Function relative_error 

pub fn relative_error(numerical: f64, reference: f64) -> f64

Normalized relative error between a numerical value and a reference.

Returns |numerical - reference| / max(|reference|, floor). With the default floor this equals the signed relative error whenever the reference is not degenerate, and collapses to an absolute error when the reference is near zero.

Examples

use solver::validation::relative_error;

assert!((relative_error(1.01, 1.0) - 0.01).abs() < 1e-15);
assert!((relative_error(0.0, 0.0)).abs() < 1e-15);