Struct SimulationBasedCalibration 

pub struct SimulationBasedCalibration {
    pub n_reps: u32,
    pub n_draws: usize,
    pub seed: u64,
}

Simulation-based calibration ranks for a scalar posterior functional.

For each replicate: draw θ* from the prior predictive, simulate data, refit, and record the rank of θ* among posterior draws of the primary effect.

Fields

n_reps: u32

Number of SBC replicates.

n_draws: usize

Draws per refit.

seed: u64

RNG seed.

Implementations

impl SimulationBasedCalibration

pub fn new(n_reps: u32) -> Self

Construct.

pub fn check( &self, estimator: &BayesianGComputationAte, problem: &PreparedBayesianProblem, identification: IdentificationStatus, workspace: &mut BayesianGCompWorkspace, ctx: &ExecutionContext, ) -> Result<SbcReport, ValidationError>

Run SBC: draw θ from the prior, simulate y from the prior predictive under the fixed design matrix, refit the Bayesian g-computation estimator, and rank the true ATE among posterior effect draws.

Errors

Fit failures.

pub fn to_report( &self, report: &SbcReport, original_ate: f64, ) -> RefutationReport

Convert to a refutation report.

Passes only when both hold:

• the mean rank fraction is in [0.35, 0.65] (catches gross location bias), and
• the χ² uniformity statistic over the 10 rank bins is below SBC_CHI2_CRITICAL_9DF_P99 (catches symmetric-about-0.5 U/M-shaped rank distributions — overdispersed / underdispersed posteriors — that a mean-only band cannot see because they average out to ≈0.5).

Trait Implementations

impl Clone for SimulationBasedCalibration

fn clone(&self) -> SimulationBasedCalibration

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SimulationBasedCalibration

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for SimulationBasedCalibration

fn default() -> Self

Returns the “default value” for a type.