Update tests except UQ

tests/calculators/test_calculator.py

@@ -34,52 +34,50 @@
 ]
 
 
-class TestCalculator:
-    def test_compute(self):
-        test_file_path = Path(__file__).parents[1].joinpath("configs_extxyz")
-        tset = Dataset(test_file_path.joinpath("Si_4"))
-        configs = tset.get_configs()
-
-        modelname = "SW_StillingerWeber_1985_Si__MO_405512056662_006"
-        model = KIMModel(modelname)
-
-        # calculator
-        calc = Calculator(model)
-        compute_arguments = calc.create(configs)
-
-        for i, ca in enumerate(compute_arguments):
-            calc.compute(ca)
-            energy = calc.get_energy(ca)
-            forces = calc.get_forces(ca)[:3]
-
-            assert energy == pytest.approx(ref_energies[i], 1e-6)
-            assert np.allclose(forces, ref_forces[i])
-
-    def test_parameter(self):
-        modelname = "SW_StillingerWeber_1985_Si__MO_405512056662_006"
-        model = KIMModel(modelname)
-
-        # parameters
-        params = model.get_model_params()
-        sigma = params["sigma"][0]
-        A = params["A"][0]
-
-        # optimizing parameters
-        # B will not be optimized, only providing initial guess
-        model.set_opt_params(
-            sigma=[["default"]], B=[["default", "fix"]], A=[["default"]]
-        )
-
-        calc = Calculator(model)
-
-        x0 = calc.get_opt_params()
-        assert x0[0] == sigma
-        assert x0[1] == A
-        assert len(x0) == 2
-        assert model.get_num_opt_params() == 2
-
-        x1 = [i + 0.1 for i in x0]
-        calc.update_model_params(x1)
-
-        assert params["sigma"][0] == sigma + 0.1
-        assert params["A"][0] == A + 0.1
+def test_compute(test_data_dir):
+    test_file_dir = test_data_dir.joinpath("configs/Si_4")
+    tset = Dataset(test_file_dir)
+    configs = tset.get_configs()
+
+    modelname = "SW_StillingerWeber_1985_Si__MO_405512056662_006"
+    model = KIMModel(modelname)
+
+    # calculator
+    calc = Calculator(model)
+    compute_arguments = calc.create(configs)
+
+    for i, ca in enumerate(compute_arguments):
+        calc.compute(ca)
+        energy = calc.get_energy(ca)
+        forces = calc.get_forces(ca)[:3]
+
+        assert energy == pytest.approx(ref_energies[i], 1e-6)
+        assert np.allclose(forces, ref_forces[i])
+
+
+def test_parameter():
+    modelname = "SW_StillingerWeber_1985_Si__MO_405512056662_006"
+    model = KIMModel(modelname)
+
+    # parameters
+    params = model.get_model_params()
+    sigma = params["sigma"][0]
+    A = params["A"][0]
+
+    # optimizing parameters
+    # B will not be optimized, only providing initial guess
+    model.set_opt_params(sigma=[["default"]], B=[["default", "fix"]], A=[["default"]])
+
+    calc = Calculator(model)
+
+    x0 = calc.get_opt_params()
+    assert x0[0] == sigma
+    assert x0[1] == A
+    assert len(x0) == 2
+    assert model.get_num_opt_params() == 2
+
+    x1 = [i + 0.1 for i in x0]
+    calc.update_model_params(x1)
+
+    assert params["sigma"][0] == sigma + 0.1
+    assert params["A"][0] == A + 0.1

tests/calculators/test_calculator_torch.py

@@ -1,6 +1,7 @@
 from pathlib import Path
 
 import numpy as np
+import pytest
 import torch
 from torch import Tensor
 
@@ -10,56 +11,90 @@
 from kliff.descriptors import SymmetryFunction
 from kliff.models import NeuralNetwork
 
