amjax 0.0.1__tar.gz

amjax-0.0.1/.github/workflows/release.yml

@@ -0,0 +1,23 @@
+name: Release
+
+on:
+  push:
+    branches:
+      - main
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Release
+        uses: patrick-kidger/action_update_python_project@v8
+        with:
+            python-version: "3.11"
+            test-script: |
+                cp -r ${{ github.workspace }}/tests ./tests
+                cp ${{ github.workspace }}/pyproject.toml ./pyproject.toml
+                uv sync --extra tests --no-install-project --inexact
+                uv run --no-sync pytest
+            pypi-token: ${{ secrets.pypi_token }}
+            github-user: vboussange
+            github-token: ${{ github.token }}  # automatically created token

amjax-0.0.1/.github/workflows/run_tests.yml

@@ -0,0 +1,14 @@
+name: Tests
+
+on: [push, pull_request]
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: astral-sh/setup-uv@v3
+      - name: Install dependencies
+        run: uv sync --extra tests
+      - name: Run tests
+        run: uv run pytest tests/ -v

amjax-0.0.1/.gitignore

@@ -0,0 +1,216 @@
+# macOS
+.DS_Store
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+#poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+#pdm.lock
+#pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+#pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# Abstra
+# Abstra is an AI-powered process automation framework.
+# Ignore directories containing user credentials, local state, and settings.
+# Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore 
+#  that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#  and can be added to the global gitignore or merged into this file. However, if you prefer, 
+#  you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Cursor
+#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
+#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
+#  refer to https://docs.cursor.com/context/ignore-files
+.cursorignore
+.cursorindexingignore
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+
+# AMJax — local benchmark config (personal, not for version control)
+benchmarks/config_local.yaml
+
+benchmarks/results/
+benchmarks/results_pinv/

amjax-0.0.1/LICENSE

@@ -0,0 +1,9 @@
+MIT License
+
+Copyright (c) 2026-present Victor Boussange <vic.boussange@gmail.com>
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

