API Docs

.github/workflows/gh-pages.yml

@@ -0,0 +1,25 @@
+name: GitHub Pages
+
+on:
+  pull_request:
+    branches: [main]
+  push:
+    branches: [main]
+  workflow_dispatch:
+
+# set permissions of GITHUB_TOKEN to allow deployment to GitHub Pages
+permissions:
+  contents: read
+  pages: write
+  id-token: write
+
+jobs:
+  build:
+    uses: janosh/workflows/.github/workflows/nodejs-gh-pages.yml@main
+    with:
+      python-version: "3.11"
+      working-directory: site
+      install-cmd: npm install --force
+      pre-build: |
+        pip install '..[docs]'
+        python make_docs.py

.gitignore

@@ -6,49 +6,15 @@ __pycache__/
 *.so
 
 # Distribution / packaging
-bin/
 build/
-develop-eggs/
 dist/
-eggs/
-lib/
-lib64/
-parts/
 sdist/
-var/
 *.egg-info/
-.installed.cfg
-*.egg
-
-# Installer logs
-pip-log.txt
-pip-delete-this-directory.txt
 
 # Unit test / coverage reports
-.tox/
 .coverage
-.cache
-nosetests.xml
 coverage.xml
 
-# Translations
-*.mo
-
-# Mr Developer
-.mr.developer.cfg
-.project
-.pydevproject
-
-# Rope
-.ropeproject
-
-# Django stuff:
-*.log
-*.pot
-
-# Sphinx documentation
-docs/_build/
-
 # Extras
 *.traj
 *.xml

.pre-commit-config.yaml

@@ -43,3 +43,27 @@ repos:
     hooks:
       - id: nbstripout
         args: [--drop-empty-cells, --keep-output]
+
+  - repo: https://github.com/pre-commit/mirrors-prettier
+    rev: v3.0.0-alpha.6
+    hooks:
+      - id: prettier
+        args: [--write] # edit files in-place
+        additional_dependencies:
+          - prettier
+          - prettier-plugin-svelte
+          - svelte
+
+  - repo: https://github.com/pre-commit/mirrors-eslint
+    rev: v8.41.0
+    hooks:
+      - id: eslint
+        types: [file]
+        args: [--fix, --plugin, "svelte3, @typescript-eslint"]
+        files: \.(js|ts|svelte)$
+        additional_dependencies:
+          - eslint
+          - typescript
+          - eslint-plugin-svelte3
+          - "@typescript-eslint/eslint-plugin"
+          - "@typescript-eslint/parser"

README.md

@@ -13,7 +13,7 @@
 
 A pretrained universal neural network potential for
 **charge**-informed atomistic modeling
-![chgnet](chgnet-logo.png)
+![chgnet](https://raw.github.com/CederGroupHub/chgnet/main/site/static/chgnet-logo.png)
 **C**rystal **H**amiltonian **G**raph neural **Net**work is pretrained on the GGA/GGA+U static and relaxation trajectories from Materials Project,
 a comprehensive dataset consisting of 1.5 Million structures from 146k compounds spanning the whole periodic table.
 
@@ -25,7 +25,7 @@ DFT magnetic moments, which carry rich information about both local ionic enviro
 
 | Notebooks                                                                                                                | Links                                                                                                           | Descriptions                                                                                                                        |
 | ------------------------------------------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------- |
-| [**CHGNet Basics**](https://github.com/CederGroupHub/chgnet/blob/main/examples/basics.ipynb)                     | [![Open in Google Colab]](https://colab.research.google.com/github/CederGroupHub/chgnet/blob/main/examples/basics.ipynb)             | Examples for loading pre-trained CHGNet, predicting energy, force, stress, magmom as well as running structure optimization and MD. |
+| [**CHGNet Basics**](https://github.com/CederGroupHub/chgnet/blob/main/examples/basics.ipynb)                             | [![Open in Google Colab]](https://colab.research.google.com/github/CederGroupHub/chgnet/blob/main/examples/basics.ipynb)                      | Examples for loading pre-trained CHGNet, predicting energy, force, stress, magmom as well as running structure optimization and MD. |
 | [**Tuning CHGNet**](https://github.com/CederGroupHub/chgnet/blob/main/examples/fine_tuning.ipynb)                        | [![Open in Google Colab]](https://colab.research.google.com/github/CederGroupHub/chgnet/blob/main/examples/fine_tuning.ipynb)                 | Examples of fine tuning the pretrained CHGNet to your system of interest.                                                           |
 | [**Visualize Relaxation**](https://github.com/CederGroupHub/chgnet/blob/main/examples/crystaltoolkit_relax_viewer.ipynb) | [![Open in Google Colab]](https://colab.research.google.com/github/CederGroupHub/chgnet/blob/main/examples/crystaltoolkit_relax_viewer.ipynb) | Crystal Toolkit that visualizes atom positions, energies and forces of a structure during CHGNet relaxation.                        |
 
