Struct MclmcStats 

pub struct MclmcStats<P: HasDims, H: Storable<P>, A: Storable<P>, Pt: Storable<P>> {
    pub chain: u64,
    pub draw: u64,
    pub num_steps: u64,
    pub energy_change: f64,
    pub log_weight: f64,
    pub tuning: bool,
    pub hamiltonian: H,
    pub adapt: A,
    pub point: Pt,
    pub average_step_size: f64,
    pub divergence: DivergenceStats,
    /* private fields */
}

Per-draw sampler statistics for MclmcChain.

Fields

chain: u64

draw: u64

num_steps: u64

energy_change: f64

log_weight: f64

Unnormalized log importance weight for this draw: log_weight = log(step_size_factor) − energy_change.

step_size_factor is the factor applied to the base step size for the final leapfrog step (1.0 when no dynamic retry occurred). log(factor) corrects for varying sampling density under dynamic step size reduction; it is 0.0 (no correction) when the step is taken at full size. The −energy_change term corrects for integration error. See Robnik & Seljak (2023), arXiv:2212.08549.

tuning: bool

hamiltonian: H

adapt: A

point: Pt

average_step_size: f64

divergence: DivergenceStats

Trait Implementations

impl<P: Debug + HasDims, H: Debug + Storable<P>, A: Debug + Storable<P>, Pt: Debug + Storable<P>> Debug for MclmcStats<P, H, A, Pt>

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

Formats the value using the given formatter.

impl<P: HasDims, H: Storable<P>, A: Storable<P>, Pt: Storable<P>> Storable<P> for MclmcStats<P, H, A, Pt>

fn names(parent: &P) -> Vec<&str>

fn item_type(parent: &P, item: &str) -> ItemType

fn dims<'a>(parent: &'a P, item: &str) -> Vec<&'a str>

fn event_dim(parent: &P, item: &str) -> Option<&'static str>

Return the name of the primary dimension for the given field, or None if the field uses the default draw dimension.

fn get_all<'a>(&'a mut self, parent: &'a P) -> Vec<(&'a str, Option<Value>)>