rewrite unittests to completely remove dependency on NeuroChem (#2)

rewrite unittests to completely remove dependency on NeuroChem

.gitattributes

 *.h5 filter=lfs diff=lfs merge=lfs -text
+tests/test_data/* filter=lfs diff=lfs merge=lfs -text
\ No newline at end of file

examples/ANI-1x-training-Adam.py

@@ -50,7 +50,8 @@ def subset_rmse(subset_dataloader):
             coordinates, energies = batch[molecule_id]
             coordinates = coordinates.to(aev_computer.device)
             energies = energies.to(aev_computer.device)
-            count, squared_error = evaluate(model, coordinates, energies, _species)
+            count, squared_error = evaluate(
+                model, coordinates, energies, _species)
             squared_error = squared_error.item()
             a.add(count, squared_error)
     mse = a.avg()
@@ -79,7 +80,8 @@ while True:
             coordinates, energies = batch[molecule_id]
             coordinates = coordinates.to(aev_computer.device)
             energies = energies.to(aev_computer.device)
-            count, squared_error = evaluate(model, coordinates, energies, _species)
+            count, squared_error = evaluate(
+                model, coordinates, energies, _species)
             a.add(count, squared_error / len(_species))
         optimize_step(a)
         step += 1

examples/ANI-1x-training.py

@@ -78,7 +78,8 @@ while True:
             coordinates, energies = batch[molecule_id]
             coordinates = coordinates.to(aev_computer.device)
             energies = energies.to(aev_computer.device)
-            count, squared_error = evaluate(model, coordinates, energies, _species)
+            count, squared_error = evaluate(
+                model, coordinates, energies, _species)
             a.add(count, squared_error / len(_species))
         optimize_step(a)
         step += 1

setup.py

@@ -17,6 +17,6 @@ setup(name='torchani',
           'h5py',
       ],
       test_suite='nose.collector',
-      tests_require=['nose', 'cntk'],
+      tests_require=['nose'],
       cmdclass=cmdclass,
       )

tests/test_aev.py

+import torch
+import numpy
+import torchani
+import unittest
+import os
+import pickle
+
+path = os.path.dirname(os.path.realpath(__file__))
+N = 97
+
+
+class TestAEV(unittest.TestCase):
+
+    def setUp(self, dtype=torchani.default_dtype):
+        self.aev = torchani.SortedAEV(dtype=dtype, device=torch.device('cpu'))
+        self.tolerance = 1e-5
+
+    def _test_molecule(self, coordinates, species, expected_radial, expected_angular):
+        radial, angular = self.aev(coordinates, species)
+        radial_diff = expected_radial - radial
+        radial_max_error = torch.max(torch.abs(radial_diff)).item()
+        angular_diff = expected_angular - angular
+        angular_max_error = torch.max(torch.abs(angular_diff)).item()
+        self.assertLess(radial_max_error, self.tolerance)
+        self.assertLess(angular_max_error, self.tolerance)
+
+    def testGDB(self):
+        for i in range(N):
+            datafile = os.path.join(path, 'test_data/{}'.format(i))
+            with open(datafile, 'rb') as f:
+                coordinates, species, radial, angular, _, _ = pickle.load(f)
+                self._test_molecule(coordinates, species, radial, angular)
+
+
+if __name__ == '__main__':
+    unittest.main()

torchani/tests/test_benchmark.py → tests/test_benchmark.py

@@ -85,7 +85,9 @@ class TestBenchmark(unittest.TestCase):
                          'terms and indices>radial terms',
                          'terms and indices>angular terms',
                          'total>terms and indices',
-                         'total>partition', 'total>assemble'])
+                         'total>combinations', 'total>assemble',
+                         'total>mask_r', 'total>mask_a'
+                         ])
 
     def testModelOnAEV(self):
         aev_computer = torchani.SortedAEV(

torchani/tests/test_data.py → tests/test_data.py

@@ -11,9 +11,13 @@ from bisect import bisect
 from pickle import dump, load
 
 
+path = os.path.dirname(os.path.realpath(__file__))
+dataset_dir = os.path.join(path, 'dataset')
+
+
 class TestDataset(unittest.TestCase):
 
-    def setUp(self, data_path=torchani.buildin_dataset_dir):
+    def setUp(self, data_path=dataset_dir):
         self.data_path = data_path
         self.ds = torchani.data.load_dataset(data_path)