@@ -39,6 +39,8 @@ You can install `chgnet` through `pip`:
 pip install chgnet
 ```
 
+<slot name='docs' />
+
 ## Usage
 
 ### Direct Inference (Static Calculation)
@@ -152,15 +154,15 @@ trainer.train(train_loader, val_loader, test_loader)
 
 1. The energy used for training should be energy/atom if you're fine-tuning the pretrained `CHGNet`.
 2. The pretrained dataset of `CHGNet` comes from GGA+U DFT with [`MaterialsProject2020Compatibility`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/entries/compatibility.py#L826-L1102).
-The parameter for VASP is described in [`MPRelaxSet`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/io/vasp/sets.py#L862-L879).
-If you're fine-tuning with [`MPRelaxSet`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/io/vasp/sets.py#L862-L879), it is recommended to apply the [`MP2020`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/entries/compatibility.py#L826-L1102)
-compatibility to your energy labels so that they're consistent with the pretrained dataset.
+   The parameter for VASP is described in [`MPRelaxSet`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/io/vasp/sets.py#L862-L879).
+   If you're fine-tuning with [`MPRelaxSet`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/io/vasp/sets.py#L862-L879), it is recommended to apply the [`MP2020`](https://github.com/materialsproject/pymatgen/blob/v2023.2.28/pymatgen/entries/compatibility.py#L826-L1102)
+   compatibility to your energy labels so that they're consistent with the pretrained dataset.
 3. If you're fine-tuning to functionals other than GGA, we recommend you refit the [`AtomRef`](https://github.com/CederGroupHub/chgnet/blob/main/chgnet/model/composition_model.py).
 4. `CHGNet` stress is in unit GPa, and the unit conversion has already been included in
-[`dataset.py`](https://github.com/CederGroupHub/chgnet/blob/main/chgnet/data/dataset.py). So `VASP` stress can be directly fed to `StructureData`
+   [`dataset.py`](https://github.com/CederGroupHub/chgnet/blob/main/chgnet/data/dataset.py). So `VASP` stress can be directly fed to `StructureData`
 5. To save time from graph conversion step for each training, we recommend you use [`GraphData`](https://github.com/CederGroupHub/chgnet/blob/main/chgnet/data/dataset.py) defined in
-[`dataset.py`](https://github.com/CederGroupHub/chgnet/blob/main/chgnet/data/dataset.py), which reads graphs directly from saved directory. To create saved graphs,
-see [`examples/make_graphs.py`](https://github.com/CederGroupHub/chgnet/blob/main/examples/make_graphs.py).
+   [`dataset.py`](https://github.com/CederGroupHub/chgnet/blob/main/chgnet/data/dataset.py), which reads graphs directly from saved directory. To create saved graphs,
+   see [`examples/make_graphs.py`](https://github.com/CederGroupHub/chgnet/blob/main/examples/make_graphs.py).
 6. Apple’s Metal Performance Shaders `MPS` is currently disabled until a stable version of `pytorch` for `MPS` is released.
 
 ## Reference

chgnet/data/dataset.py

@@ -669,7 +669,7 @@ def get_train_val_test_loader(
 
 
 def collate_graphs(batch_data: list):
-    """Collate of list of (graph, target) into batch data,.
+    """Collate of list of (graph, target) into batch data.
 
     Args:
         batch_data (list): list of (graph, target(dict))

chgnet/model/dynamics.py

@@ -71,16 +71,14 @@ def __init__(
 
         # mps is disabled before stable version of pytorch on apple mps is released
         if use_device == "mps":
-            raise NotImplementedError("mps is not supported yet")
