__(self, hidden_channels, act=swish): + super(ResidualLayer, self).__init__() + self.act = act + self.lin1 = Linear(hidden_channels, hidden_channels) + self.lin2 = Linear(hidden_channels, hidden_channels) + + self.reset_parameters() + + def reset_parameters(self): + glorot_orthogonal(self.lin1.weight, scale=2.0) + self.lin1.bias.data.fill_(0) + glorot_orthogonal(self.lin2.weight, scale=2.0) + self.lin2.bias.data.fill_(0) + + def forward(self, x): + return x + self.act(self.lin2(self.act(self.lin1(x)))) + + +class init(torch.nn.Module): + def __init__(self, num_radial, hidden_channels, act=swish, use_node_features=True): + super(init, self).__init__() + self.act = act + self.use_node_features = use_node_features + if self.use_node_features: + self.emb = Embedding(95, hidden_channels) + else: # option to use no node features and a learned embedding vector for each node instead + self.node_embedding = nn.Parameter(torch.empty((hidden_channels,))) + nn.init.normal_(self.node_embedding) + self.lin_rbf_0 = Linear(num_radial, hidden_channels) + self.lin = Linear(3 * hidden_channels, hidden_channels) + self.lin_rbf_1 = nn.Linear(num_radial, hidden_channels, bias=False) + self.reset_parameters() + + def reset_parameters(self): + if self.use_node_features: + self.emb.weight.data.uniform_(-sqrt(3), sqrt(3)) + self.lin_rbf_0.reset_parameters() + self.lin.reset_parameters() + glorot_orthogonal(self.lin_rbf_1.weight, scale=2.0) + + def forward(self, x, emb, i, j): + rbf,_,_ = emb + if self.use_node_features: + x = self.emb(x) + else: + x = self.node_embedding[None, :].expand(x.shape[0], -1) + rbf0 = self.act(self.lin_rbf_0(rbf)) + e1 = self.act(self.lin(torch.cat([x[i], x[j], rbf0], dim=-1))) + e2 = self.lin_rbf_1(rbf) * e1 + + return e1, e2 + + +class update_e(torch.nn.Module): + def __init__(self, hidden_channels, int_emb_size, basis_emb_size_dist, basis_emb_size_angle, basis_emb_size_torsion, num_spherical, num_radial, + num_before_skip, num_after_skip, act=swish): + super(update_e, self).__init__() + self.act = act + self.lin_rbf1 = nn.Linear(num_radial, basis_emb_size_dist, bias=False) + self.lin_rbf2 = nn.Linear(basis_emb_size_dist, hidden_channels, bias=False) + self.lin_sbf1 = nn.Linear(num_spherical * num_radial, basis_emb_size_angle, bias=False) + self.lin_sbf2 = nn.Linear(basis_emb_size_angle, int_emb_size, bias=False) + self.lin_t1 = nn.Linear(num_spherical * num_spherical * num_radial, basis_emb_size_torsion, bias=False) + self.lin_t2 = nn.Linear(basis_emb_size_torsion, int_emb_size, bias=False) + self.lin_rbf = nn.Linear(num_radial, hidden_channels, bias=False) + + self.lin_kj = nn.Linear(hidden_channels, hidden_channels) + self.lin_ji = nn.Linear(hidden_channels, hidden_channels) + + self.lin_down = nn.Linear(hidden_channels, int_emb_size, bias=False) + self.lin_up = nn.Linear(int_emb_size, hidden_channels, bias=False) + + self.layers_before_skip = torch.nn.ModuleList([ + ResidualLayer(hidden_channels, act) + for _ in range(num_before_skip) + ]) + self.lin = nn.Linear(hidden_channels, hidden_channels) + self.layers_after_skip = torch.nn.ModuleList([ + ResidualLayer(hidden_channels, act) + for _ in range(num_after_skip) + ]) + + self.reset_parameters() + + def reset_parameters(self): + glorot_orthogonal(self.lin_rbf1.weight, scale=2.0) + glorot_orthogonal(self.lin_rbf2.weight, scale=2.0) + glorot_orthogonal(self.lin_sbf1.weight, scale=2.0) + glorot_orthogonal(self.lin_sbf2.weight, scale=2.0) + glorot_orthogonal(self.lin_t1.weight, scale=2.0) + glorot_orthogonal(self.lin_t2.weight, scale=2.0) + + glorot_orthogonal(self.lin_kj.weight, scale=2.0) + self.lin_kj.bias.data.fill_(0) + glorot_orthogonal(self.lin_ji.weight, scale=2.0) + self.lin_ji.bias.data.fill_(0) + + glorot_orthogonal(self.lin_down.weight, scale=2.0) + glorot_orthogonal(self.lin_up.weight, scale=2.0) + + for res_layer in self.layers_before_skip: + res_layer.reset_parameters() + glorot_orthogonal(self.lin.weight, scale=2.0) + self.lin.bias.data.fill_(0) + for res_layer in self.layers_after_skip: + res_layer.reset_parameters() + + glorot_orthogonal(self.lin_rbf.weight, scale=2.0) + + def forward(self, x, emb, idx_kj, idx_ji): + rbf0, sbf, t = emb + x1,_ = x + + x_ji = self.act(self.lin_ji(x1)) + x_kj = self.act(self.lin_kj(x1)) + + rbf = self.lin_rbf1(rbf0) + rbf = self.lin_rbf2(rbf) + x_kj = x_kj * rbf + + x_kj = self.act(self.lin_down(x_kj)) + + sbf = self.lin_sbf1(sbf) + sbf = self.lin_sbf2(sbf) + x_kj = x_kj[idx_kj] * sbf + + t = self.lin_t1(t) + t = self.lin_t2(t) + x_kj = x_kj * t + + x_kj = scatter(x_kj, idx_ji, dim=0, dim_size=x1.size(0)) + x_kj = self.act(self.lin_up(x_kj)) + + e1 = x_ji + x_kj + for layer in self.layers_before_skip: + e1 = layer(e1) + e1 = self.act(self.lin(e1)) + x1 + for layer in self.layers_after_skip: + e1 = layer(e1) + e2 = self.lin_rbf(rbf0) * e1 + + return e1, e2 + + +class update_v(torch.nn.Module): + def __init__(self, hidden_channels, out_emb_channels, out_channels, num_output_layers, act, output_init): + super(update_v, self).__init__() + self.act = act + self.output_init = output_init + + self.lin_up = nn.Linear(hidden_channels, out_emb_channels, bias=True) + self.lins = torch.nn.ModuleList() + for _ in range(num_output_layers): + self.lins.append(nn.Linear(out_emb_channels, out_emb_channels)) + self.lin = nn.Linear(out_emb_channels, out_channels, bias=False) + + self.reset_parameters() + + def reset_parameters(self): + glorot_orthogonal(self.lin_up.weight, scale=2.0) + for lin in self.lins: + glorot_orthogonal(lin.weight, scale=2.0) + lin.bias.data.fill_(0) + if self.output_init == 'zeros': + self.lin.weight.data.fill_(0) + if self.output_init == 'GlorotOrthogonal': + glorot_orthogonal(self.lin.weight, scale=2.0) + + def forward(self, e, i): + _, e2 = e + v = scatter(e2, i, dim=0) + v = self.lin_up(v) + for lin in self.lins: + v = self.act(lin(v)) + v = self.lin(v) + return v + + +class update_u(torch.nn.Module): + def __init__(self): + super(update_u, self).__init__() + + def forward(self, u, v, batch): + u += scatter(v, batch, dim=0) + return u + +# Based on the code from: https://github.com/klicperajo/dimenet, +# https://github.com/rusty1s/pytorch_geometric/blob/master/torch_geometric/nn/models/dimenet_utils.py + + +def Jn(r, n): + return np.sqrt(np.pi / (2 * r)) * sp.jv(n + 0.5, r) + + +def Jn_zeros(n, k): + zerosj = np.zeros((n, k), dtype='float32') + zerosj[0] = np.arange(1, k + 1) * np.pi + points = np.arange(1, k + n) * np.pi + racines = np.zeros(k + n - 1, dtype='float32') + for i in range(1, n): + for j in range(k + n - 1 - i): + foo = brentq(Jn, points[j], points[j + 1], (i, )) + racines[j] = foo + points = racines + zerosj[i][:k] = racines[:k] + + return zerosj + + +def spherical_bessel_formulas(n): + x = sym.symbols('x') + + f = [sym.sin(x) / x] + a = sym.sin(x) / x + for i in range(1, n): + b = sym.diff(a, x) / x + f += [sym.simplify(b * (-x)**i)] + a = sym.simplify(b) + return f + + +def bessel_basis(n, k): + zeros = Jn_zeros(n, k) + normalizer = [] + for order in range(n): + normalizer_tmp = [] + for i in range(k): + normalizer_tmp += [0.5 * Jn(zeros[order, i], order + 1)**2] + normalizer_tmp = 1 / np.array(normalizer_tmp)**0.5 + normalizer += [normalizer_tmp] + + f = spherical_bessel_formulas(n) + x = sym.symbols('x') + bess_basis = [] + for order in range(n): + bess_basis_tmp = [] + for i in range(k): + bess_basis_tmp += [ + sym.simplify(normalizer[order][i] * + f[order].subs(x, zeros[order, i] * x)) + ] + bess_basis += [bess_basis_tmp] + return bess_basis + + +def sph_harm_prefactor(k, m): + return ((2 * k + 1) * np.math.factorial(k - abs(m)) / + (4 * np.pi * np.math.factorial(k + abs(m))))**0.5 + + +def associated_legendre_polynomials(k, zero_m_only=True): + z = sym.symbols('z') + P_l_m = [[0] * (j + 1) for j in range(k)] + + P_l_m[0][0] = 1 + if k > 0: + P_l_m[1][0] = z + + for j in range(2, k): + P_l_m[j][0] = sym.simplify(((2 * j - 1) * z * P_l_m[j - 1][0] - + (j - 1) * P_l_m[j - 2][0]) / j) + if not zero_m_only: + for i in range(1, k): + P_l_m[i][i] = sym.simplify((1 - 2 * i) * P_l_m[i - 1][i - 