Struct rusty_machine::stats::dist::gaussian::Gaussian

pub struct Gaussian { /* fields omitted */ }

A Gaussian random variable.

This struct stores both the variance and the standard deviation. This is to minimize the computation required for computing the distribution functions and sampling.

It is most efficient to construct the struct using the from_std_dev constructor.

Methods

impl Gaussian

fn new(mean: f64, variance: f64) -> Gaussian

Creates a new Gaussian random variable from a given mean and variance.

fn from_std_dev(mean: f64, std_dev: f64) -> Gaussian

Creates a new Gaussian random variable from a given mean and standard deviation.

Trait Implementations

impl Debug for Gaussian

fn fmt(&self, __arg_0: &mut Formatter) -> Result

Formats the value using the given formatter.

impl Clone for Gaussian

fn clone(&self) -> Gaussian

Returns a copy of the value.

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

Performs copy-assignment from source.

impl Copy for Gaussian

impl Default for Gaussian

The default Gaussian random variable. This is the Standard Normal random variable.

The defaults are:

• mean = 0
• variance = 1

fn default() -> Gaussian

Returns the "default value" for a type.

impl Distribution<f64> for Gaussian

The distribution of the gaussian random variable.

Accurately computes the PDF and log PDF. Estimates the CDF accurate only to 0.003.

fn pdf(&self, x: f64) -> f64

The pdf of the normal distribution

Examples

use rusty_machine::stats::dist::Gaussian;
use rusty_machine::stats::dist::Distribution;
use rusty_machine::stats::dist::consts;

let gauss = Gaussian::default();

let lpdf_zero = gauss.pdf(0f64);

// The value should be very close to 1/sqrt(2 * pi)
assert!((lpdf_zero - (1f64/consts::SQRT_2_PI).abs()) < 1e-20);

fn logpdf(&self, x: f64) -> f64

The log pdf of the normal distribution.

Examples

use rusty_machine::stats::dist::Gaussian;
use rusty_machine::stats::dist::Distribution;
use rusty_machine::stats::dist::consts;

let gauss = Gaussian::default();

let lpdf_zero = gauss.logpdf(0f64);

// The value should be very close to -0.5*Ln(2 * pi)
assert!((lpdf_zero + 0.5*consts::LN_2_PI).abs() < 1e-20);

fn cdf(&self, x: f64) -> f64

Rough estimate for the cdf of the gaussian distribution. Accurate to 0.003.

Examples

use rusty_machine::stats::dist::Gaussian;
use rusty_machine::stats::dist::Distribution;

let gauss = Gaussian::new(10f64, 5f64);
let cdf_mid = gauss.cdf(10f64);

assert!((0.5 - cdf_mid).abs() < 0.004);

A slightly more involved test:

use rusty_machine::stats::dist::Gaussian;
use rusty_machine::stats::dist::Distribution;

let gauss = Gaussian::new(10f64, 4f64);
let cdf = gauss.cdf(9f64);

assert!((0.5*(1f64 - 0.382924922548) - cdf).abs() < 0.004);

impl Sample<f64> for Gaussian

fn sample<R: Rng>(&mut self, rng: &mut R) -> f64

Generate a random value of Support, using rng as the source of randomness.

impl IndependentSample<f64> for Gaussian

fn ind_sample<R: Rng>(&self, rng: &mut R) -> f64

Generate a random value.