+            raise NotImplementedError("'mps' backend is not supported yet")
         # elif torch.backends.mps.is_available():
         #     self.device = 'mps'
         # Determine the device to use
         self.device = use_device or ("cuda" if torch.cuda.is_available() else "cpu")
 
         # Move the model to the specified device
-        if model is None:
-            model = CHGNet.load()
-        self.model = model.to(self.device)
+        self.model = (model or CHGNet.load()).to(self.device)
         self.stress_weight = stress_weight
         print(f"CHGNet will run on {self.device}")
 

chgnet/utils/vasp_utils.py

@@ -134,17 +134,17 @@ def solve_charge_by_mag(
     Args:
         structure: input pymatgen structure
         default_ox (dict[str, float]): default oxidation state for elements.
-            Default = {"Li": 1, "O": -2}
-        ox_ranges (dict[str, dict[tuple[float, float], int]]): user defined range to
+            Default = dict(Li=1, O=-2)
+        ox_ranges (dict[str, dict[tuple[float, float], int]]): user-defined range to
             convert magmoms into formal valence.
             Example for Mn (Default):
-                {"Mn": {
-                (0.5, 1.5): 2,
-                (1.5, 2.5): 3,
-                (2.5, 3.5): 4,
-                (3.5, 4.2): 3,
-                (4.2, 5): 2
-                }}
+                ("Mn": (
+                    (0.5, 1.5): 2,
+                    (1.5, 2.5): 3,
+                    (2.5, 3.5): 4,
+                    (3.5, 4.2): 3,
+                    (4.2, 5): 2
+                ))
     """
     ox_list = []
     solved_ox = True
@@ -153,17 +153,15 @@ def solve_charge_by_mag(
         "Mn": {(0.5, 1.5): 2, (1.5, 2.5): 3, (2.5, 3.5): 4, (3.5, 4.2): 3, (4.2, 5): 2}
     }
 
-    mag_key = (
-        "final_magmom" if "final_magmom" in structure.site_properties else "magmom"
+    mag = structure.site_properties.get(
+        "final_magmom", structure.site_properties.get("magmom")
     )
 
-    mag = structure.site_properties[mag_key]
-
-    for site_i, site in enumerate(structure.sites):
+    for idx, site in enumerate(structure):
         assigned = False
         if site.species_string in ox_ranges:
             for (minmag, maxmag), magox in ox_ranges[site.species_string].items():
-                if mag[site_i] >= minmag and mag[site_i] < maxmag:
+                if mag[idx] >= minmag and mag[idx] < maxmag:
                     ox_list.append(magox)
                     assigned = True
                     break

pyproject.toml

@@ -33,6 +33,7 @@ classifiers = [
 test = ["pytest", "pytest-cov"]
 # needed to run interactive example notebooks
 crystal-toolkit = ["crystal-toolkit"]
+docs = ["lazydocs"]
 
 [project.urls]
 Source = "https://github.com/CederGroupHub/chgnet"

site/.eslintrc.yml

@@ -0,0 +1,28 @@
+root: true
+env:
+  browser: true
+  es2020: true
+  node: true
+parser: '@typescript-eslint/parser'
+parserOptions:
+  sourceType: module
+  ecmaVersion: latest
+plugins: [svelte3, '@typescript-eslint']
+extends:
+  - eslint:recommended
+  - plugin:@typescript-eslint/recommended
+overrides:
+  - files: ['*.svelte']
+    processor: svelte3/svelte3
+settings:
+  svelte3/typescript: true
+rules:
+  indent: [error, 2, SwitchCase: 1]
+  '@typescript-eslint/quotes': [error, backtick, avoidEscape: true]
+  semi: [error, never]
+  linebreak-style: [error, unix]
+  no-console: [error, allow: [warn, error]]
+  no-var: error
