Struct WeightedMoments 

pub struct WeightedMoments { /* private fields */ }

Bounded exact sufficient statistics for one ordered scalar moment.

Storage is independent of occurrence count. Every binary64 input is exact; only requested results and first derivatives are rounded to binary64. The shared rational bound applies to every intermediate operation. No sample identity, independence, population or unbiasedness claim follows from this arithmetic owner; the caller must bind those meanings separately.

Implementations

§

impl WeightedMoments

Signature links:WeightedMoments

pub fn new( moment: ProbabilityMoment, types: &[ValueType], ) -> Result<WeightedMoments, Diagnostic>

Signature links:newProbabilityMomentValueTypeResultWeightedMomentsDiagnostic

Establish the ordered integrand types before consuming occurrences.

Errors

Rejects types or arity outside the moment’s admitted scalar profile.

pub fn push( &mut self, row: &[ValueLiteral], weight: f64, ) -> Result<(), Diagnostic>

Signature links:pushValueLiteralf64Result()Diagnostic

Consume one occurrence, including duplicate values and zero weights.

Errors

Rejects wrong types, nonfinite/negative weights, count overflow or an exhausted arithmetic bound. Failure leaves the accumulator unchanged.

pub fn value(&self, correction: usize) -> Result<ValueLiteral, Diagnostic>

Signature links:valueusizeResultValueLiteralDiagnostic

Exact normalized moment, with an explicit finite-count correction.

Zero correction uses the total weight. A positive correction requires equal positive weights and divides centered moments by n-correction, instead of n. This defines arithmetic only, never an IID claim.

Errors

Rejects empty/zero mass, invalid correction, or unrepresentable output.

pub fn sum(&self) -> Result<ValueLiteral, Diagnostic>

Signature links:sumResultValueLiteralDiagnostic

Unnormalized weighted sum of a single quantity, with its original units.

Errors

Rejects empty/zero mass, paired/centered moments or unrepresentable output.

pub fn mean_standard_error(&self) -> Result<ValueLiteral, Diagnostic>

Signature links:mean_standard_errorResultValueLiteralDiagnostic

Estimated standard error of an equal-weight mean, in the original units.

Computes sqrt(sum((x-mean)^2)/(n*(n-1))) with an exact radicand and one correctly rounded square root. The caller must establish IID sampling and assume a finite population second moment; this is arithmetic only.

Errors

Rejects other moments, unequal/nonpositive weights, n < 2 and range errors.

pub fn derivative( &self, row: &[ValueLiteral], weight: f64, correction: usize, ) -> Result<Vec<f64>, Diagnostic>

Signature links:derivativeValueLiteralf64usizeResultVecf64Diagnostic

Derivative of this normalized finite statistic with respect to one row.

Weights are fixed. Results are coherent-SI coordinate coefficients in ordered integrand order; coefficient j has statistic-unit/input-unit-j. The caller owns membership and must supply an actual accepted occurrence.

Errors

Rejects invalid rows, weights, normalization or unrepresentable coefficients.

Trait Implementations

§

impl Clone for WeightedMoments

Signature links:CloneWeightedMoments

§

fn clone(&self) -> WeightedMoments

Signature links:cloneWeightedMoments

Returns a duplicate of the value.

§

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

Signature links:unstableSourceclone_from

Performs copy-assignment from source.

§

impl Debug for WeightedMoments

Signature links:DebugWeightedMoments

§

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

Signature links:fmtFormatterResult()Error

Formats the value using the given formatter.