check for namespace and device compatibility (#3149)

Author: david-cortes-intel

sklearnex/cluster/k_means.py

@@ -40,13 +40,15 @@
     from onedal.cluster import KMeans as onedal_KMeans
     from onedal.utils.validation import _is_arraylike_not_scalar, _is_csr
 
-    from .._config import get_config
     from .._device_offload import dispatch, wrap_output_data
     from .._utils import PatchingConditionsChain
     from ..base import oneDALEstimator
     from ..utils._array_api import enable_array_api, get_namespace
     from ..utils.validation import validate_data
 
+    if sklearn_check_version("1.9"):
+        from sklearn.utils._array_api import check_same_namespace
+
     @enable_array_api
     @control_n_jobs(decorated_methods=["fit", "fit_transform", "predict", "score"])
     class KMeans(oneDALEstimator, _sklearn_KMeans):
@@ -383,7 +385,10 @@ def predict(
                 )
 
         def _onedal_predict(self, X, sample_weight=None, queue=None):
-
+            if sklearn_check_version("1.9"):
+                check_same_namespace(
+                    X, self, attribute="cluster_centers_", method="predict"
+                )
             xp, _ = get_namespace(X)
             X = validate_data(
                 self,
@@ -450,7 +455,10 @@ def transform(self, X):
             )
 
         def _onedal_transform(self, X, queue=None):
-
+            if sklearn_check_version("1.9"):
+                check_same_namespace(
+                    X, self, attribute="cluster_centers_", method="transform"
+                )
             xp, is_array_api = get_namespace(X)
             X = validate_data(
                 self,
@@ -492,7 +500,10 @@ def score(self, X, y=None, sample_weight=None):
             )
 
         def _onedal_score(self, X, y=None, sample_weight=None, queue=None):
-
+            if sklearn_check_version("1.9"):
+                check_same_namespace(
+                    X, self, attribute="cluster_centers_", method="score"
+                )
             xp, _ = get_namespace(X)
             X = validate_data(
                 self,

sklearnex/cluster/tests/test_kmeans.py

@@ -26,9 +26,11 @@
 from onedal.tests.utils._dataframes_support import (
     _as_numpy,
     _convert_to_dataframe,
+    dpnp_available,
     get_dataframes_and_queues,
     get_queues,
 )
+from onedal.tests.utils._device_selection import is_sycl_device_available
 from sklearnex import config_context
 from sklearnex.cluster import KMeans
 from sklearnex.tests.utils import _IS_INTEL
@@ -224,3 +226,38 @@ def test_array_api_dispatch_output_type(dataframe, queue):
         assert type(trans) == type(X)
         assert type(km.cluster_centers_) == type(X)
         assert isinstance(sc, float)
+
+
+@pytest.mark.skipif(
+    not sklearn_check_version("1.9"),
+    reason="Relies on functionality introduced in later scikit-learn versions.",
+)
+@pytest.mark.skipif(not dpnp_available, reason="Functionality to test requires DPNP.")
+@pytest.mark.skipif(
+    not is_sycl_device_available("gpu"), reason="Test for GPU-specific functionality."
+)
+def test_cov_error_on_incompatible_devices(with_array_api):
+    import dpnp
+
+    rng = np.random.default_rng(seed=123)
+    X = rng.random(size=(50, 3), dtype=np.float32)
+    X_cpu = dpnp.array(X, device="cpu")
+    X_gpu = dpnp.array(X, device="gpu")
+
+    err_match = "device|queue"
+
+    model = KMeans(algorithm="lloyd").fit(X_gpu)
+    with pytest.raises(ValueError, match=err_match):
+        _ = model.predict(X_cpu)
+    with pytest.raises(ValueError, match=err_match):
+        _ = model.transform(X_cpu)
+    with pytest.raises(ValueError, match=err_match):
+        _ = model.score(X_cpu)
+
+    model.fit(X_cpu)
+    with pytest.raises(ValueError, match=err_match):
+        _ = model.predict(X_gpu)
+    with pytest.raises(ValueError, match=err_match):
+        _ = model.transform(X_gpu)
+    with pytest.raises(ValueError, match=err_match):
+        _ = model.score(X_gpu)