+  # allow triple slash for typescript file referencing https://git.io/JCeqO
+  spaced-comment: [error, always, { markers: [/] }]
+  '@typescript-eslint/no-inferrable-types': off

site/.gitignore

@@ -0,0 +1,7 @@
+# dependencies
+node_modules
+
+# production build
+.svelte-kit
+build
+src/routes/api/*.md

site/make_docs.py

@@ -0,0 +1,43 @@
+"""This script auto-generates markdown files from Python docstrings using lazydocs
+and tweaks the output for
+- prettier badges linking to source code on GitHub
+- remove bold tags since they break inline code.
+"""
+from __future__ import annotations
+
+import json
+import os
+import subprocess
+from glob import glob
+
+ROOT = os.path.dirname(os.path.dirname(__file__))
+os.chdir(ROOT)
+with open("site/package.json") as file:
+    pkg = json.load(file)
+route = "site/src/routes/api"
+
+for path in glob(f"{route}/*.md"):
+    os.remove(path)
+
+subprocess.run(
+    f"lazydocs {pkg['name']} --output-path {route} "
+    f"--no-watermark --src-base-url {pkg['repository']}/blob/main",
+    shell=True,
+)
+
+for path in glob(f"{route}/*.md"):
+    with open(path) as file:
+        markdown = file.read()
+    # remove <b> tags from generated markdown as they break inline code
+    markdown = markdown.replace("<b>", "").replace("</b>", "")
+    # improve style of badges linking to source code on GitHub
+    markdown = markdown.replace(
+        'src="https://img.shields.io/badge/-source-cccccc?style=flat-square"',
+        'src="https://img.shields.io/badge/source-blue?style=flat" alt="source link"',
+    )
+    # remove "Global Variables" section if only contains TYPE_CHECKING
+    markdown = markdown.replace(
+        "\n**Global Variables**\n---------------\n- **TYPE_CHECKING**\n\n", ""
+    )
+    with open(path, "w") as file:
+        file.write(markdown)

site/package.json

@@ -0,0 +1,54 @@
+{
+  "name": "chgnet",
+  "description": "Universal neural network potential for charge-informed atomistic modeling",
+  "author": "Deng et al.",
+  "homepage": "https://CederGroupHub.github.io/chgnet",
+  "repository": "https://github.com/CederGroupHub/chgnet",
+  "license": "MIT",
+  "type": "module",
+  "bugs": "https://github.com/CederGroupHub/chgnet/issues",
+  "scripts": {
+    "dev": "vite dev",
+    "build": "vite build",
+    "preview": "vite preview",
+    "serve": "vite build && vite preview",
+    "changelog": "npx auto-changelog --package --output ../changelog.md --unreleased-only --hide-credit --commit-limit false"
+  },
+  "devDependencies": {
+    "@sveltejs/adapter-static": "^2.0.2",
+    "@sveltejs/kit": "^1.20.5",
+    "@sveltejs/vite-plugin-svelte": "^2.4.2",
+    "@typescript-eslint/eslint-plugin": "^5.60.1",
+    "@typescript-eslint/parser": "^5.60.1",
+    "eslint": "^8.43.0",
+    "eslint-plugin-svelte3": "^4.0.0",
+    "hastscript": "^7.2.0",
+    "mdsvex": "^0.11.0",
+    "prettier": "^2.8.8",
+    "prettier-plugin-svelte": "^2.10.1",
+    "rehype-autolink-headings": "^6.1.1",
+    "rehype-slug": "^5.1.0",
+    "svelte": "^4.0.0",
+    "svelte-check": "^3.4.4",
+    "svelte-multiselect": "^9.0.0",
+    "svelte-preprocess": "^5.0.4",
+    "svelte-toc": "^0.5.5",
+    "svelte-zoo": "^0.4.8",
+    "svelte2tsx": "^0.6.16",
+    "tslib": "^2.5.3",
+    "typescript": "^5.1.3",
+    "vite": "^4.3.9"
+  },
+  "prettier": {
+    "semi": false,
+    "singleQuote": true,
+    "overrides": [
+      {
+        "files": "*.svelte",
+        "options": {
+          "printWidth": 90
+        }
+      }
+    ]
+  }
+}