amjax-0.0.1/PKG-INFO

@@ -0,0 +1,128 @@
+Metadata-Version: 2.4
+Name: amjax
+Version: 0.0.1
+Summary: JAX Algebraic Multigrid Solvers in Python
+Project-URL: Documentation, https://github.com/vboussange/AMJax#readme
+Project-URL: Source, https://github.com/vboussange/AMJax
+Author-email: Victor Boussange <vic.boussange@gmail.com>, Fanny Missillier <fanny.missillier@epfl.ch>
+Maintainer-email: Victor Boussange <vic.boussange@gmail.com>, Fanny Missillier <fanny.missillier@epfl.ch>
+License-File: LICENSE
+Requires-Python: >=3.11
+Requires-Dist: jax>=0.6.1
+Requires-Dist: matplotlib>=3.0.0
+Requires-Dist: numpy>2.4.0
+Requires-Dist: pyamg>=4.0.0
+Requires-Dist: scipy>=1.11.0
+Provides-Extra: tests
+Requires-Dist: pytest; extra == 'tests'
+Description-Content-Type: text/markdown
+
+# AMJax
+
+AMJax bridges PyAMG and JAX for algebraic multigrid (AMG) solvers: it converts PyAMG-constructed hierarchies into `jax.{jit,grad,vmap}`-compatible, multi-level solvers and preconditioners for large sparse linear systems.
+
+## Installation
+
+Install directly from GitHub (PyPI release coming soon):
+
+```bash
+uv add git+https://github.com/vboussange/AMJax.git
+```
+
+## Usage
+
+### Direct solve
+
+```python
+import pyamg
+import jax
+import jax.numpy as jnp
+from amjax import AMJAXSolver
+
+A  = pyamg.gallery.poisson((100, 100), format="csr")
+b  = jnp.ones(A.shape[0])
+
+ml = AMJAXSolver.from_pyamg(pyamg.ruge_stuben_solver(A))
+
+solve = jax.jit(lambda b: ml.solve(b, tol=1e-10, maxiter=100))
+x = solve(b)
+```
+
+### Preconditioning
+
+`AMJAXSolver` exposes a preconditioner compatible with any JAX Krylov solver:
+
+```python
+from jax.experimental import sparse as jsparse
+
+A_jax = jsparse.BCOO.from_scipy_sparse(A)
+M = ml.aspreconditioner(cycle='V')
+
+x, info = jax.scipy.sparse.linalg.cg(A_jax, b, M=M, tol=1e-10, maxiter=30)
+```
+
+### Batched solve with `jax.vmap`
+
+```python
+import numpy as np
+
+B = jnp.array(np.random.rand(4, A.shape[0]))  # (n_rhs, n)
+solve_batch = jax.jit(jax.vmap(lambda b: ml.solve(b, tol=1e-8, maxiter=100)))
+X = solve_batch(B)
+```
+
+### Differentiating through the solve with `jax.grad`
+
+```python
+f = lambda b: jnp.sum(ml.solve(b, tol=1e-10, maxiter=100))
+grad = jax.grad(f)(b)
+```
+
+### Differentiation with preconditioning
+
+```python
+f = lambda b: jnp.sum(jax.scipy.sparse.linalg.cg(A_jax, b, M=M, tol=1e-10)[0])
+grad = jax.grad(f)(b)
+```
+
+## Features
+
+- V, W and F cycles compiled with `jax.jit`
+- Coarse solvers: `jacobi`, `lu`, `qr`, `pinv`
+- Smoothers: `jacobi`
+- AMG preconditioning for JAX Krylov solvers (e.g. `jax.scipy.sparse.linalg.cg`)
+- `jax.vmap` support for batched right-hand sides
+- `jax.grad` support through both direct solve and preconditioned Krylov solvers
+
+## Solvers
+
+`AMJAXSolver.from_pyamg` accepts any hierarchy produced by a PyAMG factory:
+
+| Factory | Intended for |
+|---------|--------------|
+| `pyamg.smoothed_aggregation_solver` | SPD systems, standard aggregation AMG |
+| `pyamg.rootnode_solver` | SPD systems, robust for anisotropic problems |
+| `pyamg.pairwise_solver` | SPD systems, fast setup, weaker convergence |
+| `pyamg.ruge_stuben_solver` | General SPD systems, classical C/F splitting |
+| `pyamg.air_solver` | Non-symmetric systems |
+
+**Current limitations:** V-cycle only. `jacobi` coarse solver only.
+
+## Benchmark
+
+An exhaustive benchmark can be run in Colab: 
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/vboussange/AMJax/blob/main/benchmarks/benchmark.ipynb) 
+<!-- And results are reported at XXX -->
+
+Some key insights on **speedup gains vs PyAMG-based counterpart**:
+
+
+| Scenario | Method | CPU | GPU |
+|----------|--------|----:|----:|
+| Single solve ($Ax=b$, $b \in \mathbb{R}^n$) | AMJax | - | ~16× |
+| Single solve ($Ax=b$, $b \in \mathbb{R}^n$) | AMJax + CG | - | ~17× |
+| Batched solve ($AX=B$, $B \in \mathbb{R}^{n \times K}$, $K=64$, `jax.vmap`) | AMJax | 0.7× | ~21× |
+| Batched solve ($AX=B$, $B \in \mathbb{R}^{n \times K}$, $K=64$, `jax.vmap`) | AMJax + CG | - | ~23× |
+
+
+> Settings: Ruge-Stüben hierarchy, V-cycle, Jacobi smoother, `pinv` coarse solver, $n = 1{,}000$, f64, rtol $= 10^{-10}$, max 100 iterations. JAX times exclude JIT compilation. GPU speedup is relative to the PyAMG CPU counterpart.