1]) + if i + 1 < k: + P_l_m[i + 1][i] = sym.simplify( + (2 * i + 1) * z * P_l_m[i][i]) + for j in range(i + 2, k): + P_l_m[j][i] = sym.simplify( + ((2 * j - 1) * z * P_l_m[j - 1][i] - + (i + j - 1) * P_l_m[j - 2][i]) / (j - i)) + + return P_l_m + + +def real_sph_harm(l, zero_m_only=False, spherical_coordinates=True): + """ + Computes formula strings of the the real part of the spherical harmonics up to order l (excluded). + Variables are either cartesian coordinates x,y,z on the unit sphere or spherical coordinates phi and theta. + """ + if not zero_m_only: + x = sym.symbols('x') + y = sym.symbols('y') + S_m = [x*0] + C_m = [1+0*x] + # S_m = [0] + # C_m = [1] + for i in range(1, l): + x = sym.symbols('x') + y = sym.symbols('y') + S_m += [x*S_m[i-1] + y*C_m[i-1]] + C_m += [x*C_m[i-1] - y*S_m[i-1]] + + P_l_m = associated_legendre_polynomials(l, zero_m_only) + if spherical_coordinates: + theta = sym.symbols('theta') + z = sym.symbols('z') + for i in range(len(P_l_m)): + for j in range(len(P_l_m[i])): + if type(P_l_m[i][j]) != int: + P_l_m[i][j] = P_l_m[i][j].subs(z, sym.cos(theta)) + if not zero_m_only: + phi = sym.symbols('phi') + for i in range(len(S_m)): + S_m[i] = S_m[i].subs(x, sym.sin( + theta)*sym.cos(phi)).subs(y, sym.sin(theta)*sym.sin(phi)) + for i in range(len(C_m)): + C_m[i] = C_m[i].subs(x, sym.sin( + theta)*sym.cos(phi)).subs(y, sym.sin(theta)*sym.sin(phi)) + + Y_func_l_m = [['0']*(2*j + 1) for j in range(l)] + for i in range(l): + Y_func_l_m[i][0] = sym.simplify(sph_harm_prefactor(i, 0) * P_l_m[i][0]) + + if not zero_m_only: + for i in range(1, l): + for j in range(1, i + 1): + Y_func_l_m[i][j] = sym.simplify( + 2**0.5 * sph_harm_prefactor(i, j) * C_m[j] * P_l_m[i][j]) + for i in range(1, l): + for j in range(1, i + 1): + Y_func_l_m[i][-j] = sym.simplify( + 2**0.5 * sph_harm_prefactor(i, -j) * S_m[j] * P_l_m[i][j]) + + return Y_func_l_m + + +class Envelope(torch.nn.Module): + def __init__(self, exponent): + super(Envelope, self).__init__() + self.p = exponent + 1 + self.a = -(self.p + 1) * (self.p + 2) / 2 + self.b = self.p * (self.p + 2) + self.c = -self.p * (self.p + 1) / 2 + + def forward(self, x): + p, a, b, c = self.p, self.a, self.b, self.c + x_pow_p0 = x.pow(p - 1) + x_pow_p1 = x_pow_p0 * x + x_pow_p2 = x_pow_p1 * x + return 1. / x + a * x_pow_p0 + b * x_pow_p1 + c * x_pow_p2 + + +class dist_emb(torch.nn.Module): + def __init__(self, num_radial, cutoff=5.0, envelope_exponent=5): + super(dist_emb, self).__init__() + self.cutoff = cutoff + self.envelope = Envelope(envelope_exponent) + + self.freq = torch.nn.Parameter(torch.Tensor(num_radial)) + + self.reset_parameters() + + def reset_parameters(self): + self.freq.data = torch.arange(1, self.freq.numel() + 1).float().mul_(PI) + + def forward(self, dist): + dist = dist.unsqueeze(-1) / self.cutoff + return self.envelope(dist) * (self.freq * dist).sin() + + +class angle_emb(torch.nn.Module): + def __init__(self, num_spherical, num_radial, cutoff=5.0, + envelope_exponent=5): + super(angle_emb, self).__init__() + assert num_radial <= 64 + self.num_spherical = num_spherical + self.num_radial = num_radial + self.cutoff = cutoff + # self.envelope = Envelope(envelope_exponent) + + bessel_forms = bessel_basis(num_spherical, num_radial) + sph_harm_forms = real_sph_harm(num_spherical) + self.sph_funcs = [] + self.bessel_funcs = [] + + x, theta = sym.symbols('x theta') + modules = {'sin': torch.sin, 'cos': torch.cos} + for i in range(num_spherical): + if i == 0: + sph1 = sym.lambdify([theta], sph_harm_forms[i][0], modules)(0) + self.sph_funcs.append(lambda x: torch.zeros_like(x) + sph1) + else: + sph = sym.lambdify([theta], sph_harm_forms[i][0], modules) + self.sph_funcs.append(sph) + for j in range(num_radial): + bessel = sym.lambdify([x], bessel_forms[i][j], modules) + self.bessel_funcs.append(bessel) + + def forward(self, dist, angle, idx_kj): + dist = dist / self.cutoff + rbf = torch.stack([f(dist) for f in self.bessel_funcs], dim=1) + # rbf = self.envelope(dist).unsqueeze(-1) * rbf + + cbf = torch.stack([f(angle) for f in self.sph_funcs], dim=1) + + n, k = self.num_spherical, self.num_radial + out = (rbf[idx_kj].view(-1, n, k) * cbf.view(-1, n, 1)).view(-1, n * k) + return out + + +class torsion_emb(torch.nn.Module): + def __init__(self, num_spherical, num_radial, cutoff=5.0, + envelope_exponent=5): + super(torsion_emb, self).__init__() + assert num_radial <= 64 + self.num_spherical = num_spherical # + self.num_radial = num_radial + self.cutoff = cutoff + # self.envelope = Envelope(envelope_exponent) + + bessel_forms = bessel_basis(num_spherical, num_radial) + sph_harm_forms = real_sph_harm(num_spherical, zero_m_only=False) + self.sph_funcs = [] + self.bessel_funcs = [] + + x = sym.symbols('x') + theta = sym.symbols('theta') + phi = sym.symbols('phi') + modules = {'sin': torch.sin, 'cos': torch.cos} + for i in range(self.num_spherical): + if i == 0: + sph1 = sym.lambdify([theta, phi], sph_harm_forms[i][0], modules) + self.sph_funcs.append(lambda x, y: torch.zeros_like(x) + torch.zeros_like(y) + sph1(0,0)) #torch.zeros_like(x) + torch.zeros_like(y) + else: + for k in range(-i, i + 1): + sph = sym.lambdify([theta, phi], sph_harm_forms[i][k+i], modules) + self.sph_funcs.append(sph) + for j in range(self.num_radial): + bessel = sym.lambdify([x], bessel_forms[i][j], modules) + self.bessel_funcs.append(bessel) + + def forward(self, dist, angle, phi, idx_kj): + dist = dist / self.cutoff + rbf = torch.stack([f(dist) for f in self.bessel_funcs], dim=1) + cbf = torch.stack([f(angle, phi) for f in self.sph_funcs], dim=1) + + n, k = self.num_spherical, self.num_radial + out = (rbf[idx_kj].view(-1, 1, n, k) * cbf.view(-1, n, n, 1)).view(-1, n * n * k) + return out + + +# Based on the code from: https://github.com/klicperajo/dimenet, +# https://github.com/rusty1s/pytorch_geometric/blob/master/torch_geometric/nn/models/dimenet.py + +def xyz_to_dat(pos, edge_index, num_nodes, use_torsion = False): + """ + Compute the diatance, angle, and torsion from geometric information. + + Args: + pos: Geometric information for every node in the graph. + edge_index: Edge index of the graph. + number_nodes: Number of nodes in the graph. + use_torsion: If set to :obj:`True`, will return distance, angle and torsion, otherwise only return distance and angle (also retrun some useful index). (default: :obj:`False`) + """ + j, i = edge_index # j->i + + # Calculate distances. # number of edges + dist = (pos[i] - pos[j]).pow(2).sum(dim=-1).sqrt() + + value = torch.arange(j.size(0), device=j.device) + adj_t = SparseTensor(row=i, col=j, value=value, sparse_sizes=(num_nodes, num_nodes)) + adj_t_row = adj_t[j] + num_triplets = adj_t_row.set_value(None).sum(dim=1).to(torch.long) + + # Node indices (k->j->i) for triplets. + idx_i = i.repeat_interleave(num_triplets) + idx_j = j.repeat_interleave(num_triplets) + idx_k = adj_t_row.storage.col() + mask = idx_i != idx_k + idx_i, idx_j, idx_k = idx_i[mask], idx_j[mask], idx_k[mask] + + # Edge indices (k-j, j->i) for triplets. + idx_kj = adj_t_row.storage.value()[mask] + idx_ji = adj_t_row.storage.row()[mask] + + # Calculate angles. 