def test_from_data(self):
es = ElementStats.from_data('megnet_1', stats=['moment:None:5'])
d = es.transform_one('Fe2O3')
self.assertTrue(d.shape == (1, 80))
es = ElementStats.from_data(['megnet_1'], stats=['moment:None:5'])
d = es.transform_one('Fe2O3')
self.assertTrue(d.shape == (1, 80))
es2 = ElementStats.from_data(['megnet_1', 'megnet_3'], stats=['moment:None:5'])
d = es2.transform_one('Fe2O3')
self.assertTrue(d.shape == (1, 160))
def test_from_data(self):
es = ElementStats.from_data("megnet_1", stats=["moment:None:5"])
d = es.transform_one("Fe2O3")
self.assertTrue(d.shape == (1, 80))
es = ElementStats.from_data(["megnet_1"], stats=["moment:None:5"])
d = es.transform_one("Fe2O3")
self.assertTrue(d.shape == (1, 80))
es2 = ElementStats.from_data(["megnet_1", "megnet_3"],
stats=["moment:None:5"])
d = es2.transform_one("Fe2O3")
self.assertTrue(d.shape == (1, 160))
def test_initialization(self):
self.assertRaises(TypeError, ElementStats,
element_properties={"H": [1, 2], "O": [1, 2]}, stats=['mean'],
property_names=['p1', 'p2'],
some_random_variable='test',
)
es = ElementStats(element_properties={"H": [1, 2], "O": [1, 2]}, stats=['mean'],
property_names=['p1', 'p2'],
n_jobs=-1)
self.assertTrue(es.n_jobs > 0)
res = es.transform(["H2O", "H2O", "H2O", "H2O"])
self.assertTrue(res.shape[0] == 4)
es.clear_cache()
def test_kpca(self):
es2 = ElementStats.from_data(["megnet_1", "megnet_3"],
stats=["moment:None:5"],
num_dim=2,
reduction_algo="kpca")
d = es2.transform_one("Fe2O3")
self.assertTrue(d.shape == (1, 10))
class TestAtomSets(TestCase):
x = ["H2O", "Fe2O3"]
y = [0.1, 0.2]
model = MLP(describer=ElementStats.from_data("megnet_3"), n_neurons=(2, 2))
def test_train(self):
self.model.train(self.x, self.y, epochs=0)
self.assertTrue(self.model.predict_objs(["H2O"]).shape == (1, 1))
def test_element_stats(self):
test = ElementStats({'H': [4, 2, 3],
'O': [2, 3, 4]},
stats=['min', 'max', 'moment:1:10'])
self.assertEqual(test.transform(['H2O']).shape, (1, 36))
res = test.transform(['H2O', 'H2O'])
self.assertEqual(res.shape, (2, 36))
res2 = test.transform([Composition('H2O'), Composition('H2O')])
np.testing.assert_allclose(res.values, res2.values)
dummy_h2o = Molecule(['H', 'H', 'O'], [[-0.5, 0, 0], [0.5, 0, 0], [0, 0, 0]])
res3 = test.transform([dummy_h2o, dummy_h2o])
np.testing.assert_allclose(res.values, res3.values)
stats = ['min', 'max', *['moment:%d:None' % i for i in range(1, 11)]]
p0_names = ['p0_%s' % i for i in stats]
p1_names = ['p1_%s' % i for i in stats]
p2_names = ['p2_%s' % i for i in stats]
all_names = []
for i, j, k in zip(p0_names, p1_names, p2_names):
all_names.extend([i, j, k])
self.assertListEqual(list(res.columns), all_names)
def test_element_stats(self):
test = ElementStats({
"H": [4, 2, 3],
"O": [2, 3, 4]
},
stats=["min", "max", "moment:1:10"])
self.assertEqual(test.transform(["H2O"]).shape, (1, 36))
res = test.transform(["H2O", "H2O"])
self.assertEqual(res.shape, (2, 36))
res2 = test.transform([Composition("H2O"), Composition("H2O")])
np.testing.assert_allclose(res.values, res2.values)
dummy_h2o = Molecule(["H", "H", "O"],
[[-0.5, 0, 0], [0.5, 0, 0], [0, 0, 0]])
res3 = test.transform([dummy_h2o, dummy_h2o])
np.testing.assert_allclose(res.values, res3.values)
