Random number Engines and Distributions

Overview

CRT provides limited capability of generating random numbers based on a seed. In C++ 11 numerous random number engines and distributions were introduced.

Details

The mechanism is a two step process. The first step involves generating uniformly distributed random values between a min and max range by the random value generators. The second step involves using different distribution mechanisms such as uniform, Normal or Binomial to achieve randomness.

Except for random_device, all the random value generators produce pseudo random values, meaning the numbers are generated using a pre determined way. Also seed can be uses to initialize these engines.

seed_seq

Pseudo random numbers are repetitive when used without a seed. A seed is an unique number that can be used. For example, the the current time in nanoseconds since the system start.

The seed_seq class  provides a way to seed random number engines or a generator that requires a lot of entropy. It uses a initial seed sequence of integer values to produce a series of unsigned integer values with 32 significant bits that can be used to seed a pseudo-random generator engine.

member types

Following properties are defined.

Name	Definition
uint_least32_t result_type	The type of the numbers generated

Constructor

Name

Description

seed_seq() template<class T>  seed_seq (initializer_list<T> il) template<class InputIterator> seed_seq (InputIterator first, InputIterator last)

The default constructor with no input. The initial seed sequence will be empty. Constructs with the values in the initialization list. The initial seed sequence will contain values from il. Constructs with the values in the range. [first, last). The initial seed sequence will contain values from the range. Example //1 seed_seq seed; //2 seed_seq seed2 = {21,9,2018}; //3 string foo = "Khrisha Rao"; seed_seq seed3 (foo.begin(),foo.end());

Methods

Name	Description
void generate (RandomAccessIterator first, RandomAccesIterator last)	Populates the output range [first, last) with 32-bit unsigned integer values that are seeds, based on the initial seed sequence. Example seed_seq seed {21,9,2018}; unsigned seeds[5]; seed.generate(seeds,seeds+5); //prints 565048846 -2000421301 2106023878 263474872 1635843284 copy ( seeds,seeds+5, ostream_iterator<unsigned> (cout," ") );
size_t size()	Returns the size of the initial seed sequence. Example     seed_seq seed {21,9,2018}; //prints 3 cout << seed.size();
void param (OutputIterator dest)	Copies the internal sequence into dest.  Example     seed_seq seed {21,9,2018}; //prints 21 9 2018 seed.param(ostream_iterator<unsigned> (cout," ") );

Generators and Engines➹

Multiple classes are provided to generate pseudo random numbers using mathematical functions. Also a truly random device is also provided.

generate_canonical()

Generates a random floating point number in range [​0​, 1) using random number engines.

It's light weight compared to distributions.

Syntax

//RealType:A floating-point type.
//n:Produce random numbers with  n bits of randomness
//URNG:A uniform random number generator class.
template <class RealType, size_t n, class URNG>  
RealType generate_canonical (URNG& g)

:

    random_device rd;
    mt19937 gen(rd());
    //prints 0.335205 0.10016 0.721013 
    for (int n = 0; n < 3; ++n)
        cout << generate_canonical<double, 4>(gen) << ' ';

Distributions➹

Multiple classes are provided to generate pseudo random numbers sequences. These classes use a pseudo random number engine to generate numbers and later distribute them into sequences using mathematical functions.