Merge pull request #150 from Helmholtz-AI-Energy/hotfix/test_surrogate

Add missing decorator and island test for dynamic surrogate

.github/workflows/python-test.yml

@@ -18,8 +18,7 @@ jobs:
             - name: Install dependencies
               run: |
                   python -m pip install --upgrade pip
-                  pip install ."[dev]"
-                  pip install genbadge"[coverage]"
+                  pip install ."[test]"
             - name: Lint with ruff
               run: |
                   # Stop the build if there are Python syntax errors or undefined names.

pyproject.toml

@@ -47,6 +47,15 @@ dev = [
     "torchmetrics",
 ]
 
+test = [
+    "coverage",
+    "genbadge[coverage]",
+    "ruff",
+    "pytest",
+    "pytest-cov",
+    "pytest-mpi",
+]
+
 tutorials = [
     "torch",
     "torchvision",

tests/test_cmaes.py

@@ -38,7 +38,7 @@ def test_cmaes_basic(cma_adapter: CMAAdapter, mpi_tmp_path: pathlib.Path) -> Non
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         checkpoint_path=mpi_tmp_path,
     )
     # Run optimization and print summary of results.

tests/test_island.py

@@ -70,7 +70,7 @@ def test_islands(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         migration_probability=0.9,
         pollination=pollination,
@@ -109,7 +109,7 @@ def test_checkpointing_isolated(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         migration_probability=0.0,
         checkpoint_path=mpi_tmp_path,
@@ -126,7 +126,7 @@ def test_checkpointing_isolated(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         migration_probability=0.0,
         checkpoint_path=mpi_tmp_path,
@@ -170,7 +170,7 @@ def test_checkpointing(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         migration_probability=0.9,
         pollination=pollination,
@@ -188,7 +188,7 @@ def test_checkpointing(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         migration_probability=0.9,
         pollination=pollination,
@@ -233,7 +233,7 @@ def test_checkpointing_unequal_populations(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         island_sizes=np.array([3, 5]),
         migration_probability=0.9,
@@ -252,7 +252,7 @@ def test_checkpointing_unequal_populations(
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         num_islands=2,
         island_sizes=np.array([3, 5]),
         migration_probability=0.9,

tests/test_multi_rank_workers.py

@@ -69,7 +69,7 @@ def test_multi_rank_workers(mpi_tmp_path: pathlib.Path) -> None:
         loss_fn=parallel_sphere,  # Loss function to be minimized
         propagator=propagator,  # Propagator, i.e., evolutionary operator to be used
         rng=rng,  # Separate random number generator for Propulate optimization
-        generations=100,  # Overall number of generations
+        generations=10,  # Overall number of generations
         num_islands=2,  # Number of islands
         migration_probability=0.9,  # Migration probability
         pollination=False,  # Whether to use pollination or migration

tests/test_nm.py

@@ -61,7 +61,7 @@ def test_cmaes(
         loss_fn=function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         checkpoint_path=mpi_tmp_path,
     )
     # Run optimization and print summary of results.

tests/test_propulator.py

@@ -61,7 +61,7 @@ def test_propulator(function_name: str, mpi_tmp_path: pathlib.Path) -> None:
         loss_fn=benchmark_function,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         checkpoint_path=mpi_tmp_path,
     )  # Set up propulator performing actual optimization.
     propulator.propulate()  # Run optimization and print summary of results.
@@ -91,7 +91,7 @@ def test_propulator_checkpointing(mpi_tmp_path: pathlib.Path) -> None:
     propulator = Propulator(
         loss_fn=benchmark_function,
         propagator=propagator,
-        generations=100,
+        generations=10,
         checkpoint_path=mpi_tmp_path,
         rng=rng,
     )  # Set up propulator performing actual optimization.
@@ -107,7 +107,7 @@ def test_propulator_checkpointing(mpi_tmp_path: pathlib.Path) -> None:
     propulator = Propulator(
         loss_fn=benchmark_function,
         propagator=propagator,
-        generations=20,
+        generations=5,
         checkpoint_path=mpi_tmp_path,
         rng=rng,
     )  # Set up new propulator starting from checkpoint.

tests/test_pso.py

@@ -77,7 +77,7 @@ def test_pso(pso_propagator: Propagator, mpi_tmp_path: pathlib.Path) -> None:
         loss_fn=sphere,
         propagator=propagator,
         rng=rng,
-        generations=100,
+        generations=10,
         checkpoint_path=mpi_tmp_path,
     )
 

tests/test_surrogate.py

@@ -10,6 +10,11 @@
 from propulate import Islands, Propulator, surrogate
 from propulate.utils import get_default_propagator, set_logger_config
 
