chore: Make some aggregator dependencies optional (#327)

* Add optional dependency groups nash_mtl, cagrad and full
* Remove cxvpy and ecos from the default dependencies
* Change tests.yml to also install the full dependency group
* Update documentation about how to install torchjd in README.md, installation.md, cagrad.py and nash_mtl.py
* Add changelog entry

Author: ValerianRey

.github/workflows/tests.yml

@@ -22,7 +22,7 @@ jobs:
         with:
           python-version: ${{ matrix.python-version }}
       - name: Install default and test dependencies
-        run: pdm install --group test --frozen-lockfile
+        run: pdm install --group full --group test --frozen-lockfile
       - name: Run unit and doc tests with coverage report
         run: pdm run pytest tests/unit tests/doc --cov=src --cov-report=xml
       - name: Upload results to Codecov

CHANGELOG.md

@@ -8,6 +8,13 @@ changes that do not affect the user.
 
 ## [Unreleased]
 
+### Changed
+
+- **BREAKING**: Changed the dependencies of `CAGrad` and `NashMTL` to be optional when installing
+  TorchJD. Users of these aggregators will have to use `pip install torchjd[cagrad]`, `pip install
+  torchjd[nash_mtl]` or `pip install torchjd[full]` to install TorchJD alongside those dependencies.
+  This should make TorchJD more lightweight.
+
 ## [0.6.0] - 2025-04-19
 
 ### Added

README.md

@@ -52,6 +52,8 @@ TorchJD can be installed directly with pip:
 pip install torchjd
 ```
 <!-- end installation -->
+Some aggregators may have additional dependencies. Please refer to the
+[installation documentation](https://torchjd.org/stable/installation) for them.
 
 ## Usage
 The main way to use TorchJD is to replace the usual call to `loss.backward()` by a call to

docs/source/installation.md

@@ -6,3 +6,17 @@
 ```
 
 Note that `torchjd` requires python 3.10, 3.11, 3.12 or 3.13 and `torch>=2.0`.
+
+Some aggregators (CAGrad and Nash-MTL) have additional dependencies that are not included by default
+when installing `torchjd`. To install them, you can use:
+```
+pip install torchjd[cagrad]
+```
+```
+pip install torchjd[nash_mtl]
+```
+
+To install `torchjd` with all of its optional dependencies, you can also use:
+```
+pip install torchjd[full]
+```

pyproject.toml

@@ -17,8 +17,6 @@ dependencies = [
     "quadprog>=0.1.9, != 0.1.10",  # Doesn't work before 0.1.9, 0.1.10 is yanked
     "numpy>=1.21.0",  # Does not work before 1.21
     "qpsolvers>=1.0.1",  # Does not work before 1.0.1
-    "cvxpy>=1.3.0",  # No Clarabel solver before 1.3.0
-    "ecos>=2.0.14",  # Does not work before 2.0.14
 ]
 classifiers = [
     "Development Status :: 4 - Beta",
@@ -67,3 +65,16 @@ plot = [
     "dash>=2.16.0",  # Recent version to avoid problems, could be relaxed
     "kaleido==0.2.1",  # Only works with locked version
 ]
+
+[project.optional-dependencies]
+nash_mtl = [
+    "cvxpy>=1.3.0",  # Could be relaxed
+    "ecos>=2.0.14",  # Does not work before 2.0.14
+]
+cagrad = [
+    "cvxpy>=1.3.0",  # No Clarabel solver before 1.3.0
+]
+full = [
+    "cvxpy>=1.3.0",  # No Clarabel solver before 1.3.0
+    "ecos>=2.0.14",  # Does not work before 2.0.14
+]

src/torchjd/aggregation/cagrad.py

@@ -29,6 +29,10 @@ class CAGrad(_WeightedAggregator):
         >>>
         >>> A(J)
         tensor([0.1835, 1.2041, 1.2041])
+
+    .. note::
+        This aggregator has dependencies that are not included by default when installing
+        ``torchjd``. To install them, use ``pip install torchjd[cagrad]``.
     """
 
     def __init__(self, c: float, norm_eps: float = 0.0001):

src/torchjd/aggregation/nash_mtl.py

@@ -60,12 +60,16 @@ class NashMTL(_WeightedAggregator):
         >>> A(J)
         tensor([0.0542, 0.7061, 0.7061])
 
+    .. note::
+        This aggregator has dependencies that are not included by default when installing
+        ``torchjd``. To install them, use ``pip install torchjd[nash_mtl]``.
+
     .. warning::
         This implementation was adapted from the `official implementation
         <https://github.com/AvivNavon/nash-mtl/tree/main>`_, which has some flaws. Use with caution.
 
     .. warning::
-        The aggregator is stateful. Its output will thus depend not only on the input matrix, but
+        This aggregator is stateful. Its output will thus depend not only on the input matrix, but
         also on its state. It thus depends on previously seen matrices. It should be reset between
         experiments.
     """