Merge branch 'main' of github.com:materialsvirtuallab/matgl

.github/workflows/testing.yml

@@ -20,7 +20,7 @@ jobs:
     strategy:
       max-parallel: 20
       matrix:
-        python-version: ["3.9", "3.11"]
+        python-version: ["3.10", "3.12"]
         os: [ "ubuntu-latest", "macos-latest" ]
 
     runs-on: ${{ matrix.os }}

README.md

@@ -28,6 +28,7 @@ Nassar, Carmelo Gonzales).
 
 Major milestones are summarized below. Please refer to the [changelog] for details.
 
+- v1.1.0 (May 7 2024): Implementation of [CHGNet] + pre-trained models.
 - v1.0.0 (Feb 14 2024): Implementation of [TensorNet] and [SO3Net].
 - v0.5.1 (Jun 9 2023): Model versioning implemented.
 - v0.5.0 (Jun 8 2023): Simplified saving and loading of models. Now models can be loaded with one line of code!
@@ -87,6 +88,8 @@ We have implemented other models in matgl as well. A non-exhaustive list is give
 - [TensorNet], an O(3)-equivariant message-passing neural network architecture that
   leverages Cartesian tensor representations.
 - [SO3Net],  a minimalist SO(3)-equivariant neural network.
+- [CHGNet],  an invariant message-passing neural network architecture that can predict
+  PES properties and magmoms.
 
 ## Installation
 
@@ -226,6 +229,21 @@ information. If you are using any of the pretrained models, please cite the rele
 > Chen, C., Ong, S.P. *A universal graph deep learning interatomic potential for the periodic table.* Nature
 > Computational Science, 2023, 2, 718–728. DOI: [10.1038/s43588-022-00349-3][m3gnet].
 
+>**CHGNet**
+>
+> Deng, B., Zhong, P., Jun, K. et al. *CHGNet: as a pretrained universal neural network potential for charge-informed atomistic modelling.*
+> Nat Mach Intell 5, 1031–1041 (2023). DOI:[10.1038/s42256-023-00716-3][chgnet]
+
+>**TensorNet**
+>
+> Simeon, G.  De Fabritiis, G. *Tensornet: Cartesian tensor representations for efficient learning of molecular potentials.*
+> Adv. Neural Info. Process. Syst. 36, (2024). DOI: [10.48550/arXiv.2306.06482][tensornet]
+
+>**SO3Net**
+>
+> Schütt, K. T., Hessmann, S. S. P., Gebauer, N. W. A., Lederer, J., Gastegger, M. *SchNetPack 2.0: A neural network toolbox for atomistic machine learning.*
+> J. Chem. Phys. 158, 144801 (2023). DOI: [10.1063/5.0138367][so3net]
+
 ## FAQs
 
 1. **The `M3GNet-MP-2021.2.8-PES` differs from the original TensorFlow (TF) implementation!**
@@ -296,3 +314,4 @@ ACI-1548562.
 [tutorials]: https://matgl.ai/tutorials "Tutorials"
 [tensornet]: https://arxiv.org/abs/2306.06482 "TensorNet"
 [so3net]: https://pubs.aip.org/aip/jcp/article-abstract/158/14/144801/2877924/SchNetPack-2-0-A-neural-network-toolbox-for "SO3Net"
+[chgnet]: https://www.nature.com/articles/s42256-023-00716-3 "CHGNet"

docs/Gemfile.lock

@@ -224,8 +224,7 @@ GEM
     rb-fsevent (0.11.2)
     rb-inotify (0.10.1)
       ffi (~> 1.0)
-    rexml (3.3.3)
-      strscan
+    rexml (3.3.9)
     rouge (3.26.0)
     ruby2_keywords (0.0.5)
     rubyzip (2.3.2)
@@ -240,7 +239,6 @@ GEM
       faraday (>= 0.17.3, < 3)
     simpleidn (0.2.1)
       unf (~> 0.1.4)
-    strscan (3.1.0)
     terminal-table (1.8.0)
       unicode-display_width (~> 1.1, >= 1.1.1)
     typhoeus (1.4.0)
@@ -251,7 +249,7 @@ GEM
       unf_ext
     unf_ext (0.0.8.2)
     unicode-display_width (1.8.0)
-    webrick (1.8.1)
+    webrick (1.8.2)
 
