pystra.calibration.factors.solve_designs#
- solve_designs(problem, *, target_beta, method='root', initial_value=None, tolerance=0.0001, max_evaluations=100, bracket=None, options=None, evaluator=None)[source]#
Solve each case to a target and retain status and residuals.
- Parameters:
problem (
FactorCalibrationProblem) – Explicit cases, roles, nominal values and design parameter; copied.target_beta (
float) – Finite normal-equivalent target for the same event/reference period.method (
str) – Root solving or normal-space design-point projection.initial_value (
float|None) – Starting design parameter; defaults to the problem’s common constant.tolerance (
float) – Positive root-solver tolerance and maximum accepted beta residual.max_evaluations (
int) – Reliability evaluation budget per case, including final verification.bracket (
tuple[float,float] |None) – Increasing two-endpoint bracket for Brent root solving. Without it, the root method uses fsolve with full diagnostics.options (
object) – Evaluator-specific settings; FORMOptions for the default FORM.evaluator (
ReliabilityEvaluator|None) – Method constructor or result callback, default FORM. Root solving requires normal-equivalent estimates. Alpha projection additionally requires a named normal-space design point/direction and an analysis exposing its matching transform.u_to_x and model.
- Returns:
Every case, its design parameter, reliability record, target residual and solve status. Inner convergence, outer success and the final residual tolerance are all required. No unsuccessful solve is accepted.
- Return type:
- Raises:
ValueError – Invalid solve settings or missing design-point/projection capabilities.