0 to pi + pos_ji = pos[idx_i] - pos[idx_j] + pos_jk = pos[idx_k] - pos[idx_j] + a = (pos_ji * pos_jk).sum(dim=-1) # cos_angle * |pos_ji| * |pos_jk| + b = torch.cross(pos_ji, pos_jk).norm(dim=-1) # sin_angle * |pos_ji| * |pos_jk| + angle = torch.atan2(b, a) + + + if use_torsion: + # Prepare torsion idxes. + idx_batch = torch.arange(len(idx_i),device=j.device) + idx_k_n = adj_t[idx_j].storage.col() + repeat = num_triplets + num_triplets_t = num_triplets.repeat_interleave(repeat)[mask] + idx_i_t = idx_i.repeat_interleave(num_triplets_t) + idx_j_t = idx_j.repeat_interleave(num_triplets_t) + idx_k_t = idx_k.repeat_interleave(num_triplets_t) + idx_batch_t = idx_batch.repeat_interleave(num_triplets_t) + mask = idx_i_t != idx_k_n + idx_i_t, idx_j_t, idx_k_t, idx_k_n, idx_batch_t = idx_i_t[mask], idx_j_t[mask], idx_k_t[mask], idx_k_n[mask], idx_batch_t[mask] + + # Calculate torsions. + pos_j0 = pos[idx_k_t] - pos[idx_j_t] + pos_ji = pos[idx_i_t] - pos[idx_j_t] + pos_jk = pos[idx_k_n] - pos[idx_j_t] + dist_ji = pos_ji.pow(2).sum(dim=-1).sqrt() + plane1 = torch.cross(pos_ji, pos_j0) + plane2 = torch.cross(pos_ji, pos_jk) + a = (plane1 * plane2).sum(dim=-1) # cos_angle * |plane1| * |plane2| + b = (torch.cross(plane1, plane2) * pos_ji).sum(dim=-1) / dist_ji + torsion1 = torch.atan2(b, a) # -pi to pi + torsion1[torsion1<=0]+=2*PI # 0 to 2pi + torsion = scatter(torsion1,idx_batch_t,reduce='min') + + return dist, angle, torsion, i, j, idx_kj, idx_ji + + else: + return dist, angle, i, j, idx_kj, idx_ji diff --git a/src/tfn_layers.py b/models/layers/tfn_layer.py similarity index 56% rename from src/tfn_layers.py rename to models/layers/tfn_layer.py index eb2173d..e260009 100644 --- a/src/tfn_layers.py +++ b/models/layers/tfn_layer.py @@ -1,38 +1,36 @@ import torch from torch_scatter import scatter - import e3nn -from e3nn import o3 -from e3nn import nn -from src.modules.irreps_tools import irreps2gate +from models.mace_modules.irreps_tools import irreps2gate class TensorProductConvLayer(torch.nn.Module): + """Tensor Field Network GNN Layer in e3nn + + Implements a Tensor Field Network equivariant GNN layer for higher-order tensors, using e3nn. + Implementation adapted from: https://github.com/gcorso/DiffDock/ + + Paper: Tensor Field Networks, Thomas, Smidt et al. + """ def __init__( - self, - in_irreps, + self, + in_irreps, out_irreps, sh_irreps, - edge_feats_dim, - hidden_dim, + edge_feats_dim, + mlp_dim, aggr="add", batch_norm=False, - gate=True + gate=False, ): - """Tensor Field Network GNN Layer - - Implements a Tensor Field Network equivariant GNN layer for higher-order tensors, using e3nn. - Implementation adapted from: https://github.com/gcorso/DiffDock/ - - Paper: Tensor Field Networks, Thomas, Smidt et al. - + """ Args: in_irreps: (e3nn.o3.Irreps) Input irreps dimensions out_irreps: (e3nn.o3.Irreps) Output irreps dimensions sh_irreps: (e3nn.o3.Irreps) Spherical harmonic irreps dimensions edge_feats_dim: (int) Edge feature dimensions - hidden_dim: (int) Hidden dimension of MLP for computing tensor product weights + mlp_dim: (int) Hidden dimension of MLP for computing tensor product weights aggr: (str) Message passing aggregator batch_norm: (bool) Whether to apply equivariant batch norm gate: (bool) Whether to apply gated non-linearity @@ -46,16 +44,20 @@ def __init__( if gate: # Optionally apply gated non-linearity - irreps_scalars, irreps_gates, irreps_gated = irreps2gate(o3.Irreps(out_irreps)) - act_scalars = [torch.nn.functional.silu for _, ir in irreps_scalars] + irreps_scalars, irreps_gates, irreps_gated = irreps2gate( + e3nn.o3.Irreps(out_irreps) + ) + act_scalars = [torch.nn.functional.silu for _, ir in irreps_scalars] act_gates = [torch.sigmoid for _, ir in irreps_gates] if irreps_gated.num_irreps == 0: - self.gate = nn.Activation(out_irreps, acts=[torch.nn.functional.silu]) + self.gate = e3nn.nn.Activation(out_irreps, acts=[torch.nn.functional.silu]) else: - self.gate = nn.Gate( - irreps_scalars, act_scalars, # scalar - irreps_gates, act_gates, # gates (scalars) - irreps_gated # gated tensors + self.gate = e3nn.nn.Gate( + irreps_scalars, + act_scalars, # scalar + irreps_gates, + act_gates, # gates (scalars) + irreps_gated, # gated tensors ) # Output irreps for the tensor product must be updated self.out_irreps = out_irreps = self.gate.irreps_in @@ -63,22 +65,24 @@ def __init__( self.gate = None # Tensor product over edges to construct messages - self.tp = o3.FullyConnectedTensorProduct(in_irreps, sh_irreps, out_irreps, shared_weights=False) + self.tp = e3nn.o3.FullyConnectedTensorProduct( + in_irreps, sh_irreps, out_irreps, shared_weights=False + ) # MLP used to compute weights of tensor product self.fc = torch.nn.Sequential( - torch.nn.Linear(edge_feats_dim, hidden_dim), + torch.nn.Linear(edge_feats_dim, mlp_dim), torch.nn.ReLU(), - torch.nn.Linear(hidden_dim, self.tp.weight_numel) + torch.nn.Linear(mlp_dim, self.tp.weight_numel), ) # Optional equivariant batch norm - self.batch_norm = nn.BatchNorm(out_irreps) if batch_norm else None + self.batch_norm = e3nn.nn.BatchNorm(out_irreps) if batch_norm else None - def forward(self, node_attr, edge_index, edge_attr, edge_feat): + def forward(self, node_attr, edge_index, edge_sh, edge_feat): src, dst = edge_index - # Compute messages - tp = self.tp(node_attr[dst], edge_attr, self.fc(edge_feat)) + # Compute messages + tp = self.tp(node_attr[dst], edge_sh, self.fc(edge_feat)) # Aggregate messages out = scatter(tp, src, dim=0, reduce=self.aggr) # Optionally apply gated non-linearity and/or batch norm diff --git a/models/mace.py b/models/mace.py new file mode 100644 index 0000000..69498a5 --- /dev/null +++ b/models/mace.py @@ -0,0 +1,190 @@ +from typing import Optional + +import torch +from torch.nn import functional as F +from torch_geometric.nn import global_add_pool, global_mean_pool +import e3nn + +from models.mace_modules.irreps_tools import reshape_irreps +from models.mace_modules.blocks import ( + EquivariantProductBasisBlock, + RadialEmbeddingBlock, +) +from models.layers.tfn_layer import TensorProductConvLayer + + +class MACEModel(torch.nn.Module): + """ + MACE model from "MACE: Higher Order Equivariant Message Passing Neural Networks". + """ + def __init__( + self, + r_max: float = 10.0, + num_bessel: int = 8, + num_polynomial_cutoff: int = 5, + max_ell: int = 2, + correlation: int = 3, + num_layers: int = 5, + emb_dim: int = 64, + hidden_irreps: Optional[e3nn.o3.Irreps] = None, + mlp_dim: int = 256, + in_dim: int = 1, + out_dim: int = 1, + aggr: str = "sum", + pool: str = "sum", + batch_norm: bool = True, + residual: bool = True, + equivariant_pred: bool = False + ): + """ + Parameters: + - r_max (float): Maximum distance for Bessel basis functions (default: 10.0) + - num_bessel (int): Number of Bessel basis functions (default: 8) + - num_polynomial_cutoff (int): Number of polynomial cutoff basis functions (default: 5) + - max_ell (int): Maximum degree of spherical harmonics basis functions (default: 2) + - correlation (int): Local correlation order = body order - 1 (default: 3) + - num_layers (int): Number of layers in the model (default: 5) + - emb_dim (int): Scalar feature embedding dimension (default: 64) + - hidden_irreps (Optional[e3nn.o3.Irreps]): Hidden irreps (default: None) + - mlp_dim (int): Dimension of MLP for computing tensor product weights (default: 256) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - aggr (str): Aggregation method to be used (default: "sum") + - pool (str): Global pooling method to be used (default: "sum") + - batch_norm (bool): Whether to use batch normalization (default: True) + - residual (bool): Whether to use residual connections (default: True) + - equivariant_pred (bool): Whether it is an equivariant prediction task (default: False) + + Note: + - If `hidden_irreps` is None, the irreps for the intermediate features are computed + using `emb_dim` and `max_ell`. + - The `equivariant_pred` parameter determines whether it is an equivariant prediction task. + If set to True, equivariant prediction will be performed. + """ + super().