 PLATFORMS
   arm64-darwin-22

docs/index.md

@@ -304,3 +304,4 @@ ACI-1548562.
 [tutorials]: https://matgl.ai/tutorials "Tutorials"
 [tensornet]: https://arxiv.org/abs/2306.06482 "TensorNet"
 [so3net]: https://pubs.aip.org/aip/jcp/article-abstract/158/14/144801/2877924/SchNetPack-2-0-A-neural-network-toolbox-for "SO3Net"
+[chgnet]: https://www.nature.com/articles/s42256-023-00716-3 "CHGNet"

examples/Relaxations and Simulations using the M3GNet Universal Potential.ipynb

@@ -7,7 +7,7 @@
    "source": [
     "# Introduction\n",
     "\n",
-    "This notebook demonstrates the use of the pre-trained M3GNet model to perform structural relaxations, molecular dynamics simulations and single-point calculations.\n",
+    "This notebook demonstrates the use of the pre-trained universal potentials to perform structural relaxations, molecular dynamics simulations and single-point calculations.\n",
     "\n",
     "Author: Tsz Wai Ko (Kenko)\n",
     "Email: [email protected]"
@@ -42,7 +42,7 @@
    "source": [
     "# Loading the pre-trained M3GNet PES model\n",
     "\n",
-    "We will first load the M3GNet PES model, which is trained on the MP-2021.2.8 dataset. This can be done with a single line of code."
+    "We will first load the M3GNet PES model, which is trained on the MP-2021.2.8 dataset. This can be done with a single line of code. Here we only use M3GNet for demonstration and users can choose other available models."
    ]
   },
   {
@@ -52,7 +52,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "pot = matgl.load_model(\"M3GNet-MP-2021.2.8-PES\")"
+    "# You can load any pretrained potentials such as CHGNet ('CHGNet-MPtrj-2023.12.1-PES-2.7M', 'CHGNet-MPtrj-2024.2.13-PES-11M')\n",
+    "# To see available models, use get_available_pretrained_models()\n",
+    "pot = matgl.load_model(\"M3GNet-MP-2021.2.8-PES\") "
    ]
   },
   {
@@ -120,7 +122,7 @@
    "source": [
     "# Single point energy calculation\n",
     "\n",
-    "Perform a single-point calculation for final structure using M3GNetCalculator."
+    "Perform a single-point calculation for final structure using PESCalculator."
    ]
   },
   {
@@ -154,7 +156,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.14"
+   "version": "3.10.9"
   }
  },
  "nbformat": 4,

pyproject.toml

@@ -50,14 +50,14 @@ classifiers = [
 ]
 dependencies = [
     "ase",
-    "dgl<=2.1.0",
+    "dgl<=2.2.1",
     "pymatgen",
     "lightning<=2.4.0",
     "torch<=2.2.1",
     "pydantic",
     "torchdata<=0.7.1",
     "boto3",
-    "numpy<2.0.0"
+    "numpy<3.0.0"
 ]
 version = "1.1.3"
 
@@ -71,6 +71,7 @@ mgl = "matgl.cli:main"
 where = ["src"]
 
 [tool.setuptools.package-data]
+matgl = ["py.typed"]
 "matgl.utils" = ["sb_roots.npy"]
 
 

requirements.txt

@@ -1,10 +1,11 @@
-dgl==2.1.0
-torch==2.4.0
+dgl==2.2.1
+torch==2.2.1
 lightning==2.4.0
-pymatgen==2024.8.9
+pymatgen==2024.10.29
 ase==3.23.0
-pydantic==2.8.2
-torchdata==0.8.0
-boto3==1.34.160
-numpy==1.26.4
-sympy==1.13.2
+pydantic==2.9.2
+numpy==2.1.3
+torchdata==0.7.1
+boto3==1.35.54
+sympy==1.13.3
+