make cueq optional dep and add special test

.github/workflows/unittest.yaml

@@ -5,25 +5,46 @@ on:
     branches: [main]
 
 jobs:
-  pytest-container:
+  pytest-general:
     runs-on: ubuntu-latest
-
     steps:
       - uses: actions/checkout@v4
       - uses: actions/setup-python@v5
         with:
           python-version: "3.10"
           cache: "pip"
 
-      - name: Install requirements
+      - name: Install requirements (general tests)
         run: |
           pip install -U pip
           pip install .[dev]
 
-      - name: Log installed environment
+      - name: Log installed environment (general tests)
+        run: |
+          python3 -m pip freeze
+
+      - name: Run general unit tests
+        run: |
+          pytest tests --ignore=tests/test_cueq.py
+
+  pytest-cueq:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.10"
+          cache: "pip"
+
+      - name: Install requirements (with cueq)
+        run: |
+          pip install -U pip
+          pip install ".[dev, cueq]"
+
+      - name: Log installed environment (with cueq)
         run: |
           python3 -m pip freeze
 
-      - name: Run unit tests
+      - name: Run cueq-specific tests
         run: |
-          pytest tests
+          pytest tests/test_cueq.py tests/test_calculator.py

setup.cfg

@@ -29,7 +29,6 @@ install_requires =
     GitPython
     pyYAML
     tqdm
-    cuequivariance-torch
     # for plotting:
     matplotlib
     pandas
@@ -60,5 +59,6 @@ dev =
     pytest-benchmark
     pylint
 schedulefree = schedulefree
+cueq = cuequivariance-torch
 cueq-cuda-11 = cuequivariance-ops-torch-cu11
 cueq-cuda-12 = cuequivariance-ops-torch-cu12

tests/test_calculator.py

@@ -586,6 +586,7 @@ def test_calculator_descriptor(fitting_configs, trained_equivariant_model):
     assert not np.allclose(desc, desc_rotated, atol=1e-6)
 
 
+@pytest.mark.skipif(not CUET_AVAILABLE, reason="cuequivariance not installed")
 def test_calculator_descriptor_cueq(fitting_configs, trained_equivariant_model_cueq):
     at = fitting_configs[2].copy()
     at_rotated = fitting_configs[2].copy()

tests/test_cueq.py

@@ -143,7 +143,7 @@ def test_bidirectional_conversion(
         loss_e3nn_back.backward()
 
         # Compare gradients for all conversions
-        tol = 1e-4 if default_dtype == torch.float32 else 1e-8
+        tol = 1e-4 if default_dtype == torch.float32 else 1e-7
 
         def print_gradient_diff(name1, p1, name2, p2, conv_type):
             if p1.grad is not None and p1.grad.shape == p2.grad.shape:
@@ -152,7 +152,7 @@ def print_gradient_diff(name1, p1, name2, p2, conv_type):
                     print(
                         f"{conv_type} - Parameter {name1}/{name2}, Max error: {error.max()}"
                     )
-                    torch.testing.assert_close(p1.grad, p2.grad, atol=tol, rtol=1e-10)
+                    torch.testing.assert_close(p1.grad, p2.grad, atol=tol, rtol=tol)
 
         # E3nn to CuEq gradients
         for (name_e3nn, p_e3nn), (name_cueq, p_cueq) in zip(