scikit-learn-intelex 2024.7.0__py311-none-win_amd64.whl → 2025.0.0__py311-none-win_amd64.whl

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/basic_statistics/tests/test_incremental_basic_statistics_spmd.py

@@ -0,0 +1,307 @@
+# ==============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+import numpy as np
+import pytest
+from numpy.testing import assert_allclose
+
+from onedal.basic_statistics.tests.test_basic_statistics import options_and_tests
+from onedal.tests.utils._dataframes_support import (
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
+from sklearnex.tests._utils_spmd import (
+    _generate_statistic_data,
+    _get_local_tensor,
+    _mpi_libs_and_gpu_available,
+)
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.parametrize("weighted", [True, False])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.mpi
+def test_incremental_basic_statistics_fit_spmd_gold(dataframe, queue, weighted, dtype):
+    # Import spmd and batch algo
+    from sklearnex.basic_statistics import IncrementalBasicStatistics
+    from sklearnex.spmd.basic_statistics import (
+        IncrementalBasicStatistics as IncrementalBasicStatistics_SPMD,
+    )
+
+    # Create gold data and process into dpt
+    data = np.array(
+        [
+            [0.0, 0.0, 0.0],
+            [0.0, 1.0, 2.0],
+            [0.0, 2.0, 4.0],
+            [0.0, 3.0, 8.0],
+            [0.0, 4.0, 16.0],
+            [0.0, 5.0, 32.0],
+            [0.0, 6.0, 64.0],
+            [0.0, 7.0, 128.0],
+        ],
+        dtype=dtype,
+    )
+    dpt_data = _convert_to_dataframe(data, sycl_queue=queue, target_df=dataframe)
+
+    local_dpt_data = _convert_to_dataframe(
+        _get_local_tensor(data), sycl_queue=queue, target_df=dataframe
+    )
+
+    if weighted:
+        # Create weights array containing the weight for each sample in the data
+        weights = np.array([0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0], dtype=dtype)
+        dpt_weights = _convert_to_dataframe(
+            weights, sycl_queue=queue, target_df=dataframe
+        )
+        local_dpt_weights = _convert_to_dataframe(
+            _get_local_tensor(weights), sycl_queue=queue, target_df=dataframe
+        )
+
+    # ensure results of batch algo match spmd
+
+    incbs_spmd = IncrementalBasicStatistics_SPMD().fit(
+        local_dpt_data, sample_weight=local_dpt_weights if weighted else None
+    )
+    incbs = IncrementalBasicStatistics().fit(
+        dpt_data, sample_weight=dpt_weights if weighted else None
+    )
+
+    for option, _, _ in options_and_tests:
+        assert_allclose(
+            getattr(incbs_spmd, option),
+            getattr(incbs, option),
+            err_msg=f"Result for {option} is incorrect",
+        )
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.parametrize("num_blocks", [1, 2])
+@pytest.mark.parametrize("weighted", [True, False])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.mpi
+def test_incremental_basic_statistics_partial_fit_spmd_gold(
+    dataframe, queue, num_blocks, weighted, dtype
+):
+    # Import spmd and batch algo
+    from sklearnex.basic_statistics import IncrementalBasicStatistics
+    from sklearnex.spmd.basic_statistics import (
+        IncrementalBasicStatistics as IncrementalBasicStatistics_SPMD,
+    )
+
+    # Create gold data and process into dpt
+    data = np.array(
+        [
+            [0.0, 0.0, 0.0],
+            [0.0, 1.0, 2.0],
+            [0.0, 2.0, 4.0],
+            [0.0, 3.0, 8.0],
+            [0.0, 4.0, 16.0],
+            [0.0, 5.0, 32.0],
+            [0.0, 6.0, 64.0],
+            [0.0, 7.0, 128.0],
+        ],
+        dtype=dtype,
+    )
+    dpt_data = _convert_to_dataframe(data, sycl_queue=queue, target_df=dataframe)
+    local_data = _get_local_tensor(data)
+    split_local_data = np.array_split(local_data, num_blocks)
+
+    if weighted:
+        # Create weights array containing the weight for each sample in the data
+        weights = np.array([0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0], dtype=dtype)
+        dpt_weights = _convert_to_dataframe(
+            weights, sycl_queue=queue, target_df=dataframe
+        )
+        local_weights = _get_local_tensor(weights)
+        split_local_weights = np.array_split(local_weights, num_blocks)
+
+    incbs_spmd = IncrementalBasicStatistics_SPMD()
+    incbs = IncrementalBasicStatistics()
+
+    for i in range(num_blocks):
+        local_dpt_data = _convert_to_dataframe(
+            split_local_data[i], sycl_queue=queue, target_df=dataframe
+        )
+        if weighted:
+            local_dpt_weights = _convert_to_dataframe(
+                split_local_weights[i], sycl_queue=queue, target_df=dataframe
+            )
+        incbs_spmd.partial_fit(
+            local_dpt_data, sample_weight=local_dpt_weights if weighted else None
+        )
+
+    incbs.fit(dpt_data, sample_weight=dpt_weights if weighted else None)
+
+    for option, _, _ in options_and_tests:
+        assert_allclose(
+            getattr(incbs_spmd, option),