amjax-0.0.1/README.md

@@ -0,0 +1,109 @@
+# AMJax
+
+AMJax bridges PyAMG and JAX for algebraic multigrid (AMG) solvers: it converts PyAMG-constructed hierarchies into `jax.{jit,grad,vmap}`-compatible, multi-level solvers and preconditioners for large sparse linear systems.
+
+## Installation
+
+Install directly from GitHub (PyPI release coming soon):
+
+```bash
+uv add git+https://github.com/vboussange/AMJax.git
+```
+
+## Usage
+
+### Direct solve
+
+```python
+import pyamg
+import jax
+import jax.numpy as jnp
+from amjax import AMJAXSolver
+
+A  = pyamg.gallery.poisson((100, 100), format="csr")
+b  = jnp.ones(A.shape[0])
+
+ml = AMJAXSolver.from_pyamg(pyamg.ruge_stuben_solver(A))
+
+solve = jax.jit(lambda b: ml.solve(b, tol=1e-10, maxiter=100))
+x = solve(b)
+```
+
+### Preconditioning
+
+`AMJAXSolver` exposes a preconditioner compatible with any JAX Krylov solver:
+
+```python
+from jax.experimental import sparse as jsparse
+
+A_jax = jsparse.BCOO.from_scipy_sparse(A)
+M = ml.aspreconditioner(cycle='V')
+
+x, info = jax.scipy.sparse.linalg.cg(A_jax, b, M=M, tol=1e-10, maxiter=30)
+```
+
+### Batched solve with `jax.vmap`
+
+```python
+import numpy as np
+
+B = jnp.array(np.random.rand(4, A.shape[0]))  # (n_rhs, n)
+solve_batch = jax.jit(jax.vmap(lambda b: ml.solve(b, tol=1e-8, maxiter=100)))
+X = solve_batch(B)
+```
+
+### Differentiating through the solve with `jax.grad`
+
+```python
+f = lambda b: jnp.sum(ml.solve(b, tol=1e-10, maxiter=100))
+grad = jax.grad(f)(b)
+```
+
+### Differentiation with preconditioning
+
+```python
+f = lambda b: jnp.sum(jax.scipy.sparse.linalg.cg(A_jax, b, M=M, tol=1e-10)[0])
+grad = jax.grad(f)(b)
+```
+
+## Features
+
+- V, W and F cycles compiled with `jax.jit`
+- Coarse solvers: `jacobi`, `lu`, `qr`, `pinv`
+- Smoothers: `jacobi`
+- AMG preconditioning for JAX Krylov solvers (e.g. `jax.scipy.sparse.linalg.cg`)
+- `jax.vmap` support for batched right-hand sides
+- `jax.grad` support through both direct solve and preconditioned Krylov solvers
+
+## Solvers
+
+`AMJAXSolver.from_pyamg` accepts any hierarchy produced by a PyAMG factory:
+
+| Factory | Intended for |
+|---------|--------------|
+| `pyamg.smoothed_aggregation_solver` | SPD systems, standard aggregation AMG |
+| `pyamg.rootnode_solver` | SPD systems, robust for anisotropic problems |
+| `pyamg.pairwise_solver` | SPD systems, fast setup, weaker convergence |
+| `pyamg.ruge_stuben_solver` | General SPD systems, classical C/F splitting |
+| `pyamg.air_solver` | Non-symmetric systems |
+
+**Current limitations:** V-cycle only. `jacobi` coarse solver only.
+
+## Benchmark
+
+An exhaustive benchmark can be run in Colab: 
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/vboussange/AMJax/blob/main/benchmarks/benchmark.ipynb) 
+<!-- And results are reported at XXX -->
+
+Some key insights on **speedup gains vs PyAMG-based counterpart**:
+
+
+| Scenario | Method | CPU | GPU |
+|----------|--------|----:|----:|
+| Single solve ($Ax=b$, $b \in \mathbb{R}^n$) | AMJax | - | ~16× |
+| Single solve ($Ax=b$, $b \in \mathbb{R}^n$) | AMJax + CG | - | ~17× |
+| Batched solve ($AX=B$, $B \in \mathbb{R}^{n \times K}$, $K=64$, `jax.vmap`) | AMJax | 0.7× | ~21× |
+| Batched solve ($AX=B$, $B \in \mathbb{R}^{n \times K}$, $K=64$, `jax.vmap`) | AMJax + CG | - | ~23× |
+
+
+> Settings: Ruge-Stüben hierarchy, V-cycle, Jacobi smoother, `pinv` coarse solver, $n = 1{,}000$, f64, rtol $= 10^{-10}$, max 100 iterations. JAX times exclude JIT compilation. GPU speedup is relative to the PyAMG CPU counterpart.

amjax-0.0.1/amjax/__init__.py

@@ -0,0 +1,10 @@
+from .multilevel import MultilevelSolver
+from .relaxation.smoothing import change_smoothers
+from .relaxation.relaxation import jacobi, inverse_diagonal
+
+__all__ = [
+    "MultilevelSolver",
+    "change_smoothers",
+    "jacobi",
+    "inverse_diagonal",
+]