Struct WeightingDoSampler 

pub struct WeightingDoSampler {
    pub treatment: VariableId,
    pub outcome: VariableId,
}

Weighting do-sampler for hard do(T=t).

• Root treatment: empirical outcomes among units with T ≈ t.
• Confounded continuous treatment: Horvitz–Thompson with a shrinking Gaussian kernel on the treatment margin (Silverman bandwidth) over the fitted conditional density f(T∣parents): wᵢ ∝ Kₕ(Tᵢ − t) / f(Tᵢ∣parents). Hard interventions have a Dirac interventional law, so the kernel is the localization numerator (there is no separate lp_do term).
• Non-density / discrete mechanisms: kernel localization alone (exact match when the bandwidth collapses on tied support).

Fields

treatment: VariableId

Treatment variable.

outcome: VariableId

Outcome variable.

Implementations

impl WeightingDoSampler

pub fn new(treatment: VariableId, outcome: VariableId) -> Self

Construct.

pub fn estimate( &self, model: &CompiledCausalModel, data: &TabularData, treatment_value: f64, _ctx: &ExecutionContext, ) -> Result<DoSampleResult, ModelError>

Estimate E[Y | do(T=t)] via matching / Horvitz–Thompson on fitted model densities.

Errors

Missing columns or unfitted treatment mechanism.

pub fn weighted_mean(result: &DoSampleResult) -> f64

Weighted mean of the sampler result.

Trait Implementations

impl Clone for WeightingDoSampler

fn clone(&self) -> WeightingDoSampler

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for WeightingDoSampler

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

Formats the value using the given formatter.