+            getattr(incbs, option),
+            err_msg=f"Result for {option} is incorrect",
+        )
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.parametrize("num_blocks", [1, 2])
+@pytest.mark.parametrize("weighted", [True, False])
+@pytest.mark.parametrize("result_option", options_and_tests)
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.mpi
+def test_incremental_basic_statistics_single_option_partial_fit_spmd_gold(
+    dataframe, queue, num_blocks, weighted, result_option, dtype
+):
+    # Import spmd and batch algo
+    from sklearnex.basic_statistics import IncrementalBasicStatistics
+    from sklearnex.spmd.basic_statistics import (
+        IncrementalBasicStatistics as IncrementalBasicStatistics_SPMD,
+    )
+
+    # Create gold data and process into dpt
+    data = np.array(
+        [
+            [0.0, 0.0, 0.0],
+            [0.0, 1.0, 2.0],
+            [0.0, 2.0, 4.0],
+            [0.0, 3.0, 8.0],
+            [0.0, 4.0, 16.0],
+            [0.0, 5.0, 32.0],
+            [0.0, 6.0, 64.0],
+            [0.0, 7.0, 128.0],
+        ],
+        dtype=dtype,
+    )
+    dpt_data = _convert_to_dataframe(data, sycl_queue=queue, target_df=dataframe)
+    local_data = _get_local_tensor(data)
+    split_local_data = np.array_split(local_data, num_blocks)
+
+    if weighted:
+        # Create weights array containing the weight for each sample in the data
+        weights = np.array([0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0], dtype=dtype)
+        dpt_weights = _convert_to_dataframe(
+            weights, sycl_queue=queue, target_df=dataframe
+        )
+        local_weights = _get_local_tensor(weights)
+        split_local_weights = np.array_split(local_weights, num_blocks)
+
+    option, _, _ = result_option
+    incbs_spmd = IncrementalBasicStatistics_SPMD(result_options=option)
+    incbs = IncrementalBasicStatistics(result_options=option)
+
+    for i in range(num_blocks):
+        local_dpt_data = _convert_to_dataframe(
+            split_local_data[i], sycl_queue=queue, target_df=dataframe
+        )
+        if weighted:
+            local_dpt_weights = _convert_to_dataframe(
+                split_local_weights[i], sycl_queue=queue, target_df=dataframe
+            )
+        incbs_spmd.partial_fit(
+            local_dpt_data, sample_weight=local_dpt_weights if weighted else None
+        )
+
+    incbs.fit(dpt_data, sample_weight=dpt_weights if weighted else None)
+
+    assert_allclose(getattr(incbs_spmd, option), getattr(incbs, option))
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.parametrize("num_blocks", [1, 2])
+@pytest.mark.parametrize("weighted", [True, False])
+@pytest.mark.parametrize("n_samples", [100, 10000])
+@pytest.mark.parametrize("n_features", [10, 100])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.mpi
+def test_incremental_basic_statistics_partial_fit_spmd_synthetic(
+    dataframe, queue, num_blocks, weighted, n_samples, n_features, dtype
+):
+    # Import spmd and batch algo
+    from sklearnex.basic_statistics import IncrementalBasicStatistics
+    from sklearnex.spmd.basic_statistics import (
+        IncrementalBasicStatistics as IncrementalBasicStatistics_SPMD,
+    )
+
+    tol = 2e-3 if dtype == np.float32 else 1e-7
+
+    # Create gold data and process into dpt
+    data = _generate_statistic_data(n_samples, n_features, dtype=dtype)
+    local_data = _get_local_tensor(data)
+    split_local_data = np.array_split(local_data, num_blocks)
+    split_data = np.array_split(data, num_blocks)
+
+    if weighted:
+        # Create weights array containing the weight for each sample in the data
+        weights = _generate_statistic_data(n_samples, dtype=dtype)
+        local_weights = _get_local_tensor(weights)
+        split_local_weights = np.array_split(local_weights, num_blocks)
+        split_weights = np.array_split(weights, num_blocks)
+
+    incbs_spmd = IncrementalBasicStatistics_SPMD()
+    incbs = IncrementalBasicStatistics()
+
+    for i in range(num_blocks):
+        local_dpt_data = _convert_to_dataframe(
+            split_local_data[i], sycl_queue=queue, target_df=dataframe
+        )
+        dpt_data = _convert_to_dataframe(
+            split_data[i], sycl_queue=queue, target_df=dataframe
+        )
+        if weighted:
+            local_dpt_weights = _convert_to_dataframe(
+                split_local_weights[i], sycl_queue=queue, target_df=dataframe
+            )
+            dpt_weights = _convert_to_dataframe(
+                split_weights[i], sycl_queue=queue, target_df=dataframe
+            )
+        incbs_spmd.partial_fit(
+            local_dpt_data, sample_weight=local_dpt_weights if weighted else None
+        )
+        incbs.partial_fit(dpt_data, sample_weight=dpt_weights if weighted else None)
+
+    for option, _, _ in options_and_tests:
+        assert_allclose(
+            getattr(incbs_spmd, option),
+            getattr(incbs, option),
+            atol=tol,
+            err_msg=f"Result for {option} is incorrect",
+        )

