std::mersenne_twister_engine<UIntType,w,n,m,r,a,u,d,s,b,t,c,l,f>::mersenne_twister_engine

From cppreference.com
C++
Numerics library
Pseudo-random number generation
Random number distributions
Uniform distributions
(C++11)
(C++11)
(C++11)
Bernoulli distributions
(C++11)
(C++11)
(C++11)
(C++11)
Poisson distributions
(C++11)
(C++11)
(C++11)
(C++11)
(C++11)
Normal distributions
(C++11)
(C++11)
(C++11)
(C++11)
(C++11)
(C++11)
Sampling distributions
(C++11)
(C++11)
(C++11)
std::mersenne_twister_engine
Member functions
mersenne_twister_engine::mersenne_twister_engine
Generation
Characteristics
Non-member functions
(C++11)(C++11)(until C++20)
(C++11)(C++11)
mersenne_twister_engine( ) : mersenne_twister_engine(default_seed) { }
(1) (since C++11)
explicit mersenne_twister_engine( result_type value );
 (2) (since C++11)
template < class SeedSeq >
explicit mersenne_twister_engine( SeedSeq& seq ) ;
(3) (since C++11)
mersenne_twister_engine( const mersenne_twister_engine& other );
 (4) (since C++11)
(implicitly declared)

Constructs the pseudo-random number engine.

1) The default constructor.
  • If the default-constructed engine is of type std::mt19937, the 10000th consecutive invocation of it produces the value 4123659995.
  • If the default-constructed engine is of type std::mt19937_64, the 10000th consecutive invocation of it produces the value 9981545732273789042
2) Constructs the engine with a seed value value. Given 2w
 as p, the engine's initial state is determined as follows:
  1. Sets X-n to value % p
  2. For each integer i in [ i - n ,  -1 ] , sets Xi to [f·(Xi-1 xor (Xi-1 rshift (w-2)))+i mod n] mod p, where xor and rshift stand for built-in bitwise XOR and bitwise right-shift
3) Constructs the engine with a seed sequence seq. Given std::size_t (w / 32 ) + 1 as k, the engine's initial state is determined as follows:
  1. Creates an invented array object a of length n * k.
  2. Calls seq.generate (a + 0, a + n * k)
  3. For each integer i in [ -n ,  -1 ] , sets Xi to (∑k-1
    j=0     ak(i+n)+j·232j
    ) mod 2w
  4. If the most significant w − r bits of X-n are zero, and if each of the other resulting Xi is ​0​, changes X-n to 2w-1
This overload participates in overload resolution only if SeedSeq meets the requirements of SeedSequence
4) The copy constructor. Upon construction, *this == other is true

Parameters

value - seed value to use in the initialization of the internal state
seq - seed sequence to use in the initialization of the internal state

Complexity

1,2) O(n).
3) Same as the complexity of the seq.generate call.
4) O(n).

Exceptions

3) If SeedSeq is not std::seed_seq, throws the exceptions thrown by the seq.generate

Example

This section is incomplete
Reason: demos for overloads (2-4) required
Run this code
#include <cassert>
#include <random>
 
int main()
{
    std::mt19937 gen32; // overload (1)
    std::mt19937_64 gen64; // overload (1)
    gen32.discard(10000 - 1);
    gen64.discard(10000 - 1);
    assert(gen32() == 4123659995);
    assert(gen64() == 9981545732273789042ull);
}

Defect reports

The following behavior-changing defect reports were applied retroactively to previously published C++ standards.

DR Applied to Behavior as published Correct behavior
LWG 2181 C++11 overload (3) would not throw even if the seq.generate call throws propagates the exception
P0935R0 C++11 the default constructor was explicit made implicit

See also

 sets the current state of the engine
(public member function)
Retrieved from "https://en.cppreference.com/mwiki/index.php?title=cpp/numeric/random/mersenne_twister_engine/mersenne_twister_engine&oldid=177250