pystra.calibration.normalized.CodeCalibration#

Use it: User guide · Worked example · Theory

class CodeCalibration(*, live_load_ratios, dead_load_ratios)[source]#

Bases: object

Evaluate candidate factors on the normalized G/P/Q code-design grid.

Parameters:

  • live_load_ratios (Sequence[float]) – Nonempty one-dimensional grids in [0, 1]. They are copied to tuples.

  • dead_load_ratios (Sequence[float]) – Nonempty one-dimensional grids in [0, 1]. They are copied to tuples.

Notes

run returns a new result; no model registration or result cache is retained. Supply the same model with different CodeFactors to compare candidates. Automated factor fitting is deliberately a separate operation.

Methods

design

Solve the normalized factored code equation for resistance scale z.

run

Evaluate candidate factors, retaining every failed grid point.
property live_load_ratios: tuple[float, ...]#

Immutable live-load grid coordinates, in evaluation order.

property dead_load_ratios: tuple[float, ...]#

Immutable dead-load grid coordinates, in evaluation order.

static design(*, live_load_ratio, dead_load_ratio, factors, nominal_values)[source]#

Solve the normalized factored code equation for resistance scale z.

Return type:

float

run(model, factors, *, options=None, evaluator=None, target_beta=None)[source]#

Evaluate candidate factors, retaining every failed grid point.

Parameters:

  • model (NormalizedReliabilityModel) – Probability inputs, nominal values and dependence, copied per run.

  • factors (CodeFactors) – Candidate dimensionless factors in the normalized design equation.

  • options (object) – Evaluator settings; FORMOptions for the default method. DDM is not supported by the built-in normalized limit-state equation.

  • evaluator (ReliabilityEvaluator | None) – Method constructor or result callback; defaults to FORM. A design point is not required. See pystra.assessment.evaluate_reliability.

  • target_beta (float | None) – Finite normal-equivalent index for the same event/reference period.

Returns:

Input snapshot and every design/result, ordered by dead-load ratio then live-load ratio. Failed table estimates and target margins are unavailable; original method records retain their diagnostics.

Return type:

CodeCalibrationResult

Notes

AnalysisError and nonconvergence are recorded. Invalid specifications and programming errors still raise. Beta and target margins are normal-equivalent, including for Student-t standard space.