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/covariance/__init__.py

@@ -0,0 +1,20 @@
+# ==============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from .covariance import EmpiricalCovariance
+from .incremental_covariance import IncrementalEmpiricalCovariance
+
+__all__ = ["EmpiricalCovariance", "IncrementalEmpiricalCovariance"]

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/covariance/incremental_covariance.py

@@ -0,0 +1,37 @@
+# ==============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from onedal.spmd.covariance import (
+    IncrementalEmpiricalCovariance as onedalSPMD_IncrementalEmpiricalCovariance,
+)
+
+from ...covariance import (
+    IncrementalEmpiricalCovariance as base_IncrementalEmpiricalCovariance,
+)
+
+
+class IncrementalEmpiricalCovariance(base_IncrementalEmpiricalCovariance):
+    """
+    Incremental distributed estimator for covariance.
+    Allows to distributely compute empirical covariance estimated by maximum
+    likelihood method if data are splitted into batches.
+
+    API is the same as for `sklearnex.covariance.IncrementalEmpiricalCovariance`
+    """
+
+    _onedal_incremental_covariance = staticmethod(
+        onedalSPMD_IncrementalEmpiricalCovariance
+    )

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/covariance/tests/test_incremental_covariance_spmd.py

@@ -0,0 +1,184 @@
+# ==============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+import numpy as np
+import pytest
+from numpy.testing import assert_allclose
+
+from onedal.tests.utils._dataframes_support import (
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
+from sklearnex.tests._utils_spmd import (
+    _generate_statistic_data,
+    _get_local_tensor,
+    _mpi_libs_and_gpu_available,
+)
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.parametrize("assume_centered", [True, False])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.mpi
+def test_incremental_covariance_fit_spmd_gold(dataframe, queue, assume_centered, dtype):
+    # Import spmd and batch algo
+    from sklearnex.covariance import IncrementalEmpiricalCovariance
+    from sklearnex.spmd.covariance import (
+        IncrementalEmpiricalCovariance as IncrementalEmpiricalCovariance_SPMD,
+    )
+
+    # Create gold data and process into dpt
+    data = np.array(
+        [
+            [0.0, 0.0, 0.0],
+            [0.0, 1.0, 2.0],
+            [0.0, 2.0, 4.0],
+            [0.0, 3.0, 8.0],
+            [0.0, 4.0, 16.0],
+            [0.0, 5.0, 32.0],
+            [0.0, 6.0, 64.0],
+            [0.0, 7.0, 128.0],
+        ],
+        dtype=dtype,
+    )
+
+    dpt_data = _convert_to_dataframe(data, sycl_queue=queue, target_df=dataframe)
+
+    local_dpt_data = _convert_to_dataframe(
+        _get_local_tensor(data), sycl_queue=queue, target_df=dataframe
+    )
+
+    # ensure results of batch algo match spmd
+    spmd_result = IncrementalEmpiricalCovariance_SPMD(
+        assume_centered=assume_centered
+    ).fit(local_dpt_data)
+    non_spmd_result = IncrementalEmpiricalCovariance(assume_centered=assume_centered).fit(
+        dpt_data
+    )
+
+    assert_allclose(spmd_result.covariance_, non_spmd_result.covariance_)
+    assert_allclose(spmd_result.location_, non_spmd_result.location_)
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.parametrize("num_blocks", [1, 2])
+@pytest.mark.parametrize("assume_centered", [True, False])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.mpi
+def test_incremental_covariance_partial_fit_spmd_gold(
+    dataframe, queue, num_blocks, assume_centered, dtype
+):
+    # Import spmd and batch algo
+    from sklearnex.covariance import IncrementalEmpiricalCovariance
+    from sklearnex.spmd.covariance import (
+        IncrementalEmpiricalCovariance as IncrementalEmpiricalCovariance_SPMD,
+    )
+