-# model
-descriptor = SymmetryFunction(
-    cut_name="cos", cut_dists={"Si-Si": 5.0}, hyperparams="set30", normalize=True
-)
-
-N1 = np.random.randint(5, 10)
-N2 = np.random.randint(5, 10)
-model = NeuralNetwork(descriptor)
-model.add_layers(
-    # first hidden layer
-    nn.Linear(descriptor.get_size(), N1),
-    nn.Tanh(),
-    # second hidden layer
-    nn.Linear(N1, N2),
-    nn.Tanh(),
-    # output layer
-    nn.Linear(N2, 1),
-)
-
-# training set
-path = Path(__file__).absolute().parents[1].joinpath("configs_extxyz/Si_4")
-data = Dataset(path)
-configs = data.get_configs()
-
-# calculator
-calc = CalculatorTorch(model, gpu=False)
-_ = calc.create(configs, reuse=False)
-loader = calc.get_compute_arguments(batch_size=100)
-
-# data on parameter sizes
-exp_sizes = [
-    torch.Size([N1, 30]),
-    torch.Size([N1]),
-    torch.Size([N2, N1]),
-    torch.Size([N2]),
-    torch.Size([1, N2]),
-    torch.Size([1]),
-]
-exp_nparams_per_layer = [N1 * 30, N1, N2 * N1, N2, N2, 1]
-exp_nparams_total = np.sum(exp_nparams_per_layer)
-
-# parameters to try
-p0 = np.zeros(exp_nparams_total)
-p1 = np.ones(exp_nparams_total)
-
-
-# Test if the functions to update parameters work
-
-
-def test_get_parameters_sizes():
+
+@pytest.fixture(scope="module")
+def N1():
+    return np.random.randint(5, 10)
+
+
+@pytest.fixture(scope="module")
+def N2():
+    return np.random.randint(5, 10)
+
+
+@pytest.fixture(scope="module")
+def calc(test_data_dir, N1, N2):
+    # model
+    descriptor = SymmetryFunction(
+        cut_name="cos", cut_dists={"Si-Si": 5.0}, hyperparams="set30", normalize=True
+    )
+
+    model = NeuralNetwork(descriptor)
+    model.add_layers(
+        # first hidden layer
+        nn.Linear(descriptor.get_size(), N1),
+        nn.Tanh(),
+        # second hidden layer
+        nn.Linear(N1, N2),
+        nn.Tanh(),
+        # output layer
+        nn.Linear(N2, 1),
+    )
+
+    # training set
+    data = Dataset(test_data_dir / "configs" / "Si_4")
+    configs = data.get_configs()
+
+    # calculator
+    calc = CalculatorTorch(model, gpu=False)
+    _ = calc.create(configs, reuse=False)
+
+    return calc
+
+
+@pytest.fixture(scope="module")
+def loader(calc, N1, N2):
+    return calc.get_compute_arguments(batch_size=100)
+
+
+@pytest.fixture(scope="module")
+def exp_sizes(N1, N2):
+    # data on parameter sizes
+    exp_sizes = [
+        torch.Size([N1, 30]),
+        torch.Size([N1]),
+        torch.Size([N2, N1]),
+        torch.Size([N2]),
+        torch.Size([1, N2]),
+        torch.Size([1]),
+    ]
+
+    return exp_sizes
+
+
+@pytest.fixture(scope="module")
+def exp_nparams_per_layer(N1, N2):
+    return [N1 * 30, N1, N2 * N1, N2, N2, 1]
+
+
+@pytest.fixture(scope="module")
+def exp_nparams_total(exp_nparams_per_layer):
+    return np.sum(exp_nparams_per_layer)
+
+
+@pytest.fixture(scope="module")
+def p0(exp_nparams_total):
+    return np.zeros(exp_nparams_total)
+
+
+@pytest.fixture(scope="module")
+def p1(exp_nparams_total):
+    return np.ones(exp_nparams_total)
+
+
+def test_get_parameters_sizes(
+    calc, exp_sizes, exp_nparams_per_layer, exp_nparams_total
+):
     """
     Test if the function to get parameters sizes works.
 
@@ -74,7 +109,7 @@ def test_get_parameters_sizes():
     assert nparams_total == exp_nparams_total, "Total number of parameters is incorrect"
 
 
-def test_parameter_values():
+def test_parameter_values(calc, p0, p1):
     """
     Test if the parameter values are updated.
 
@@ -89,19 +124,20 @@ def test_parameter_values():
     ), "Either `update_model_params` or `get_opt_params` not working"
 
 
-def test_predictions_change():
+def test_predictions_change(calc, loader, p0, p1):
     """
     Test if changing parameters affect the predictions.
 
     There are two steps of this test. The first one, if we set all the parameters to be
     zero, then the (forces) predictions should also be zero.
 
