Struct Welford 

pub struct Welford { /* private fields */ }

Online (streaming) mean and variance via Welford’s algorithm.

Naive “sum of squares minus square of sum” variance loses precision catastrophically once the values share a large common offset (see this module’s tests for a worked example); Welford’s algorithm instead updates the mean and the sum of squared deviations from the running mean one sample at a time, and never needs to retain the samples themselves – exactly the shape an adaptive Monte Carlo driver needs: an unbounded trial count held in O(1) memory.

Implementations

impl Welford

pub fn new() -> Self

A fresh accumulator with no observations.

pub fn push(&mut self, x: f64)

Folds one more observation into the running mean and variance.

pub fn count(&self) -> u64

Number of observations folded in so far.

pub fn mean(&self) -> f64

Running mean. 0.0 when no observations have been pushed yet.

pub fn sample_variance(&self) -> f64

Unbiased (Bessel-corrected) sample variance, m2 / (n - 1).

0.0 when fewer than two observations have been pushed – a single point has no defined spread.

pub fn population_variance(&self) -> f64

Population variance, m2 / n. 0.0 when no observations have been pushed yet.

pub fn sample_std(&self) -> f64

Sample standard deviation, sqrt(sample_variance()).

Trait Implementations

impl Clone for Welford

fn clone(&self) -> Welford

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Welford

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

Formats the value using the given formatter.

impl Default for Welford

fn default() -> Welford

Returns the “default value” for a type.