__init__() + + self.r_max = r_max + self.max_ell = max_ell + self.num_layers = num_layers + self.emb_dim = emb_dim + self.mlp_dim = mlp_dim + self.residual = residual + self.batch_norm = batch_norm + self.hidden_irreps = hidden_irreps + self.equivariant_pred = equivariant_pred + + # Edge embedding + self.radial_embedding = RadialEmbeddingBlock( + r_max=r_max, + num_bessel=num_bessel, + num_polynomial_cutoff=num_polynomial_cutoff, + ) + sh_irreps = e3nn.o3.Irreps.spherical_harmonics(max_ell) + self.spherical_harmonics = e3nn.o3.SphericalHarmonics( + sh_irreps, normalize=True, normalization="component" + ) + + # Embedding lookup for initial node features + self.emb_in = torch.nn.Embedding(in_dim, emb_dim) + + # Set hidden irreps if none are provided + if hidden_irreps is None: + hidden_irreps = (sh_irreps * emb_dim).sort()[0].simplify() + # Note: This defaults to O(3) equivariant layers + # It is possible to use SO(3) equivariance by passing the appropriate irreps + + self.convs = torch.nn.ModuleList() + self.prods = torch.nn.ModuleList() + self.reshapes = torch.nn.ModuleList() + + # First layer: scalar only -> tensor + self.convs.append( + TensorProductConvLayer( + in_irreps=e3nn.o3.Irreps(f'{emb_dim}x0e'), + out_irreps=hidden_irreps, + sh_irreps=sh_irreps, + edge_feats_dim=self.radial_embedding.out_dim, + mlp_dim=mlp_dim, + aggr=aggr, + batch_norm=batch_norm, + gate=False, + ) + ) + self.reshapes.append(reshape_irreps(hidden_irreps)) + self.prods.append( + EquivariantProductBasisBlock( + node_feats_irreps=hidden_irreps, + target_irreps=hidden_irreps, + correlation=correlation, + element_dependent=False, + num_elements=in_dim, + use_sc=residual + ) + ) + + # Intermediate layers: tensor -> tensor + for _ in range(num_layers - 1): + self.convs.append( + TensorProductConvLayer( + in_irreps=hidden_irreps, + out_irreps=hidden_irreps, + sh_irreps=sh_irreps, + edge_feats_dim=self.radial_embedding.out_dim, + mlp_dim=mlp_dim, + aggr=aggr, + batch_norm=batch_norm, + gate=False, + ) + ) + self.reshapes.append(reshape_irreps(hidden_irreps)) + self.prods.append( + EquivariantProductBasisBlock( + node_feats_irreps=hidden_irreps, + target_irreps=hidden_irreps, + correlation=correlation, + element_dependent=False, + num_elements=in_dim, + use_sc=residual + ) + ) + + # Global pooling/readout function + self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] + + if self.equivariant_pred: + # Linear predictor for equivariant tasks using geometric features + self.pred = torch.nn.Linear(hidden_irreps.dim, out_dim) + else: + # MLP predictor for invariant tasks using only scalar features + self.pred = torch.nn.Sequential( + torch.nn.Linear(emb_dim, emb_dim), + torch.nn.ReLU(), + torch.nn.Linear(emb_dim, out_dim) + ) + + def forward(self, batch): + # Node embedding + h = self.emb_in(batch.atoms) # (n,) -> (n, d) + + # Edge features + vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] + lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] + + edge_sh = self.spherical_harmonics(vectors) + edge_feats = self.radial_embedding(lengths) + + for conv, reshape, prod in zip(self.convs, self.reshapes, self.prods): + # Message passing layer + h_update = conv(h, batch.edge_index, edge_sh, edge_feats) + + # Update node features + sc = F.pad(h, (0, h_update.shape[-1] - h.shape[-1])) + h = prod(reshape(h_update), sc, None) + + out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) + + if not self.equivariant_pred: + # Select only scalars for invariant prediction + out = out[:,:self.emb_dim] + + return self.pred(out) # (batch_size, out_dim) diff --git a/src/modules/__init__.py b/models/mace_modules/__init__.py similarity index 100% rename from src/modules/__init__.py rename to models/mace_modules/__init__.py diff --git a/src/modules/blocks.py b/models/mace_modules/blocks.py similarity index 99% rename from src/modules/blocks.py rename to models/mace_modules/blocks.py index 54cbb83..60aaed2 100644 --- a/src/modules/blocks.py +++ b/models/mace_modules/blocks.py @@ -1,7 +1,7 @@ ########################################################################################### # Elementary Block for Building O(3) Equivariant Higher Order Message Passing Neural Network # Authors: Ilyes Batatia, Gregor Simm -# This program is distributed under the ASL License (see ASL.md) +# This program is distributed under the MIT License (see MIT.md) ########################################################################################### from abc import ABC, abstractmethod diff --git a/src/modules/cg.py b/models/mace_modules/cg.py similarity index 98% rename from src/modules/cg.py rename to models/mace_modules/cg.py index b1fc237..8f944a0 100644 --- a/src/modules/cg.py +++ b/models/mace_modules/cg.py @@ -1,7 +1,7 @@ ########################################################################################### # Higher Order Real Clebsch Gordan (based on e3nn by Mario Geiger) # Authors: Ilyes Batatia -# This program is distributed under the ASL License (see ASL.md) +# This program is distributed under the MIT License (see MIT.md) ########################################################################################### import collections diff --git a/src/modules/irreps_tools.py b/models/mace_modules/irreps_tools.py similarity index 98% rename from src/modules/irreps_tools.py rename to models/mace_modules/irreps_tools.py index 09a1758..1989da8 100644 --- a/src/modules/irreps_tools.py +++ b/models/mace_modules/irreps_tools.py @@ -1,7 +1,7 @@ ########################################################################################### # Elementary tools for handling irreducible representations # Authors: Ilyes Batatia, Gregor Simm -# This program is distributed under the ASL License (see ASL.md) +# This program is distributed under the MIT License (see MIT.md) ########################################################################################### from typing import List, Tuple diff --git a/src/modules/radial.py b/models/mace_modules/radial.py similarity index 89% rename from src/modules/radial.py rename to models/mace_modules/radial.py index c8686a7..61368ec 100644 --- a/src/modules/radial.py +++ b/models/mace_modules/radial.py @@ -1,11 +1,12 @@ ########################################################################################### # Radial basis and cutoff # Authors: Ilyes Batatia, Gregor Simm -# This program is distributed under the ASL License (see ASL.md) +# This program is distributed under the MIT License (see MIT.md) ########################################################################################### import numpy as np import torch +from e3nn.util.jit import compile_mode class BesselBasis(torch.nn.Module): @@ -40,7 +41,7 @@ def __init__(self, r_max: float, num_basis=8, trainable=False): torch.tensor(np.sqrt(2.0 / r_max), dtype=torch.get_default_dtype()), ) - def forward(self, x: torch.Tensor,) -> torch.Tensor: # [..., 1] + def forward(self, x: torch.Tensor) -> torch.Tensor: # [..., 1] numerator = torch.sin(self.bessel_weights * x) # [..., num_basis] return self.prefactor * (numerator / x) @@ -68,17 +69,13 @@ def __init__(self, r_max: float, p=6): ) def forward(self, x: torch.Tensor) -> torch.Tensor: - # yapf: disable envelope = ( 1.0 - ((self.p + 1.0) * (self.p + 2.0) / 2.0) * torch.pow(x / self.r_max, self.p) + self.p * (self.p + 2.0) * torch.pow(x / self.r_max, self.p + 1) - (self.p * (self.p + 1.0) / 2) * torch.pow(x / self.r_max, self.p + 2) ) - # yapf: enable - - # noinspection PyUnresolvedReferences - return envelope * (x < self.r_max).type(torch.get_default_dtype()) + return envelope * (x < self.r_max) def __repr__(self): return f"{self.__class__.__name__}(p={self.p}, r_max={self.r_max})" diff --git a/src/modules/symmetric_contraction.py b/models/mace_modules/symmetric_contraction.py similarity index 99% rename from src/modules/symmetric_contraction.py rename to models/mace_modules/symmetric_contraction.py index 85da8a0..7334a00 100644 --- a/src/modules/symmetric_contraction.py +++ b/models/mace_modules/symmetric_contraction.py @@ -2,7 +2,7 @@ # Implementation of the symmetric contraction algorithm presented in the MACE paper # (Batatia et al, MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields , Eq.10 and 11) # Authors: Ilyes Batatia -# This program is distributed under the ASL License (see ASL.md) +# This program is distributed under the MIT License (see MIT.md) ########################################################################################### from typing import Dict, Optional, Union diff --git a/models/schnet.py b/models/schnet.py new file mode 100644 index 0000000..67a21cd --- /dev/null +++ b/models/schnet.py @@ -0,0 +1,80 @@ +from typing import Optional + +import torch +from torch.nn import functional as F +from torch_geometric.nn import SchNet +from torch_geometric.nn import global_add_pool, global_mean_pool + + +class SchNetModel(SchNet): + """ + SchNet model from "Schnet - a deep learning architecture for molecules and materials". + + This class extends the SchNet base class for PyG. + """ + def __init__( + self, + hidden_channels: int = 128, + in_dim: int = 1, + out_dim: int = 1, + num_filters: int = 128, + num_layers: int = 6, + num_gaussians: int = 50, + cutoff: float = 10, + max_num_neighbors: int = 32, + pool: str = 'sum' + ): + """ + Initializes an instance of the SchNetModel class with the provided parameters. + + Parameters: + - hidden_channels (int): Number of channels in the hidden layers (default: 128) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - num_filters (int): Number of filters used in convolutional layers (default: 128) + - num_layers (int): Number of convolutional layers in the model (default: 6) + - num_gaussians (int): Number of Gaussian functions used for radial filters (default: 50) + - cutoff (float): Cutoff distance for interactions (default: 10) + - max_num_neighbors (int): Maximum number of neighboring atoms to consider (default: 32) + - pool (str): Global pooling method to be used (default: "sum") + """ + super().