def test_save_load(tmpdir):
path = os.path.join(tmpdir, "dissimilarity_matrix.npy")
matrix = np.asarray([[1, 0], [0.5, 0.25]], dtype=np.float32)
reference = cf.DissimilarityMatrix(matrix)
reference.save(path)
loaded = cf.DissimilarityMatrix.load(path)
assert np.all(matrix == loaded.get_dissimilarity_matrix())
def to_prec_fit(args):
matrix = cc.DissimilarityMatrix(args[4])
return (
args[0],
args[1],
args[2],
args[3],
matrix.calc_precision(),
matrix.calc_fitness(),
)
def test_calc_fitness_precision_1():
matrix = cf.DissimilarityMatrix(
np.asarray([
# dissimilarities are maximal for all model/real trace pairs
# thus, precision and fitness is minimal
[1.0, 1.0],
[1.0, 1.0],
]))
assert matrix.calc_precision() == pytest.approx(0.0)
assert matrix.calc_fitness() == pytest.approx(0.0)
def test_calc_fitness_precision_5():
# this example just checks for floating point semantics
# all dissimilarities are high, thus precision and fitness is low
matrix = cf.DissimilarityMatrix(
np.asarray([
[1 - 1e-9, 1 - 1e-9],
[1 - 1e-9, 1 - 1e-9],
]))
assert matrix.calc_precision() == pytest.approx(1e-9)
assert matrix.calc_fitness() == pytest.approx(1e-9)
def test_calc_fitness_precision_4():
matrix = cf.DissimilarityMatrix(
np.asarray([
# each model trace (row) is equal to a real trace (column)
# but the model does not cover all real traces
# thus, precision is maximal but fitness is low
[1.0, 0.0],
[1.0, 0.0],
]))
assert matrix.calc_precision() == pytest.approx(1.0)
assert matrix.calc_fitness() == pytest.approx(0.5)
def test_calc_fitness_precision_3():
matrix = cf.DissimilarityMatrix(
np.asarray([
# each real trace (column) can be represented by the model (rows),
# but the model allows for extra traces not covered by the real log
# thus, fitness is maximal but precision is low
[0.0, 0.0],
[1.0, 1.0],
]))
assert matrix.calc_precision() == pytest.approx(0.5)
assert matrix.calc_fitness() == pytest.approx(1.0)