Module arithmetic 

Extended precision arithmetic for scientific computing.

This module provides double-double arithmetic (~31 decimal digits of precision) using error-free transformations (EFTs) based on Dekker/Knuth algorithms.

Overview

Type	Description
DoubleDouble	Double-double floating-point number (hi + lo)

Functions

Function	Description
dot_dd	Accumulated dot product in double-double precision
sum_dd	Compensated summation (Ogita-Rump-Oishi)

Example

use scirs2_core::arithmetic::{DoubleDouble, sum_dd};

// Catastrophic cancellation handled correctly
let values = [1.0_f64, 1e100, 1.0, -1e100];
let result = sum_dd(&values);
assert!((result.to_f64() - 2.0).abs() < 1e-10);

Re-exports

pub use double_double::dot_dd;

pub use double_double::sum_dd;

pub use double_double::DoubleDouble;

Modules

double_double

Double-double arithmetic (~31 decimal digits of precision).