__init__( + hidden_channels, + num_filters, + num_layers, + num_gaussians, + cutoff, + interaction_graph=None, + max_num_neighbors=max_num_neighbors, + readout=pool, + dipole=False, + mean=None, + std=None, + atomref=None + ) + + # Global pooling/readout function + self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] + + # Overwrite atom embedding and final predictor + self.lin2 = torch.nn.Linear(hidden_channels // 2, out_dim) + + def forward(self, batch): + + h = self.embedding(batch.atoms) # (n,) -> (n, d) + + row, col = batch.edge_index + edge_weight = (batch.pos[row] - batch.pos[col]).norm(dim=-1) + edge_attr = self.distance_expansion(edge_weight) + + for interaction in self.interactions: + # # Message passing layer: (n, d) -> (n, d) + h = h + interaction(h, batch.edge_index, edge_weight, edge_attr) + + out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) + + out = self.lin1(out) + out = self.act(out) + out = self.lin2(out) # (batch_size, out_dim) + + return out diff --git a/models/spherenet.py b/models/spherenet.py new file mode 100644 index 0000000..4cb384c --- /dev/null +++ b/models/spherenet.py @@ -0,0 +1,110 @@ +from typing import Callable + +import torch +from torch.nn import functional as F +from torch_scatter import scatter + +from models.layers.spherenet_layer import * + + +class SphereNetModel(torch.nn.Module): + """ + SphereNet model from "Spherical Message Passing for 3D Molecular Graphs". + """ + def __init__( + self, + cutoff: float = 10, + num_layers: int = 4, + hidden_channels: int = 128, + in_dim: int = 1, + out_dim: int = 1, + int_emb_size: int = 64, + basis_emb_size_dist: int = 8, + basis_emb_size_angle: int = 8, + basis_emb_size_torsion: int = 8, + out_emb_channels: int = 128, + num_spherical: int = 7, + num_radial: int = 6, + envelope_exponent: int = 5, + num_before_skip: int = 1, + num_after_skip: int = 2, + num_output_layers: int = 2, + act: Callable = swish, + output_init: str = 'GlorotOrthogonal', + use_node_features: bool = True + ): + """ + Initializes an instance of the SphereNetModel class with the following parameters: + + Parameters: + - cutoff (int): Cutoff distance for interactions (default: 10) + - num_layers (int): Number of layers in the model (default: 4) + - hidden_channels (int): Number of channels in the hidden layers (default: 128) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - int_emb_size (int): Embedding size for interaction features (default: 64) + - basis_emb_size_dist (int): Embedding size for distance basis functions (default: 8) + - basis_emb_size_angle (int): Embedding size for angle basis functions (default: 8) + - basis_emb_size_torsion (int): Embedding size for torsion basis functions (default: 8) + - out_emb_channels (int): Number of channels in the output embeddings (default: 128) + - num_spherical (int): Number of spherical harmonics (default: 7) + - num_radial (int): Number of radial basis functions (default: 6) + - envelope_exponent (int): Exponent of the envelope function (default: 5) + - num_before_skip (int): Number of layers before the skip connections (default: 1) + - num_after_skip (int): Number of layers after the skip connections (default: 2) + - num_output_layers (int): Number of output layers (default: 2) + - act (function): Activation function (default: swish) + - output_init (str): Initialization method for the output layer (default: 'GlorotOrthogonal') + - use_node_features (bool): Whether to use node features (default: True) + """ + super().__init__() + + self.cutoff = cutoff + + self.init_e = init(num_radial, hidden_channels, act, use_node_features=use_node_features) + self.init_v = update_v(hidden_channels, out_emb_channels, out_dim, num_output_layers, act, output_init) + self.init_u = update_u() + self.emb = emb(num_spherical, num_radial, self.cutoff, envelope_exponent) + + self.update_vs = torch.nn.ModuleList([ + update_v(hidden_channels, out_emb_channels, out_dim, num_output_layers, act, output_init) for _ in range(num_layers)]) + + self.update_es = torch.nn.ModuleList([ + update_e(hidden_channels, int_emb_size, basis_emb_size_dist, basis_emb_size_angle, basis_emb_size_torsion, num_spherical, num_radial, num_before_skip, num_after_skip,act) for _ in range(num_layers)]) + + self.update_us = torch.nn.ModuleList([update_u() for _ in range(num_layers)]) + + self.reset_parameters() + + def reset_parameters(self): + self.init_e.reset_parameters() + self.init_v.reset_parameters() + self.emb.reset_parameters() + for update_e in self.update_es: + update_e.reset_parameters() + for update_v in self.update_vs: + update_v.reset_parameters() + + + def forward(self, batch_data): + z, pos, batch = batch_data.atoms, batch_data.pos, batch_data.batch + edge_index = batch_data.edge_index + num_nodes = z.size(0) + dist, angle, torsion, i, j, idx_kj, idx_ji = xyz_to_dat(pos, edge_index, num_nodes, use_torsion=True) + + emb = self.emb(dist, angle, torsion, idx_kj) + + # Initialize edge, node, graph features + e = self.init_e(z, emb, i, j) + v = self.init_v(e, i) + # Disable virutal node trick + # u = self.init_u(torch.zeros_like(scatter(v, batch, dim=0)), v, batch) + + for update_e, update_v, update_u in zip(self.update_es, self.update_vs, self.update_us): + e = update_e(e, emb, idx_kj, idx_ji) + v = update_v(e, i) + # Disable virutal node trick + # u = update_u(u, v, batch) + + out = scatter(v, batch, dim=0, reduce='add') + return out diff --git a/models/tfn.py b/models/tfn.py new file mode 100644 index 0000000..30ea2ec --- /dev/null +++ b/models/tfn.py @@ -0,0 +1,160 @@ +from typing import Optional + +import torch +from torch.nn import functional as F +from torch_geometric.nn import global_add_pool, global_mean_pool +import e3nn + +from models.mace_modules.blocks import RadialEmbeddingBlock +from models.layers.tfn_layer import TensorProductConvLayer + + +class TFNModel(torch.nn.Module): + """ + Tensor Field Network model from "Tensor Field Networks". + """ + def __init__( + self, + r_max: float = 10.0, + num_bessel: int = 8, + num_polynomial_cutoff: int = 5, + max_ell: int = 2, + num_layers: int = 5, + emb_dim: int = 64, + hidden_irreps: Optional[e3nn.o3.Irreps] = None, + mlp_dim: int = 256, + in_dim: int = 1, + out_dim: int = 1, + aggr: str = "sum", + pool: str = "sum", + gate: bool = True, + batch_norm: bool = False, + residual: bool = True, + equivariant_pred: bool = False + ): + """ + Parameters: + - r_max (float): Maximum distance for Bessel basis functions (default: 10.0) + - num_bessel (int): Number of Bessel basis functions (default: 8) + - num_polynomial_cutoff (int): Number of polynomial cutoff basis functions (default: 5) + - max_ell (int): Maximum degree of spherical harmonics basis functions (default: 2) + - num_layers (int): Number of layers in the model (default: 5) + - emb_dim (int): Scalar feature embedding dimension (default: 64) + - hidden_irreps (Optional[e3nn.o3.Irreps]): Hidden irreps (default: None) + - mlp_dim (int): Dimension of MLP for computing tensor product weights (default: 256) + - in_dim (int): Input dimension of the model (default: 1) + - out_dim (int): Output dimension of the model (default: 1) + - aggr (str): Aggregation method to be used (default: "sum") + - pool (str): Global pooling method to be used (default: "sum") + - gate (bool): Whether to use gated equivariant non-linearity (default: True) + - batch_norm (bool): Whether to use batch normalization (default: False) + - residual (bool): Whether to use residual connections (default: True) + - equivariant_pred (bool): Whether it is an equivariant prediction task (default: False) + + Note: + - If `hidden_irreps` is None, the irreps for the intermediate features are computed + using `emb_dim` and `max_ell`. + - The `equivariant_pred` parameter determines whether it is an equivariant prediction task. + If set to True, equivariant prediction will be performed. + - At present, only one of `gate` and `batch_norm` can be True. + """ + super().