Module rand

Random floating point number generation with the rand crate.

There are two new distributions for generating random floats. The first one is Uniform01, which supports generating floats between 0 and 1. This is also the underlying implementation of the builtin rand distributions Standard, Open01, OpenClosed01. The other one is UniformFBig, which supports generating floats in a certain range. This is also the backend for the SampleUniform trait.

Examples

use dashu_float::rand::Uniform01;

type FBig = dashu_float::FBig;

// generate FBigs in a [0, 1) or [0, 1] with a given precision
let a: FBig = thread_rng().sample(Uniform01::new(10));
let b: FBig = thread_rng().sample(Uniform01::new_closed(10));
let c: FBig = thread_rng().gen(); // the default distribution generates in [0, 1)
assert!(a >= FBig::ZERO && a < FBig::ONE);
assert!(b >= FBig::ZERO && b <= FBig::ONE);
assert!(c >= FBig::ZERO && c < FBig::ONE);

// generate FBigs in a range
let a = thread_rng().gen_range(FBig::from(3)..FBig::from(10));
let b = thread_rng().sample(Uniform::new(FBig::from(-5), &a));
assert!(a >= FBig::from(3) && a < FBig::from(10));
assert!(b >= FBig::from(-5) && b < a);

Precision and Rounding

The precision of a float generated by different distributions is explained below:

• Uniform01 will generate floats with the precision decided by the constructor.
• Standard, Open01, OpenClosed01 will generate floats with the max precision such that the significand fits in a DoubleWord.
• UniformFBig (and therefore Uniform) will generate floats the precision being the maximum between the interval boundaries.

The rounding of the FBig type doesn’t affect the number generation process.

Structs

Uniform01

A uniform distribution between 0 and 1. It can be used to replace the Standard, Open01, OpenClosed01 distributions from the rand crate when you want to customize the precision of the generated float number.

UniformFBig

The back-end implementing UniformSampler for FBig (and DBig).