+pytestmark = pytest.mark.filterwarnings(
+    "ignore::DeprecationWarning",
+    match="Assigning the 'data' attribute is an inherently unsafe operation and will be removed in the future.",
+)
+
 log = logging.getLogger(__name__)  # Get logger instance.
 set_logger_config(level=logging.DEBUG)
 
@@ -51,7 +56,7 @@ def test_static(mpi_tmp_path: Path) -> None:
     rng = random.Random(
         MPI.COMM_WORLD.rank + 100
     )  # Set up separate random number generator for evolutionary optimizer.
-    num_generations = 50
+    num_generations = 4
 
     propagator = get_default_propagator(  # Get default evolutionary operator.
         pop_size=pop_size,  # Breeding population size
@@ -74,6 +79,7 @@ def test_static(mpi_tmp_path: Path) -> None:
     MPI.COMM_WORLD.barrier()
 
 
+@pytest.mark.mpi(min_size=8)
 def test_static_island(mpi_tmp_path: Path) -> None:
     """Test static surrogate using a dummy function."""
     pop_size = 2 * MPI.COMM_WORLD.size  # Breeding population size
@@ -84,7 +90,7 @@ def test_static_island(mpi_tmp_path: Path) -> None:
     rng = random.Random(
         MPI.COMM_WORLD.rank + 100
     )  # Set up separate random number generator for evolutionary optimizer.
-    num_generations = 50
+    num_generations = 4
 
     propagator = get_default_propagator(  # Get default evolutionary operator.
         pop_size=pop_size,  # Breeding population size
@@ -111,13 +117,47 @@ def test_static_island(mpi_tmp_path: Path) -> None:
     MPI.COMM_WORLD.barrier()
 
 
+def test_dynamic(mpi_tmp_path: Path) -> None:
+    """Test static surrogate using a dummy function."""
+    pop_size = 2 * MPI.COMM_WORLD.size  # Breeding population size
+    limits: Dict[str, Union[Tuple[int, int], Tuple[float, float], Tuple[str, ...]]] = {
+        "start": (0.1, 7.0),
+        "limit": (-1.0, 1.0),
+    }  # Define search space.
+    rng = random.Random(
+        MPI.COMM_WORLD.rank + 100
+    )  # Set up separate random number generator for evolutionary optimizer.
+    num_generations = 4
+
+    propagator = get_default_propagator(  # Get default evolutionary operator.
+        pop_size=pop_size,  # Breeding population size
+        limits=limits,  # Search space
+        crossover_prob=0.7,  # Crossover probability
+        mutation_prob=0.4,  # Mutation probability
+        random_init_prob=0.1,  # Random-initialization probability
+        rng=rng,  # Random number generator for evolutionary optimizer
+    )
+    propulator = Propulator(
+        loss_fn=ind_loss,
+        propagator=propagator,
+        generations=num_generations,
+        checkpoint_path=mpi_tmp_path,
+        rng=rng,
+        surrogate_factory=lambda: surrogate.DynamicSurrogate(limits),
+    )  # Set up propulator performing actual optimization.
+
+    propulator.propulate()  # Run optimization and print summary of results.
+    MPI.COMM_WORLD.barrier()
+
+
+@pytest.mark.mpi(min_size=8)
 @pytest.mark.filterwarnings(
     "ignore::DeprecationWarning",
     match="Assigning the 'data' attribute is an inherently unsafe operation and will be removed in the future.",
 )
-def test_dynamic(mpi_tmp_path: Path) -> None:
+def test_dynamic_island(mpi_tmp_path: Path) -> None:
     """Test dynamic surrogate using a dummy function."""
-    num_generations = 10  # Number of generations
+    num_generations = 4  # Number of generations
     pop_size = 2 * MPI.COMM_WORLD.size  # Breeding population size
     limits: Dict[str, Union[Tuple[int, int], Tuple[float, float], Tuple[str, ...]]] = {
         "start": (0.1, 7.0),
@@ -137,7 +177,7 @@ def test_dynamic(mpi_tmp_path: Path) -> None:
         propagator=propagator,  # Evolutionary operator
         rng=rng,  # Random number generator
         generations=num_generations,  # Number of generations per worker
-        num_islands=1,  # Number of islands
+        num_islands=2,  # Number of islands
         checkpoint_path=mpi_tmp_path,
         surrogate_factory=lambda: surrogate.DynamicSurrogate(limits),
     )