In the program, often in different classes are generated random numbers. So I want to create a class that returns a single instance of the generator std::mt19937. I also take into account that some compilers do not work with std::random_device (To do this, check the value of entropy). I have created a class singleton.
#include <iostream>
#include <random>
#include <chrono>
class RandomGenerator
{
public:
static RandomGenerator& Instance() {
static RandomGenerator s;
return s;
}
std::mt19937 get();
private:
RandomGenerator();
~RandomGenerator() {}
RandomGenerator(RandomGenerator const&) = delete;
RandomGenerator& operator= (RandomGenerator const&) = delete;
std::mt19937 mt;
};
RandomGenerator::RandomGenerator() {
std::random_device rd;
if (rd.entropy() != 0) {
mt.seed(rd());
}
else {
auto seed = std::chrono::high_resolution_clock::now().time_since_epoch().count();
mt.seed(seed);
}
}
std::mt19937 RandomGenerator::get() {
return mt;
}
int main() {
std::mt19937 &mt = RandomGenerator::Instance().get();
std::uniform_real_distribution<double> dist(0.0, 1.0);
for (std::size_t i = 0; i < 5; i++)
std::cout << dist(mt) << "\n";
std::cout << "\n";
std::mt19937 &mt2 = RandomGenerator::Instance().get();
std::uniform_real_distribution<double> dist2(0.0, 1.0);
for (std::size_t i = 0; i < 5; i++)
std::cout << dist2(mt2) << "\n";
return 0;
}
But when I get out of class generator std::mt19937, random numbers begin to repeat. How to avoid it?
0.389459
0.68052
0.508421
0.0758856
0.0137491
0.389459
0.68052
0.508421
0.0758856
0.0137491
P.S. Is there a better way to initialize the generator than time?
Solution
Tested this under the following compilers: Visual Studio, MinGW, DevC++.
#include <iostream>
#include <random>
#include <chrono>
class RandomGenerator
{
public:
static RandomGenerator& Instance() {
static RandomGenerator s;
return s;
}
std::mt19937 & get();
private:
RandomGenerator();
~RandomGenerator() {}
RandomGenerator(RandomGenerator const&) = delete;
RandomGenerator& operator= (RandomGenerator const&) = delete;
std::mt19937 mt;
};
RandomGenerator::RandomGenerator() {
std::random_device rd;
if (rd.entropy() != 0) {
mt.seed(rd());
}
else {
auto seed = std::chrono::high_resolution_clock::now().time_since_epoch().count();
mt.seed(seed);
}
}
std::mt19937 & RandomGenerator::get() {
return mt;
}
int main() {
std::mt19937 &mt = RandomGenerator::Instance().get();
std::uniform_real_distribution<double> dist(0.0, 1.0);
for (std::size_t i = 0; i < 5; i++)
std::cout << dist(mt) << "\n";
std::cout << "\n";
std::mt19937 &mt2 = RandomGenerator::Instance().get();
std::uniform_real_distribution<double> dist2(0.0, 1.0);
for (std::size_t i = 0; i < 5; i++)
std::cout << dist2(mt2) << "\n";
return 0;
}
std::mt19937 get(); returns a copy. Every time you call get() you make a copy of the initial state of the engine. mt19937 is a pseudo-random engine, each state produces a predetermined sequence. If the state of two instances are identical, they will produce the same sequence. Make the function return a reference so that the state of the singleton instance is updated with each new number generated.
std::mt19937 & RandomGenerator::get() {
return mt;
}