__init__() + + self.r_max = r_max + self.max_ell = max_ell + self.num_layers = num_layers + self.emb_dim = emb_dim + self.mlp_dim = mlp_dim + self.residual = residual + self.batch_norm = batch_norm + self.gate = gate + self.hidden_irreps = hidden_irreps + self.equivariant_pred = equivariant_pred + + # Edge embedding + self.radial_embedding = RadialEmbeddingBlock( + r_max=r_max, + num_bessel=num_bessel, + num_polynomial_cutoff=num_polynomial_cutoff, + ) + sh_irreps = e3nn.o3.Irreps.spherical_harmonics(max_ell) + self.spherical_harmonics = e3nn.o3.SphericalHarmonics( + sh_irreps, normalize=True, normalization="component" + ) + + # Embedding lookup for initial node features + self.emb_in = torch.nn.Embedding(in_dim, emb_dim) + + # Set hidden irreps if none are provided + if hidden_irreps is None: + hidden_irreps = (sh_irreps * emb_dim).sort()[0].simplify() + # Note: This defaults to O(3) equivariant layers + # It is possible to use SO(3) equivariance by passing the appropriate irreps + + self.convs = torch.nn.ModuleList() + # First conv layer: scalar only -> tensor + self.convs.append( + TensorProductConvLayer( + in_irreps=e3nn.o3.Irreps(f'{emb_dim}x0e'), + out_irreps=hidden_irreps, + sh_irreps=sh_irreps, + edge_feats_dim=self.radial_embedding.out_dim, + mlp_dim=mlp_dim, + aggr=aggr, + batch_norm=batch_norm, + gate=gate, + ) + ) + # Intermediate conv layers: tensor -> tensor + for _ in range(num_layers - 1): + conv = TensorProductConvLayer( + in_irreps=hidden_irreps, + out_irreps=hidden_irreps, + sh_irreps=sh_irreps, + edge_feats_dim=self.radial_embedding.out_dim, + mlp_dim=mlp_dim, + aggr=aggr, + batch_norm=batch_norm, + gate=gate, + ) + self.convs.append(conv) + + # Global pooling/readout function + self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] + + if self.equivariant_pred: + # Linear predictor for equivariant tasks using geometric features + self.pred = torch.nn.Linear(hidden_irreps.dim, out_dim) + else: + # MLP predictor for invariant tasks using only scalar features + self.pred = torch.nn.Sequential( + torch.nn.Linear(emb_dim, emb_dim), + torch.nn.ReLU(), + torch.nn.Linear(emb_dim, out_dim) + ) + + def forward(self, batch): + # Node embedding + h = self.emb_in(batch.atoms) # (n,) -> (n, d) + + # Edge features + vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] + lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] + + edge_sh = self.spherical_harmonics(vectors) + edge_feats = self.radial_embedding(lengths) + + for conv in self.convs: + # Message passing layer + h_update = conv(h, batch.edge_index, edge_sh, edge_feats) + + # Update node features + h = h_update + F.pad(h, (0, h_update.shape[-1] - h.shape[-1])) if self.residual else h_update + + out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) + + if not self.equivariant_pred: + # Select only scalars for invariant prediction + out = out[:,:self.emb_dim] + + return self.pred(out) # (batch_size, out_dim) diff --git a/src/models.py b/src/models.py deleted file mode 100644 index 0875fc9..0000000 --- a/src/models.py +++ /dev/null @@ -1,521 +0,0 @@ -from typing import Callable, Optional, Union -import torch -from torch.nn import functional as F -import torch_geometric -from torch_geometric.nn import SchNet, DimeNetPlusPlus, global_add_pool, global_mean_pool -import torch_scatter -from torch_scatter import scatter -from e3nn import o3 - -from src.modules.blocks import ( - EquivariantProductBasisBlock, - RadialEmbeddingBlock, -) -from src.modules.irreps_tools import reshape_irreps - -from src.egnn_layers import MPNNLayer, EGNNLayer -from src.tfn_layers import TensorProductConvLayer -import src.gvp_layers as gvp - - -class MACEModel(torch.nn.Module): - def __init__( - self, - r_max=10.0, - num_bessel=8, - num_polynomial_cutoff=5, - max_ell=2, - correlation=3, - num_layers=5, - emb_dim=64, - in_dim=1, - out_dim=1, - aggr="sum", - pool="sum", - residual=True, - scalar_pred=True - ): - super().__init__() - self.r_max = r_max - self.emb_dim = emb_dim - self.num_layers = num_layers - self.residual = residual - self.scalar_pred = scalar_pred - # Embedding - self.radial_embedding = RadialEmbeddingBlock( - r_max=r_max, - num_bessel=num_bessel, - num_polynomial_cutoff=num_polynomial_cutoff, - ) - sh_irreps = o3.Irreps.spherical_harmonics(max_ell) - self.spherical_harmonics = o3.SphericalHarmonics( - sh_irreps, normalize=True, normalization="component" - ) - - # Embedding lookup for initial node features - self.emb_in = torch.nn.Embedding(in_dim, emb_dim) - - self.convs = torch.nn.ModuleList() - self.prods = torch.nn.ModuleList() - self.reshapes = torch.nn.ModuleList() - hidden_irreps = (sh_irreps * emb_dim).sort()[0].simplify() - irrep_seq = [ - o3.Irreps(f'{emb_dim}x0e'), - # o3.Irreps(f'{emb_dim}x0e + {emb_dim}x1o + {emb_dim}x2e'), - # o3.Irreps(f'{emb_dim//2}x0e + {emb_dim//2}x0o + {emb_dim//2}x1e + {emb_dim//2}x1o + {emb_dim//2}x2e + {emb_dim//2}x2o'), - hidden_irreps - ] - for i in range(num_layers): - in_irreps = irrep_seq[min(i, len(irrep_seq) - 1)] - out_irreps = irrep_seq[min(i + 1, len(irrep_seq) - 1)] - conv = TensorProductConvLayer( - in_irreps=in_irreps, - out_irreps=out_irreps, - sh_irreps=sh_irreps, - edge_feats_dim=self.radial_embedding.out_dim, - hidden_dim=emb_dim, - gate=False, - aggr=aggr, - ) - self.convs.append(conv) - self.reshapes.append(reshape_irreps(out_irreps)) - prod = EquivariantProductBasisBlock( - node_feats_irreps=out_irreps, - target_irreps=out_irreps, - correlation=correlation, - element_dependent=False, - num_elements=in_dim, - use_sc=residual - ) - self.prods.append(prod) - - # Global pooling/readout function - self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] - - if self.scalar_pred: - # Predictor MLP - self.pred = torch.nn.Sequential( - torch.nn.Linear(emb_dim, emb_dim), - torch.nn.ReLU(), - torch.nn.Linear(emb_dim, out_dim) - ) - else: - self.pred = torch.nn.Linear(hidden_irreps.dim, out_dim) - - def forward(self, batch): - h = self.emb_in(batch.atoms) # (n,) -> (n, d) - - # Edge features - vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] - lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] - edge_attrs = self.spherical_harmonics(vectors) - edge_feats = self.radial_embedding(lengths) - - for conv, reshape, prod in zip(self.convs, self.reshapes, self.prods): - # Message passing layer - h_update = conv(h, batch.edge_index, edge_attrs, edge_feats) - # Update node features - sc = F.pad(h, (0, h_update.shape[-1] - h.shape[-1])) - h = prod(reshape(h_update), sc, None) - - if self.scalar_pred: - # Select only scalars for prediction - h = h[:,:self.emb_dim] - out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) - return self.pred(out) # (batch_size, out_dim) - - -class TFNModel(torch.nn.Module): - def __init__( - self, - r_max=10.0, - num_bessel=8, - num_polynomial_cutoff=5, - max_ell=2, - num_layers=5, - emb_dim=64, - in_dim=1, - out_dim=1, - aggr="sum", - pool="sum", - residual=True, - scalar_pred=True - ): - super().