def _resolve_optimizer(optimizer_info):
"""Resolve the optimization routine from the given information dictionary.
Args:
optimizer_info (dict): the optimization dictionary with at least 'name' for the optimizer and settings
for the optimizer settings
Returns:
optimizer: the optimization routine
"""
name = optimizer_info['name']
settings = deepcopy(optimizer_info.get('settings', {}) or {})
optimizer = get_optimizer_by_name(name)
if 'optimizers' in settings and settings['optimizers']:
settings['optimizers'] = [
_resolve_optimizer(info) for info in settings['optimizers']
]
if 'optimizer' in settings and settings['optimizer']:
settings['optimizer'] = _resolve_optimizer(settings['optimizer'])
if 'starting_point_generator' in settings and settings[
'starting_point_generator']:
cls = getattr(mot.cl_routines.optimizing.random_restart,
list(settings['starting_point_generator'].keys())[0])
settings['starting_point_generator'] = cls(
**list(settings['starting_point_generator'].values())[0])
return optimizer(**settings)
def __init__(self):
"""Storage class for communication between the options dialog and the main frame"""
self.use_model_default_optimizer = True
self.double_precision = False
self.optimizer = mdt.configuration.get_general_optimizer_name()
self.patience = mdt.configuration.get_general_optimizer_settings(
)['patience']
if self.patience is None:
self.patience = get_optimizer_by_name(
self.optimizer).default_patience
self.recalculate_all = False
def get_optimizer(self):
if self.use_model_default_optimizer:
return None
optimizer = get_optimizer_by_name(self.optimizer)
return optimizer(patience=self.patience)
def _update_default_patience(self):
optimizer = get_optimizer_by_name(OptimOptions.optim_routines[
self.optimizationRoutine.currentText()])
self.patience.setText(str(optimizer.default_patience))