stats = ["min", "max", *["moment:%d:None" % i for i in range(1, 11)]]
p0_names = ["p0_%s" % i for i in stats]
p1_names = ["p1_%s" % i for i in stats]
p2_names = ["p2_%s" % i for i in stats]
all_names = []
for i, j, k in zip(p0_names, p1_names, p2_names):
all_names.extend([i, j, k])
self.assertListEqual(list(res.columns), all_names)
def test_error(self):
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={"H": [1, 2], "O": [1, 2, 3]}, stats=['mean'])
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={"H": [1, 2], "O": [1, 2]}, stats=['mean'],
property_names=['p1'])
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={"H": [1, 2], "O": [1, 2]}, stats=['super_std'],
property_names=['p1', 'p2'])
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={"H": [1, 2], "O": [1, 2]}, stats=['super_std:-1'],
property_names=['p1', 'p2'])
with self.assertRaises(ValueError):
ElementStats.from_file(os.path.join(CWD, 'test_data/wrong_dummy_property.json'), stats=['max'])
with self.assertRaises(ValueError):
ElementStats.from_data('megnet_22', stats=['max'])
def test_error(self):
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={
"H": [1, 2],
"O": [1, 2, 3]
},
stats=["mean"])
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={
"H": [1, 2],
"O": [1, 2]
},
stats=["mean"],
property_names=["p1"])
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={
"H": [1, 2],
"O": [1, 2]
},
stats=["super_std"],
property_names=["p1", "p2"])
with self.assertRaises(ValueError):
_ = ElementStats(element_properties={
"H": [1, 2],
"O": [1, 2]
},
stats=["super_std:-1"],
property_names=["p1", "p2"])
with self.assertRaises(ValueError):
ElementStats.from_file(os.path.join(
CWD, "test_data/wrong_dummy_property.json"),
stats=["max"])
with self.assertRaises(ValueError):
ElementStats.from_data("megnet_22", stats=["max"])
def test_keys(self):
es = ElementStats.from_file(os.path.join(CWD, 'test_data/dummy_property.json'), stats=['max'])
self.assertListEqual(es.stats, ['mean'])
def test_geometric_mean(self):
es2 = ElementStats.from_data(['megnet_1', 'megnet_3'], stats=['shifted_geometric_mean:100'])
d = es2.transform_one('Fe2O3')
self.assertTrue(d.shape == (1, 32))
def test_pca(self):
es2 = ElementStats.from_data(["megnet_1", "megnet_3"],
stats=["moment:None:5"],
num_dim=4)
d = es2.transform_one("Fe2O3")
self.assertTrue(d.shape == (1, 20))
def setUpClass(cls) -> None:
cls.x = ['H2O', 'Fe2O3']
cls.y = [0.1, 0.2]
cls.model = MLP(describer=ElementStats.from_data("megnet_3"), n_neurons=(2, 2))
def test_geometric_mean(self):
es2 = ElementStats.from_data(["megnet_1", "megnet_3"],
stats=["shifted_geometric_mean:100"])
d = es2.transform_one("Fe2O3")
self.assertTrue(d.shape == (1, 32))
def test_pca(self):
es2 = ElementStats.from_data(['megnet_1', 'megnet_3'], stats=['moment:None:5'], num_dim=4)
d = es2.transform_one('Fe2O3')
self.assertTrue(d.shape == (1, 20))
def test_kpca(self):
es2 = ElementStats.from_data(['megnet_1', 'megnet_3'], stats=['moment:None:5'], num_dim=2,
reduction_algo='kpca')
d = es2.transform_one('Fe2O3')
self.assertTrue(d.shape == (1, 10))
def test_keys(self):
es = ElementStats.from_file(os.path.join(
CWD, "test_data/dummy_property.json"),
stats=["max"])
self.assertListEqual(es.stats, ["mean"])