Sam tov jax implementation

CI/unit_tests/data/test_double_well_potential.py

@@ -9,6 +9,7 @@
 Description: Test the double_well_potential module.
 """
 import unittest
+
 from symsuite.data.double_well_potential import DoubleWellPotential
 
 
@@ -52,5 +53,5 @@ def test_double_well(self):
         self.assertEqual(len(self.generator.image), 500)
 
 
-if __name__ == '__main__':
+if __name__ == "__main__":
     unittest.main()

CI/unit_tests/distance_metrics/test_angular_distance.py

@@ -0,0 +1,72 @@
+"""
+ZnRND: A Zincwarecode package.
+License
+-------
+This program and the accompanying materials are made available under the terms
+of the Eclipse Public License v2.0 which accompanies this distribution, and is
+available at https://www.eclipse.org/legal/epl-v20.html
+SPDX-License-Identifier: EPL-2.0
+Copyright Contributors to the Zincwarecode Project.
+Contact Information
+-------------------
+email: [email protected]
+github: https://github.com/zincware
+web: https://zincwarecode.com/
+Citation
+--------
+If you use this module please cite us with:
+Summary
+-------
+Test the angular distance module.
+"""
+import os
+
+os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
+
+import jax.numpy as np
+from numpy.testing import assert_array_almost_equal
+
+from symsuite.distance_metrics.angular_distance import AngularDistance
+
+
+class TestAngularDistance:
+    """
+    Class to test the cosine distance measure module.
+    """
+
+    def test_angular_distance(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = AngularDistance()
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1, 0]])
+        point_2 = np.array([[0, 1]])
+        assert_array_almost_equal(metric(point_1, point_2), [0.5])
+
+        # Test parallel vectors
+        point_1 = np.array([[1, 0]])
+        point_2 = np.array([[1, 1]])
+        assert_array_almost_equal(metric(point_1, point_2), [0.25])
+
+    def test_multiple_distances(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = AngularDistance()
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1, 0], [1, 0]])
+        point_2 = np.array([[0, 1], [1, 1]])
+        assert_array_almost_equal(metric(point_1, point_2), [0.5, 0.25])

CI/unit_tests/distance_metrics/test_cosine_distance.py

@@ -0,0 +1,75 @@
+"""
+ZnRND: A Zincwarecode package.
+License
+-------
+This program and the accompanying materials are made available under the terms
+of the Eclipse Public License v2.0 which accompanies this distribution, and is
+available at https://www.eclipse.org/legal/epl-v20.html
+SPDX-License-Identifier: EPL-2.0
+Copyright Contributors to the Zincwarecode Project.
+Contact Information
+-------------------
+email: [email protected]
+github: https://github.com/zincware
+web: https://zincwarecode.com/
+Citation
+--------
+If you use this module please cite us with:
+Summary
+-------
+Test the cosine distance module.
+"""
+import os
+
+os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
+
+import jax.numpy as np
+from numpy.testing import assert_array_almost_equal
+
+from symsuite.distance_metrics.cosine_distance import CosineDistance
+
+
+class TestCosineDistance:
+    """
+    Class to test the cosine distance measure module.
+    """
+
+    def test_cosine_distance(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = CosineDistance()
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1, 0, 0, 0]])
+        point_2 = np.array([[0, 1, 0, 0]])
+        assert_array_almost_equal(metric(point_1, point_2), [1])
+
+        # Test parallel vectors
+        assert_array_almost_equal(metric(point_1, point_1), [0])
+
+        # Somewhere in between
+        point_1 = np.array([[1.0, 0, 0, 0]])
+        point_2 = np.array([[0.5, 1.0, 0, 3.0]])
+        assert_array_almost_equal(metric(point_1, point_2), [0.84382623])
+
+    def test_multiple_distances(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = CosineDistance()
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1, 0, 0, 0], [1, 0, 0, 0], [1.0, 0, 0, 0]])
+        point_2 = np.array([[0, 1, 0, 0], [1, 0, 0, 0], [0.5, 1.0, 0, 3.0]])
+        assert_array_almost_equal(metric(point_1, point_2), [1, 0, 0.843826], decimal=6)