__init__() - self.r_max = r_max - self.emb_dim = emb_dim - self.num_layers = num_layers - self.residual = residual - self.scalar_pred = scalar_pred - # Embedding - self.radial_embedding = RadialEmbeddingBlock( - r_max=r_max, - num_bessel=num_bessel, - num_polynomial_cutoff=num_polynomial_cutoff, - ) - sh_irreps = o3.Irreps.spherical_harmonics(max_ell) - self.spherical_harmonics = o3.SphericalHarmonics( - sh_irreps, normalize=True, normalization="component" - ) - - # Embedding lookup for initial node features - self.emb_in = torch.nn.Embedding(in_dim, emb_dim) - - self.convs = torch.nn.ModuleList() - hidden_irreps = (sh_irreps * emb_dim).sort()[0].simplify() - irrep_seq = [ - o3.Irreps(f'{emb_dim}x0e'), - # o3.Irreps(f'{emb_dim}x0e + {emb_dim}x1o + {emb_dim}x2e'), - # o3.Irreps(f'{emb_dim//2}x0e + {emb_dim//2}x0o + {emb_dim//2}x1e + {emb_dim//2}x1o + {emb_dim//2}x2e + {emb_dim//2}x2o'), - hidden_irreps - ] - for i in range(num_layers): - in_irreps = irrep_seq[min(i, len(irrep_seq) - 1)] - out_irreps = irrep_seq[min(i + 1, len(irrep_seq) - 1)] - conv = TensorProductConvLayer( - in_irreps=in_irreps, - out_irreps=out_irreps, - sh_irreps=sh_irreps, - edge_feats_dim=self.radial_embedding.out_dim, - hidden_dim=emb_dim, - gate=True, - aggr=aggr, - ) - self.convs.append(conv) - - # Global pooling/readout function - self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] - - if self.scalar_pred: - # Predictor MLP - self.pred = torch.nn.Sequential( - torch.nn.Linear(emb_dim, emb_dim), - torch.nn.ReLU(), - torch.nn.Linear(emb_dim, out_dim) - ) - else: - self.pred = torch.nn.Linear(hidden_irreps.dim, out_dim) - - def forward(self, batch): - h = self.emb_in(batch.atoms) # (n,) -> (n, d) - - # Edge features - vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] - lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] - edge_attrs = self.spherical_harmonics(vectors) - edge_feats = self.radial_embedding(lengths) - - for conv in self.convs: - # Message passing layer - h_update = conv(h, batch.edge_index, edge_attrs, edge_feats) - - # Update node features - h = h_update + F.pad(h, (0, h_update.shape[-1] - h.shape[-1])) if self.residual else h_update - - if self.scalar_pred: - # Select only scalars for prediction - h = h[:,:self.emb_dim] - out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) - return self.pred(out) # (batch_size, out_dim) - - -class GVPGNNModel(torch.nn.Module): - def __init__( - self, - r_max=10.0, - num_bessel=8, - num_polynomial_cutoff=5, - num_layers=5, - emb_dim=64, - in_dim=1, - out_dim=1, - aggr="sum", - pool="sum", - residual=True - ): - super().__init__() - _DEFAULT_V_DIM = (emb_dim, emb_dim) - _DEFAULT_E_DIM = (emb_dim, 1) - activations = (F.relu, None) - - self.r_max = r_max - self.emb_dim = emb_dim - self.num_layers = num_layers - # Embedding - self.radial_embedding = RadialEmbeddingBlock( - r_max=r_max, - num_bessel=num_bessel, - num_polynomial_cutoff=num_polynomial_cutoff, - ) - self.emb_in = torch.nn.Embedding(in_dim, emb_dim) - self.W_e = torch.nn.Sequential( - gvp.LayerNorm((self.radial_embedding.out_dim, 1)), - gvp.GVP((self.radial_embedding.out_dim, 1), _DEFAULT_E_DIM, - activations=(None, None), vector_gate=True) - ) - self.W_v = torch.nn.Sequential( - gvp.LayerNorm((emb_dim, 0)), - gvp.GVP((emb_dim, 0), _DEFAULT_V_DIM, - activations=(None, None), vector_gate=True) - ) - - # Stack of GNN layers - self.layers = torch.nn.ModuleList( - gvp.GVPConvLayer(_DEFAULT_V_DIM, _DEFAULT_E_DIM, - activations=activations, vector_gate=True, - residual=residual) - for _ in range(num_layers)) - - self.W_out = torch.nn.Sequential( - gvp.LayerNorm(_DEFAULT_V_DIM), - gvp.GVP(_DEFAULT_V_DIM, (emb_dim, 0), - activations=activations, vector_gate=True) - ) - - # Global pooling/readout function - self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] - - # Predictor MLP - self.pred = torch.nn.Sequential( - torch.nn.Linear(emb_dim, emb_dim), - torch.nn.ReLU(), - torch.nn.Linear(emb_dim, out_dim) - ) - - def forward(self, batch): - - # Edge features - vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] - lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] - - h_V = self.emb_in(batch.atoms) # (n,) -> (n, d) - h_E = (self.radial_embedding(lengths), torch.nan_to_num(torch.div(vectors, lengths)).unsqueeze_(-2)) - - h_V = self.W_v(h_V) - h_E = self.W_e(h_E) - - for layer in self.layers: - h_V = layer(h_V, batch.edge_index, h_E) - - out = self.W_out(h_V) - - out = self.pool(out, batch.batch) # (n, d) -> (batch_size, d) - return self.pred(out) # (batch_size, out_dim) - - -class EGNNModel(torch.nn.Module): - def __init__( - self, - num_layers=5, - emb_dim=128, - in_dim=1, - out_dim=1, - activation="relu", - norm="layer", - aggr="sum", - pool="sum", - residual=True - ): - """E(n) Equivariant GNN model - - Args: - num_layers: (int) - number of message passing layers - emb_dim: (int) - hidden dimension - in_dim: (int) - initial node feature dimension - out_dim: (int) - output number of classes - activation: (str) - non-linearity within MLPs (swish/relu) - norm: (str) - normalisation layer (layer/batch) - aggr: (str) - aggregation function `\oplus` (sum/mean/max) - pool: (str) - global pooling function (sum/mean) - residual: (bool) - whether to use residual connections - """ - super().__init__() - - # Embedding lookup for initial node features - self.emb_in = torch.nn.Embedding(in_dim, emb_dim) - - # Stack of GNN layers - self.convs = torch.nn.ModuleList() - for layer in range(num_layers): - self.convs.append(EGNNLayer(emb_dim, activation, norm, aggr)) - - # Global pooling/readout function - self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] - - # Predictor MLP - self.pred = torch.nn.Sequential( - torch.nn.Linear(emb_dim, emb_dim), - torch.nn.ReLU(), - torch.nn.Linear(emb_dim, out_dim) - ) - self.residual = residual - - def forward(self, batch): - - h = self.emb_in(batch.atoms) # (n,) -> (n, d) - pos = batch.pos # (n, 3) - - for conv in self.convs: - # Message passing layer - h_update, pos_update = conv(h, pos, batch.edge_index) - - # Update node features (n, d) -> (n, d) - h = h + h_update if self.residual else h_update - - # Update node coordinates (no residual) (n, 3) -> (n, 3) - pos = pos_update - - out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) - return self.pred(out) # (batch_size, out_dim) - - -class MPNNModel(torch.nn.Module): - def __init__( - self, - num_layers=5, - emb_dim=128, - in_dim=1, - out_dim=1, - activation="relu", - norm="layer", - aggr="sum", - pool="sum", - residual=True - ): - """Vanilla Message Passing GNN model - - Args: - num_layers: (int) - number of message passing layers - emb_dim: (int) - hidden dimension - in_dim: (int) - initial node feature dimension - out_dim: (int) - output number of classes - activation: (str) - non-linearity within MLPs (swish/relu) - norm: (str) - normalisation layer (layer/batch) - aggr: (str) - aggregation function `\oplus` (sum/mean/max) - pool: (str) - global pooling function (sum/mean) - residual: (bool) - whether to use residual connections - """ - super().__init__() - - # Embedding lookup for initial node features - self.emb_in = torch.nn.Embedding(in_dim, emb_dim) - - # Stack of GNN layers - self.convs = torch.nn.ModuleList() - for layer in range(num_layers): - self.convs.append(MPNNLayer(emb_dim, activation, norm, aggr)) - - # Global pooling/readout function - self.pool = {"mean": global_mean_pool, "sum": global_add_pool}[pool] - - # Predictor MLP - self.pred = torch.nn.Sequential( - torch.nn.Linear(emb_dim, emb_dim), - torch.nn.ReLU(), - torch.nn.Linear(emb_dim, out_dim) - ) - self.residual = residual - - def forward(self, batch): - - h = self.emb_in(batch.atoms) # (n,) -> (n, d) - - for conv in self.convs: - # Message passing layer and residual connection - h = h + conv(h, batch.edge_index) if self.residual else conv(h, batch.edge_index) - - out = self.pool(h, batch.batch) # (n, d) -> (batch_size, d) - return self.pred(out) # (batch_size, out_dim) - - -class SchNetModel(SchNet): - def __init__( - self, - hidden_channels: int = 128, - in_dim: int = 1, - out_dim: int = 1, - num_filters: int = 128, - num_layers: int = 6, - num_gaussians: int = 50, - cutoff: float = 10, - max_num_neighbors: int = 32, - readout: str = 'add', - dipole: bool = False, - mean: Optional[float] = None, - std: Optional[float] = None, - atomref: Optional[torch.Tensor] = None, - ): - super().