def prepare_optimization(self,
initial_optimization_parameters: np.ndarray):
"""Prepare for the next optimization.
Parameters
----------
initial_optimization_parameters : array
shape (num_t, num_ctrl)
Data stored in this class might be overwritten.
"""
self._min_costs = np.inf
self._min_costs_par = None
self._n_cost_fkt_eval = 0
self._n_jac_fkt_eval = 0
self.pulse_shape = initial_optimization_parameters.shape
if self.save_intermediary_steps:
self.optim_iter_summary = \
optimization_data.OptimizationSummary(
indices=self.system_simulator.cost_indices
)
self._opt_start_time = time.time()
if self.system_simulator.stats is not None:
# If the system simulator wants to write down statistics, then
# initialise a fresh instance
self.system_simulator.stats = \
performance_statistics.PerformanceStatistics()
self.system_simulator.stats.start_t_opt = float(
self._opt_start_time)
self.system_simulator.stats.indices = \
self.system_simulator.cost_indices
def __init__(
self,
solvers: Optional[Sequence[solver_algorithms.Solver]],
cost_fktns: Optional[Sequence[cost_functions.CostFunction]],
optimization_parameters=None,
num_ctrl=None,
times=None,
num_times=None,
record_performance_statistics: bool = True,
numeric_jacobian: bool = False,
cost_fktn_weights: Optional[Sequence[float]] = None
):
self._num_ctrl = num_ctrl
self._num_times = num_times
self._optimization_parameteres = optimization_parameters
self._times = times
self.solvers = solvers
self.cost_fktns = cost_fktns
self.stats = (performance_statistics.PerformanceStatistics()
if record_performance_statistics else None)
self.numeric_jacobian = numeric_jacobian
self.cost_fktn_weights = cost_fktn_weights
if self.cost_fktn_weights is not None:
if len(self.cost_fktn_weights) == 0:
self.cost_fktn_weights = None
elif not len(self.cost_fktns) == len(self.cost_fktn_weights):
raise ValueError('A cost function weight must be specified for'
'each cost function or for none at all.')
def __init__(self,
solvers: Optional[Sequence[solver_algorithms.Solver]],
cost_fktns: Optional[Sequence[cost_functions.CostFunction]],
optimization_parameters=None,
num_ctrl=None,
times=None,
num_times=None,
record_performance_statistics: bool = True,
numeric_jacobian: bool = False):
self._num_ctrl = num_ctrl
self._num_times = num_times
self._optimization_parameteres = optimization_parameters
self._times = times
self.solvers = solvers
self.cost_fktns = cost_fktns
self.stats = (performance_statistics.PerformanceStatistics()
if record_performance_statistics else None)
self.numeric_jacobian = numeric_jacobian