def main(data_sizes: List, approximation_ranks: List, increments: List,
singular_values: RowVector, used_data_factory: Callable,
results_path: str, experiment_type: str, power_method_iterations,
_seed) -> None:
""" The main function of this project
This functions performs the desired experiment according to the given configuration.
The function runs the random_svd and random_id for every combination of data_size, approximation rank and increment
given in the config and saves all the results to a csv file in the results folder (given in the configuration).
"""
seed(_seed) # Seed the interpolative package of Scipy for later usage.
results_log = DataLog(LogFields) # Initializing an empty results log.
random_svd_with_run_time: Callable = measure_time(random_svd)
random_id_with_run_time: Callable = measure_time(random_id)
for data_size in data_sizes:
data_matrix: Matrix = used_data_factory(data_size, singular_values)
for approximation_rank in approximation_ranks:
next_singular_value: Scalar = singular_values[approximation_rank + 1] if \
approximation_rank < len(singular_values) else singular_values[-1]
for increment in increments:
# Executing all the tested methods.
print(
f'n={data_size}, k={approximation_rank}, l={approximation_rank + increment}'
)
random_svd_approximation, svd_duration = random_svd_with_run_time(
data_matrix, approximation_rank, increment)
random_svd_accuracy: Scalar = estimate_spectral_norm_diff(
data_matrix, random_svd_approximation,
power_method_iterations)
print(
f'runtime={svd_duration}, accuracy={random_svd_accuracy}')
random_id_approximation, id_duration = random_id_with_run_time(
data_matrix, approximation_rank, increment)
random_id_accuracy: Scalar = estimate_spectral_norm_diff(
data_matrix, random_id_approximation,
power_method_iterations)
print(f'runtime={id_duration}, accuracy={random_id_accuracy}')
# Appending all the experiment results to the log.
results_log.append(LogFields.DataSize, data_size)
results_log.append(LogFields.ApproximationRank,
approximation_rank)
results_log.append(LogFields.Increment,
increment + approximation_rank)
results_log.append(LogFields.NextSingularValue,
next_singular_value)
results_log.append(LogFields.RandomSVDAccuracy,
random_svd_accuracy)
results_log.append(LogFields.RandomIDAccuracy,
random_id_accuracy)
results_log.append(LogFields.RandomSVDDuration, svd_duration)
results_log.append(LogFields.RandomIDDuration, id_duration)
results_log.save_log(experiment_type + " results",
results_folder_path=results_path)
def test_rand(self):
pymatrixid.seed('default')
assert_(np.allclose(pymatrixid.rand(2), [0.8932059, 0.64500803], 1e-4))
pymatrixid.seed(1234)
x1 = pymatrixid.rand(2)
assert_(np.allclose(x1, [0.7513823, 0.06861718], 1e-4))
np.random.seed(1234)
pymatrixid.seed()
x2 = pymatrixid.rand(2)
np.random.seed(1234)
pymatrixid.seed(np.random.rand(55))
x3 = pymatrixid.rand(2)
assert_allclose(x1, x2)
assert_allclose(x1, x3)
def test_rand(self):
pymatrixid.seed('default')
assert_(np.allclose(pymatrixid.rand(2), [0.8932059, 0.64500803], 1e-4))
pymatrixid.seed(1234)
x1 = pymatrixid.rand(2)
assert_(np.allclose(x1, [0.7513823, 0.06861718], 1e-4))
np.random.seed(1234)
pymatrixid.seed()
x2 = pymatrixid.rand(2)
np.random.seed(1234)
pymatrixid.seed(np.random.rand(55))
x3 = pymatrixid.rand(2)
assert_allclose(x1, x2)
assert_allclose(x1, x3)