__init__(hidden_channels, num_filters, num_layers, num_gaussians, cutoff, max_num_neighbors, readout, dipole, mean, std, atomref) - - # Overwrite atom embedding and final predictor - self.lin2 = torch.nn.Linear(hidden_channels // 2, out_dim) - - def forward(self, batch): - h = self.embedding(batch.atoms) - - row, col = batch.edge_index - edge_weight = (batch.pos[row] - batch.pos[col]).norm(dim=-1) - edge_attr = self.distance_expansion(edge_weight) - - for interaction in self.interactions: - h = h + interaction(h, batch.edge_index, edge_weight, edge_attr) - - h = self.lin1(h) - h = self.act(h) - h = self.lin2(h) - - out = scatter(h, batch.batch, dim=0, reduce=self.readout) - return out - - -class DimeNetPPModel(DimeNetPlusPlus): - def __init__( - self, - hidden_channels: int = 128, - in_dim: int = 1, - out_dim: int = 1, - num_layers: 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, - cutoff: float = 10, - max_num_neighbors: int = 32, - envelope_exponent: int = 5, - num_before_skip: int = 1, - num_after_skip: int = 2, - num_output_layers: int = 3, - act: Union[str, Callable] = 'swish' - ): - super().__init__(hidden_channels, out_dim, num_layers, int_emb_size, basis_emb_size, out_emb_channels, num_spherical, num_radial, cutoff, max_num_neighbors, envelope_exponent, num_before_skip, num_after_skip, num_output_layers, act) - - def forward(self, batch): - - i, j, idx_i, idx_j, idx_k, idx_kj, idx_ji = self.triplets( - batch.edge_index, num_nodes=batch.atoms.size(0)) - - # Calculate distances. - dist = (batch.pos[i] - batch.pos[j]).pow(2).sum(dim=-1).sqrt() - - # Calculate angles. - pos_i = batch.pos[idx_i] - pos_ji, pos_ki = batch.pos[idx_j] - pos_i, batch.pos[idx_k] - pos_i - a = (pos_ji * pos_ki).sum(dim=-1) - b = torch.cross(pos_ji, pos_ki).norm(dim=-1) - angle = torch.atan2(b, a) - - rbf = self.rbf(dist) - sbf = self.sbf(dist, angle, idx_kj) - - # Embedding block. - x = self.emb(batch.atoms, rbf, i, j) - P = self.output_blocks[0](x, rbf, i, num_nodes=batch.pos.size(0)) - - # Interaction blocks. - for interaction_block, output_block in zip(self.interaction_blocks, - self.output_blocks[1:]): - x = interaction_block(x, rbf, sbf, idx_kj, idx_ji) - P += output_block(x, rbf, i) - - return P.sum(dim=0) if batch is None else scatter(P, batch.batch, dim=0) diff --git a/src/modules/model.py b/src/modules/model.py deleted file mode 100644 index 4595b52..0000000 --- a/src/modules/model.py +++ /dev/null @@ -1,171 +0,0 @@ -from typing import Callable, Optional, Type -import torch -from torch_scatter import scatter -from e3nn import o3 - -from src.modules.blocks import ( - EquivariantProductBasisBlock, - InteractionBlock, - LinearNodeEmbeddingBlock, - LinearReadoutBlock, - NonLinearReadoutBlock, - RadialEmbeddingBlock, -) -from src.modules import ( - interaction_classes, - gate_dict -) - - -class OriginalMACEModel(torch.nn.Module): - def __init__( - self, - r_max: float = 10.0, - num_bessel: int = 8, - num_polynomial_cutoff: int = 5, - max_ell: int = 2, - interaction_cls: Type[InteractionBlock] = interaction_classes["RealAgnosticResidualInteractionBlock"], - interaction_cls_first: Type[InteractionBlock] = interaction_classes["RealAgnosticInteractionBlock"], - num_interactions: int = 2, - num_elements: int = 1, - hidden_irreps: o3.Irreps = o3.Irreps("64x0e + 64x1o + 64x2e"), - MLP_irreps: o3.Irreps = o3.Irreps("64x0e"), - irreps_out: o3.Irreps = o3.Irreps("1x0e"), - avg_num_neighbors: int = 1, - correlation: int = 3, - gate: Optional[Callable] = gate_dict["silu"], - num_layers=2, - in_dim=1, - out_dim=1, - ): - super().__init__() - self.r_max = r_max - self.num_elements = num_elements - # Embedding - node_attr_irreps = o3.Irreps([(num_elements, (0, 1))]) - node_feats_irreps = o3.Irreps([(hidden_irreps.count(o3.Irrep(0, 1)), (0, 1))]) - self.node_embedding = LinearNodeEmbeddingBlock( - irreps_in=node_attr_irreps, irreps_out=node_feats_irreps - ) - self.radial_embedding = RadialEmbeddingBlock( - r_max=r_max, - num_bessel=num_bessel, - num_polynomial_cutoff=num_polynomial_cutoff, - ) - edge_feats_irreps = o3.Irreps(f"{self.radial_embedding.out_dim}x0e") - - sh_irreps = o3.Irreps.spherical_harmonics(max_ell) - num_features = hidden_irreps.count(o3.Irrep(0, 1)) - interaction_irreps = (sh_irreps * num_features).sort()[0].simplify() - self.spherical_harmonics = o3.SphericalHarmonics( - sh_irreps, normalize=True, normalization="component" - ) - - # Interactions and readout - self.atomic_energies_fn = LinearReadoutBlock(node_feats_irreps, irreps_out) - - inter = interaction_cls_first( - node_attrs_irreps=node_attr_irreps, - node_feats_irreps=node_feats_irreps, - edge_attrs_irreps=sh_irreps, - edge_feats_irreps=edge_feats_irreps, - target_irreps=interaction_irreps, - hidden_irreps=hidden_irreps, - avg_num_neighbors=avg_num_neighbors, - ) - self.interactions = torch.nn.ModuleList([inter]) - - # Use the appropriate self connection at the first layer for proper E0 - use_sc_first = False - if "Residual" in str(interaction_cls_first): - use_sc_first = True - - node_feats_irreps_out = inter.target_irreps - prod = EquivariantProductBasisBlock( - node_feats_irreps=node_feats_irreps_out, - target_irreps=hidden_irreps, - correlation=correlation, - element_dependent=True, - num_elements=num_elements, - use_sc=use_sc_first, - ) - self.products = torch.nn.ModuleList([prod]) - - self.readouts = torch.nn.ModuleList() - self.readouts.append(LinearReadoutBlock(hidden_irreps, irreps_out)) - - for i in range(num_interactions - 1): - if i == num_interactions - 2: - hidden_irreps_out = str( - hidden_irreps[0] - ) # Select only scalars for last layer - else: - hidden_irreps_out = hidden_irreps - inter = interaction_cls( - node_attrs_irreps=node_attr_irreps, - node_feats_irreps=hidden_irreps, - edge_attrs_irreps=sh_irreps, - edge_feats_irreps=edge_feats_irreps, - target_irreps=interaction_irreps, - hidden_irreps=hidden_irreps_out, - avg_num_neighbors=avg_num_neighbors, - ) - self.interactions.append(inter) - prod = EquivariantProductBasisBlock( - node_feats_irreps=interaction_irreps, - target_irreps=hidden_irreps_out, - correlation=correlation, - element_dependent=True, - num_elements=num_elements, - use_sc=True - ) - self.products.append(prod) - if i == num_interactions - 2: - self.readouts.append( - NonLinearReadoutBlock(hidden_irreps_out, MLP_irreps, gate, irreps_out) - ) - else: - self.readouts.append(LinearReadoutBlock(hidden_irreps, irreps_out)) - - def forward(self, batch): - # MACE expects one-hot-ified input - batch.atoms.unsqueeze_(-1) - shape = batch.atoms.shape[:-1] + (self.num_elements,) - node_attrs = torch.zeros(shape, device=batch.atoms.device).view(shape) - node_attrs.scatter_(dim=-1, index=batch.atoms, value=1) - - # Node embeddings - node_feats = self.node_embedding(node_attrs) - node_e0 = self.atomic_energies_fn(node_feats) - e0 = scatter(node_e0, batch.batch, dim=0, reduce="sum") # [n_graphs, irreps_out] - - # Edge features - vectors = batch.pos[batch.edge_index[0]] - batch.pos[batch.edge_index[1]] # [n_edges, 3] - lengths = torch.linalg.norm(vectors, dim=-1, keepdim=True) # [n_edges, 1] - edge_attrs = self.spherical_harmonics(vectors) - edge_feats = self.radial_embedding(lengths) - - # Interactions - energies = [e0] - for interaction, product, readout in zip( - self.interactions, self.products, self.readouts - ): - node_feats, sc = interaction( - node_attrs=node_attrs, - node_feats=node_feats, - edge_attrs=edge_attrs, - edge_feats=edge_feats, - edge_index=batch.edge_index, - ) - node_feats = product( - node_feats=node_feats, sc=sc, node_attrs=node_attrs - ) - node_energies = readout(node_feats).squeeze(-1) # [n_nodes, irreps_out] - energy = scatter(node_energies, batch.batch, dim=0, reduce="sum") # [n_graphs, irreps_out] - energies.append(energy) - - # Sum over energy contributions - contributions = torch.stack(energies, dim=-1) - total_energy = torch.sum(contributions, dim=-1) # [n_graphs, irreps_out] - - return total_energy