+    # Create gold data and process into dpt
+    data = np.array(
+        [
+            [0.0, 0.0, 0.0],
+            [0.0, 1.0, 2.0],
+            [0.0, 2.0, 4.0],
+            [0.0, 3.0, 8.0],
+            [0.0, 4.0, 16.0],
+            [0.0, 5.0, 32.0],
+            [0.0, 6.0, 64.0],
+            [0.0, 7.0, 128.0],
+        ],
+        dtype=dtype,
+    )
+
+    dpt_data = _convert_to_dataframe(data, sycl_queue=queue, target_df=dataframe)
+
+    local_data = _get_local_tensor(data)
+    split_local_data = np.array_split(local_data, num_blocks)
+
+    inccov_spmd = IncrementalEmpiricalCovariance_SPMD(assume_centered=assume_centered)
+    inccov = IncrementalEmpiricalCovariance(assume_centered=assume_centered)
+
+    for i in range(num_blocks):
+        local_dpt_data = _convert_to_dataframe(
+            split_local_data[i], sycl_queue=queue, target_df=dataframe
+        )
+        inccov_spmd.partial_fit(local_dpt_data)
+
+    inccov.fit(dpt_data)
+
+    assert_allclose(inccov_spmd.covariance_, inccov.covariance_)
+    assert_allclose(inccov_spmd.location_, inccov.location_)
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize("n_samples", [100, 10000])
+@pytest.mark.parametrize("n_features", [10, 100])
+@pytest.mark.parametrize("num_blocks", [1, 2])
+@pytest.mark.parametrize("assume_centered", [True, False])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.mpi
+def test_incremental_covariance_partial_fit_spmd_synthetic(
+    n_samples, n_features, num_blocks, assume_centered, dataframe, queue, dtype
+):
+    # Import spmd and batch algo
+    from sklearnex.covariance import IncrementalEmpiricalCovariance
+    from sklearnex.spmd.covariance import (
+        IncrementalEmpiricalCovariance as IncrementalEmpiricalCovariance_SPMD,
+    )
+
+    # Generate data and process into dpt
+    data = _generate_statistic_data(n_samples, n_features, dtype=dtype)
+
+    dpt_data = _convert_to_dataframe(data, sycl_queue=queue, target_df=dataframe)
+
+    local_data = _get_local_tensor(data)
+    split_local_data = np.array_split(local_data, num_blocks)
+
+    inccov_spmd = IncrementalEmpiricalCovariance_SPMD(assume_centered=assume_centered)
+    inccov = IncrementalEmpiricalCovariance(assume_centered=assume_centered)
+
+    for i in range(num_blocks):
+        local_dpt_data = _convert_to_dataframe(
+            split_local_data[i], sycl_queue=queue, target_df=dataframe
+        )
+        inccov_spmd.partial_fit(local_dpt_data)
+
+    inccov.fit(dpt_data)
+
+    tol = 1e-7
+
+    assert_allclose(inccov_spmd.covariance_, inccov.covariance_, atol=tol)
+    assert_allclose(inccov_spmd.location_, inccov.location_, atol=tol)

{scikit_learn_intelex-2024.7.0.data/data/Lib/site-packages/sklearnex/preview → scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/decomposition}/__init__.py

@@ -14,4 +14,7 @@
 # limitations under the License.
 # ==============================================================================
 
-__all__ = ["cluster", "covariance", "decomposition", "linear_model"]
+from .incremental_pca import IncrementalPCA
+from .pca import PCA
+
+__all__ = ["IncrementalPCA", "PCA"]

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/decomposition/incremental_pca.py

@@ -0,0 +1,30 @@
+# ==============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from onedal.spmd.decomposition import IncrementalPCA as onedalSPMD_IncrementalPCA
+
+from ...preview.decomposition import IncrementalPCA as base_IncrementalPCA
+
+
+class IncrementalPCA(base_IncrementalPCA):
+    """
+    Distributed incremental estimator for PCA based on sklearnex implementation.
+    Allows for distributed PCA computation if data is split into batches.
+
+    API is the same as for `sklearnex.decomposition.IncrementalPCA`
+    """
+
+    _onedal_incremental_pca = staticmethod(onedalSPMD_IncrementalPCA)