OptimizationHyperparameters#
- class OptimizationHyperparameters(opttol=0.01, gradConverge=False, min_inner_iter=5, max_restart=inf, penalty_ratio=0.01, penalty_reduction=0.1, break_iter_period=50, verbose=0)[source]#
Bases:
objectHyperparameters for optimization algorithms.
- Parameters:
opttol (float)
gradConverge (bool)
min_inner_iter (int)
max_restart (float)
penalty_ratio (float)
penalty_reduction (float)
break_iter_period (int)
verbose (int)
- opttol#
Optimization tolerance for convergence. Default: 1e-2.
- Type:
float
- gradConverge#
Whether to check for gradient convergence. Default: False.
- Type:
bool
- min_inner_iter#
Minimum number of inner iterations for fixed penalty convergence. Default: 5.
- Type:
int
- max_restart#
Maximum number of outer iterations that reduce penalties. Default: np.inf.
- Type:
float
- penalty_ratio#
Initial boundary penalty values, as a factor of dualvalue. Default: 1e-2.
- Type:
float
- penalty_reduction#
Factor by which penalty ratio is reduced per outer iteration. Default: 0.1.
- Type:
float
- break_iter_period#
Period of iterations for checking break conditions. Default: 50.
- Type:
int
- verbose#
Verbosity level (0 = silent). Default: 0.
- Type:
int
Attributes Summary
Attributes Documentation
- break_iter_period: int = 50#
- gradConverge: bool = False#
- max_restart: float = inf#
- min_inner_iter: int = 5#
- opttol: float = 0.01#
- penalty_ratio: float = 0.01#
- penalty_reduction: float = 0.1#
- verbose: int = 0#