CI/unit_tests/distance_metrics/test_hyper_sphere_distance.py

@@ -0,0 +1,87 @@
+"""
+ZnRND: A Zincwarecode package.
+
+License
+-------
+This program and the accompanying materials are made available under the terms
+of the Eclipse Public License v2.0 which accompanies this distribution, and is
+available at https://www.eclipse.org/legal/epl-v20.html
+SPDX-License-Identifier: EPL-2.0
+Copyright Contributors to the Zincwarecode Project.
+
+Contact Information
+-------------------
+email: [email protected]
+github: https://github.com/zincware
+web: https://zincwarecode.com/
+
+Citation
+--------
+If you use this module please cite us with:
+
+Summary
+-------
+Test the hyper sphere distance module.
+"""
+import os
+
+os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
+
+import jax.numpy as np
+from numpy.testing import assert_array_almost_equal
+
+from symsuite.distance_metrics.hyper_sphere_distance import HyperSphere
+
+
+class TestCosineDistance:
+    """
+    Class to test the cosine distance measure module.
+    """
+
+    def test_hyper_sphere_distance(self):
+        """
+        Test the hyper sphere distance.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = HyperSphere(order=2)
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1, 0, 0, 0]])
+        point_2 = np.array([[0, 1, 0, 0]])
+        assert_array_almost_equal(metric(point_1, point_2), [1.41421356])
+
+        # Test parallel vectors
+        point_1 = np.array([[1, 0, 0, 0]])
+        point_2 = np.array([[1, 0, 0, 0]])
+        assert_array_almost_equal(metric(point_1, point_2), [0])
+
+        # Somewhere in between
+        point_1 = np.array([[1.0, 0, 0, 0]])
+        point_2 = np.array([[0.5, 1.0, 0, 3.0]])
+        assert_array_almost_equal(
+            metric(point_1, point_2), [0.84382623 * np.sqrt(10.25)]
+        )
+
+    def test_multiple_distances(self):
+        """
+        Test the hyper sphere distance.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = HyperSphere(order=2)
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1, 0, 0, 0], [1, 0, 0, 0], [1.0, 0, 0, 0]])
+        point_2 = np.array([[0, 1, 0, 0], [1, 0, 0, 0], [0.5, 1.0, 0, 3.0]])
+        assert_array_almost_equal(
+            metric(point_1, point_2),
+            [np.sqrt(2), 0, 0.84382623 * np.sqrt(10.25)],
+            decimal=6,
+        )

CI/unit_tests/distance_metrics/test_l_p_norm.py

@@ -0,0 +1,85 @@
+"""
+ZnRND: A Zincwarecode package.
+License
+-------
+This program and the accompanying materials are made available under the terms
+of the Eclipse Public License v2.0 which accompanies this distribution, and is
+available at https://www.eclipse.org/legal/epl-v20.html
+SPDX-License-Identifier: EPL-2.0
+Copyright Contributors to the Zincwarecode Project.
+Contact Information
+-------------------
+email: [email protected]
+github: https://github.com/zincware
+web: https://zincwarecode.com/
+Citation
+--------
+If you use this module please cite us with:
+
+Summary
+-------
+Test the l_p norm metric.
+"""
+import os
+
+os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
+
+import jax.numpy as np
+from numpy.testing import assert_almost_equal, assert_array_almost_equal
+
+from symsuite.distance_metrics.l_p_norm import LPNorm
+
+
+class TestLPNorm:
+    """
+    Class to test the cosine distance measure module.
+    """
+
+    def test_l_2_distance(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = LPNorm(order=2)
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1.0, 7.0, 0.0, 0.0]])
+        point_2 = np.array([[1.0, 1.0, 0.0, 0.0]])
+
+        assert_array_almost_equal(metric(point_1, point_2), [6.0])
+
+    def test_l_3_distance(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = LPNorm(order=3)
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1.0, 7.0, 0.0, 0.0]])
+        point_2 = np.array([[1.0, 1.0, 0.0, 0.0]])
+        assert_almost_equal(metric(point_1, point_2), [6.0], decimal=4)
+
+    def test_multi_distance(self):
+        """
+        Test the cosine similarity measure.
+
+        Returns
+        -------
+        Assert the correct answer is returned for orthogonal, parallel, and
+        somewhere in between.
+        """
+        metric = LPNorm(order=1)
+
+        # Test orthogonal vectors
+        point_1 = np.array([[1.0, 7.0, 0.0, 0.0], [4, 7, 2, 1]])
+        point_2 = np.array([[1.0, 1.0, 0.0, 0.0], [6, 3, 1, 8]])
+        assert_array_almost_equal(metric(point_1, point_2), [6.0, 14.0], decimal=4)