import time
from typing import List, Tuple
import numpy as np
from quara.loss_function.probability_based_loss_function import (
ProbabilityBasedLossFunction,
ProbabilityBasedLossFunctionOption,
)
from quara.minimization_algorithm.minimization_algorithm import (
MinimizationAlgorithm,
MinimizationAlgorithmOption,
)
from quara.protocol.qtomography.standard.standard_qtomography import StandardQTomography
from quara.protocol.qtomography.standard.standard_qtomography_estimator import (
StandardQTomographyEstimator,
StandardQTomographyEstimationResult,
)
[docs]class LossMinimizationEstimationResult(StandardQTomographyEstimationResult):
def __init__(
self,
qtomography: StandardQTomography,
data,
estimated_var_sequence: List[np.ndarray],
computation_times: List[float],
):
super().__init__(qtomography, data, estimated_var_sequence, computation_times)
[docs]class LossMinimizationEstimator(StandardQTomographyEstimator):
def __init__(self):
super().__init__()
[docs] def calc_estimate(
self,
qtomography: StandardQTomography,
empi_dists: List[Tuple[int, np.ndarray]],
loss: ProbabilityBasedLossFunction,
loss_option: ProbabilityBasedLossFunctionOption,
algo: MinimizationAlgorithm,
algo_option: MinimizationAlgorithmOption,
is_computation_time_required: bool = False,
) -> StandardQTomographyEstimationResult:
"""calculates estimate variables.
Notice: this function updates ``loss`` and ``algo`` properties.
Parameters
----------
qtomography : StandardQTomography
StandardQTomography to calculates estimate variables.
empi_dists : List[Tuple[int, np.ndarray]]
empirical distributions to calculates estimate variables.
loss : ProbabilityBasedLossFunction
ProbabilityBasedLossFunction to calculates estimate variables.
loss_option : ProbabilityBasedLossFunctionOption
ProbabilityBasedLossFunctionOption to calculates estimate variables.
algo : MinimizationAlgorithm
MinimizationAlgorithm to calculates estimate variables.
algo_option : MinimizationAlgorithmOption
MinimizationAlgorithmOption to calculates estimate variables.
is_computation_time_required : bool, optional
whether to include computation time in the return value or not, by default False.
Returns
-------
StandardQTomographyEstimationResult
estimation result.
Raises
------
ValueError
loss.is_option_sufficient() returns False.
ValueError
algo.is_loss_sufficient() returns False.
ValueError
algo.is_option_sufficient() returns False.
ValueError
algo.is_loss_and_option_sufficient() returns False.
"""
result = self.calc_estimate_sequence(
qtomography,
[empi_dists],
loss,
loss_option,
algo,
algo_option,
is_computation_time_required=is_computation_time_required,
)
return result
[docs] def calc_estimate_sequence(
self,
qtomography: StandardQTomography,
empi_dists_sequence: List[List[Tuple[int, np.ndarray]]],
loss: ProbabilityBasedLossFunction,
loss_option: ProbabilityBasedLossFunctionOption,
algo: MinimizationAlgorithm,
algo_option: MinimizationAlgorithmOption,
is_computation_time_required: bool = False,
) -> StandardQTomographyEstimationResult:
"""calculates sequence of estimate variables.
Notice: this function updates ``loss`` and ``algo`` properties.
Parameters
----------
qtomography : StandardQTomography
StandardQTomography to calculates estimate variables.
empi_dists_sequence : List[List[Tuple[int, np.ndarray]]]
sequence of empirical distributions to calculates estimate variables.
loss : ProbabilityBasedLossFunction
ProbabilityBasedLossFunction to calculates estimate variables.
loss_option : ProbabilityBasedLossFunctionOption
ProbabilityBasedLossFunctionOption to calculates estimate variables.
algo : MinimizationAlgorithm
MinimizationAlgorithm to calculates estimate variables.
algo_option : MinimizationAlgorithmOption
MinimizationAlgorithmOption to calculates estimate variables.
is_computation_time_required : bool, optional
whether to include computation time in the return value or not, by default False.
Returns
-------
StandardQTomographyEstimationResult
estimation result.
Raises
------
ValueError
loss.is_option_sufficient() returns False.
ValueError
algo.is_loss_sufficient() returns False.
ValueError
algo.is_option_sufficient() returns False.
ValueError
algo.is_loss_and_option_sufficient() returns False.
"""
estimated_var_sequence = []
computation_times = [] if is_computation_time_required else None
for empi_dists in empi_dists_sequence:
if is_computation_time_required:
start_time = time.time()
# set loss settings
loss.set_from_standard_qtomography_option_data(
qtomography,
loss_option,
empi_dists,
algo.is_gradient_required,
algo.is_hessian_required,
)
# set algorithm settings
algo.set_from_option(algo_option)
algo.set_constraint_from_standard_qt_and_option(qtomography, algo_option)
algo.set_from_loss(loss)
# validate
if loss.is_option_sufficient() == False:
raise ValueError(
"loss.is_option_sufficient() must return True. But returns False"
)
if algo.is_loss_sufficient() == False:
raise ValueError(
"algo.is_loss_sufficient() must return True. But returns False"
)
if algo.is_option_sufficient() == False:
raise ValueError(
"algo.is_option_sufficient() must return True. But returns False"
)
if algo.is_loss_and_option_sufficient() == False:
raise ValueError(
"algo.is_loss_and_option_sufficient() must return True. But returns False"
)
if is_computation_time_required:
prepare_time = time.time() - start_time
# optimize
algo_result = algo.optimize(
loss,
loss_option,
algo_option,
on_iteration_history=is_computation_time_required,
)
# post-processing
estimated_var_sequence.append(algo_result.value)
if is_computation_time_required:
computation_times.append(prepare_time + algo_result.computation_time)
result = LossMinimizationEstimationResult(
qtomography, empi_dists_sequence, estimated_var_sequence, computation_times
)
return result