-    Then, if we change the parameters to some other values, the predictions should change
-    and they should not be zero, unless there is something special with the
+    Then, if we change the parameters to some other values, the predictions should
+    change and they should not be zero, unless there is something special with the
     configurations.
     """
     # Test if predictions are zeros when all parameters are zeros
     calc.update_model_params(p0)
+
     for batch in loader:
         calc.compute(batch)
     # We will only look at the forces
@@ -110,7 +146,7 @@ def test_predictions_change():
     for f0 in forces0:
         all_zeros.append(Tensor.all(f0 == 0.0))
     assert np.all(all_zeros), (
-        "Problem in predicitons calculation: "
+        "Problem in prediction calculation: "
         + "there are non-zero forces when all parameters are zero"
     )
 
@@ -124,9 +160,3 @@ def test_predictions_change():
         change.append(not Tensor.all(f0 - f1 == 0.0))
     # Use any since there might be special configurations
     assert np.any(change), "Changing parameters doesn't change predictions"
-
-
-if __name__ == "__main__":
-    test_get_parameters_sizes()
-    test_parameter_values()
-    test_predictions_change()

tests/dataset/test_extxyz.py

@@ -1,18 +1,28 @@
-from pathlib import Path
-
 import numpy as np
+import pytest
 
 from kliff.dataset.dataset import Configuration, Dataset
 
 
-def test_configuration(e=True, f=False, s=False, order=False):
-    path = Path(__file__).parents[1].joinpath("configs_extxyz/MoS2")
+@pytest.mark.parametrize(
+    "f,s,order",
+    (
+        [False, False, False],
+        [True, False, False],
+        [True, True, False],
+        [True, True, True],
+    ),
+)
+def test_configuration(test_data_dir, f, s, order, e=True):
+    path = test_data_dir.joinpath("configs/MoS2")
+
     if e:
-        fname = path.joinpath("MoS2_energy.xyz")
+        fname = path / "MoS2_energy.xyz"
     if f:
-        fname = path.joinpath("MoS2_energy_forces.xyz")
+        fname = path / "MoS2_energy_forces.xyz"
     if s:
-        fname = path.joinpath("MoS2_energy_forces_stress.xyz")
+        fname = path / "MoS2_energy_forces_stress.xyz"
+
     fname = fname.as_posix()
 
     config = Configuration.from_file(fname, file_format="xyz")
@@ -71,15 +81,7 @@ def test_configuration(e=True, f=False, s=False, order=False):
     assert natoms_by_species["S"] == 192
 
 
-def test_config():
-    test_configuration(True, False, False)
-    test_configuration(True, True, False)
-    test_configuration(True, True, True)
-    test_configuration(True, True, True, order=True)
-
-
-def test_dataset():
-    directory = Path(__file__).parents[1].joinpath("configs_extxyz/MoS2").as_posix()
-    tset = Dataset(directory)
+def test_dataset(test_data_dir):
+    tset = Dataset(test_data_dir / "configs/MoS2")
     configs = tset.get_configs()
     assert len(configs) == 3

tests/uq/test_bootstrap_empirical.py

@@ -186,15 +186,3 @@ def test_multi_calc_cas_generator():
     assert (
         len(bootstrap_cas[0][0]) + len(bootstrap_cas[0][1]) == ncas_energy + ncas_forces
     ), "For each sample, generator should generate the same number of cas in total as the original"
-
-
-if __name__ == "__main__":
-    test_wrapper()
-    test_error()
-    test_bootstrap_cas_generator()
-    test_callback()
-    test_run()
-    test_appending_cas()
-    test_save_load_cas()
-    test_reset()
-    test_multi_calc_cas_generator()

tests/uq/test_bootstrap_nn.py

@@ -146,14 +146,3 @@ def test_reset():
     # Check reset bootstrap samples
     assert BS._nsamples_prepared == 0, "Reset bootstrap cas failed"
     assert BS._nsamples_done == 0, "Reset ensembles failed"
-
-
-if __name__ == "__main__":
-    test_wrapper()
-    test_error()
-    test_original_state()
-    test_bootstrap_cas_generator()
-    test_run()
-    test_appending_cas()
-    test_save_load_cas()
-    test_reset()

tests/uq/test_bootstrap_nn_separate_species.py

@@ -101,9 +101,3 @@ def test_run():
         shape == exp_shape
     ), f"Samples doesn't have the right shape; expected {exp_shape}, got {shape}"
     assert BS._nsamples_done == nsamples
-
-
-if __name__ == "__main__":
-    test_model()
-    test_original_state()
-    test_run()