Enum VarianceReduction 

pub enum VarianceReduction {
    None,
    Antithetic,
    ControlVariate {
        control_fn: fn(f64) -> f64,
        expectation: f64,
    },
    ImportanceSampling {
        sample_fn: fn(&mut SimRng) -> f64,
        likelihood_ratio: fn(f64) -> f64,
    },
    SelfNormalizingIS {
        sample_fn: fn(&mut SimRng) -> f64,
        weight_fn: fn(f64) -> f64,
    },
    Stratified {
        num_strata: usize,
    },
}

Variance reduction technique.

Variants

None

No variance reduction.

Antithetic

Antithetic variates: use (U, 1-U) pairs.

ControlVariate

Control variate with known expectation.

Fields

control_fn: fn(f64) -> f64

Control function.

expectation: f64

Known expectation of control.

ImportanceSampling

Importance sampling with proposal distribution.

Standard importance sampling: E_p[f] = E_q[f * p/q]

Fields

sample_fn: fn(&mut SimRng) -> f64

Sample from proposal distribution q(x).

likelihood_ratio: fn(f64) -> f64

Likelihood ratio p(x)/q(x) where p is target, q is proposal.

SelfNormalizingIS

Self-normalizing importance sampling.

More robust when the normalizing constant is unknown. Uses: E_p[f] ≈ Σ(w_i * f(x_i)) / Σ(w_i)

Fields

sample_fn: fn(&mut SimRng) -> f64

Sample from proposal distribution q(x).

weight_fn: fn(f64) -> f64

Unnormalized weight w(x) ∝ p(x)/q(x).

Stratified

Stratified sampling.

Fields

num_strata: usize

Number of strata.

Trait Implementations

impl Clone for VarianceReduction

fn clone(&self) -> VarianceReduction

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for VarianceReduction

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

Formats the value using the given formatter.

impl Default for VarianceReduction

fn default() -> VarianceReduction

Returns the “default value” for a type.