ppforest2::stats::Uniform Class Reference

Discrete uniform random integer generator over [min, max]. More...

#include <Uniform.hpp>

Public Member Functions
	Uniform (int min, int max)
	Construct a uniform integer generator over [min, max].
std::vector< int >	distinct (int count, RNG &rng)
	Sample without replacement from [min, max].
std::vector< int >	operator() (int count, RNG &rng) const
	Generate multiple uniform random integers (with replacement).
int	operator() (RNG &rng) const
	Generate a single uniform random integer in [min, max].

Detailed Description

Discrete uniform random integer generator over [min, max].

All random integers are produced via Lemire's nearly-divisionless method, which replaces the classical modulo reduction with a multiply-and-shift, falling back to rejection sampling only when needed to eliminate bias. This is both faster and provably unbiased.

The distinct() method builds on this by performing a Fisher-Yates (Knuth) shuffle to sample without replacement — the foundation of bootstrap sampling and random variable selection in forests.

Note

This class is used instead of std::uniform_int_distribution to guarantee identical output across compilers and platforms.

See also

https://arxiv.org/abs/1805.10941

Constructor & Destructor Documentation

Uniform()

ppforest2::stats::Uniform::Uniform	(	int	min,
		int	max )

Construct a uniform integer generator over [min, max].

Parameters

min	Inclusive lower bound (must be ≥ 0).
max	Inclusive upper bound (must be ≥ min).

Member Function Documentation

distinct()

std::vector< int > ppforest2::stats::Uniform::distinct	(	int	count,
		RNG &	rng )

Sample without replacement from [min, max].

Implements the Fisher-Yates (Knuth) shuffle: fills a vector with [min, max], shuffles it using gen_lemire() for each swap, and returns the first count elements. The result is a uniformly distributed subset of size count with no repeats.

This is the method used by DRUniformStrategy and bootstrap sampling, ensuring reproducible variable selection across platforms.

Parameters

count	Number of distinct integers to draw (≤ range size).
rng	Random number generator (pcg32).

Returns

Vector of count distinct integers from [min, max].

operator()() [1/2]

std::vector< int > ppforest2::stats::Uniform::operator()	(	int	count,
		RNG &	rng ) const

Generate multiple uniform random integers (with replacement).

Parameters

count	Number of samples to generate.
rng	Random number generator (pcg32).

Returns

Vector of count i.i.d. integers from [min, max].

operator()() [2/2]

int ppforest2::stats::Uniform::operator()

(

RNG &

rng

)

const

Generate a single uniform random integer in [min, max].

Parameters

rng	Random number generator (pcg32).

Returns

Uniformly distributed integer in [min, max].

---

The documentation for this class was generated from the following file:

• core/src/stats/Uniform.hpp