robust-rs-core 0.1.0

Core abstractions for robust statistics: loss functions, robust scale and influence-function theory.
Documentation 
1
2
3
4
5
6
7
8
9
10

//! Efficiency relative to the Gaussian MLE.

use super::variance::asymptotic_variance;
use crate::rho::RhoFunction;

/// Asymptotic efficiency at the Gaussian: `(E[ψ'])² / E[ψ²]`; the reciprocal of
/// the asymptotic variance, since the location MLE attains variance 1.
pub fn gaussian_efficiency(rho: &dyn RhoFunction, quad_points: usize) -> f64 {
    1.0 / asymptotic_variance(rho, quad_points)
}