Each simulation of normal distribution is the same (C++)

问题

I write a code to simulate a normal distribution in C++. But each time it seems the result is the same. My question is what't the reason of this phenomenon and How to fix it? I never have this problem with Python. Any references are very appreciated.

// Simulation.cpp : Defines the entry point for the console application.
#include "stdafx.h"
#include <iostream>
#include<random>

void main(){

     // create default engine as source of randomness
     // The maxtime we do expriements is 10000
     // Record to the sum of the maxtimes sample
    std::default_random_engine dre; 
    const int maxtimes = 10000;
    double sum = 0.0 ;
    // Generate the normal distribution witn mean 0 and variaiton 1.
    std::normal_distribution<double> distribution(0.0, 1.0);
    // Output the result and Record their sum. 
    for( int i=0; i<maxtimes; ++i)
      {

        double x = distribution(dre);
        std::cout << x << ":";
        sum +=x; 
        x =0.0; 
      }
    std::cout<<std::endl;
    std::cout <<" The average sum is: " << sum/10000 <<std::endl; 
  }

My code runs in visual C++ 2010.

回答1:

You're constructing the default_random_engine from the same seed every time: Since you aren't giving it a seed to construct with, it simply uses the default, which is the same every run, so you get the same "random" numbers each run. http://www.cplusplus.com/reference/random/linear_congruential_engine/linear_congruential_engine/

Use random_device to seed the generator.

std::default_random_engine dre(std::random_device()()); 

回答2:

Try:

std::random_device mch;
std::default_random_engine generator(mch());
std::normal_distribution<double> distribution(0.0, 1.0);

来源：https://stackoverflow.com/questions/22921853/each-simulation-of-normal-distribution-is-the-same-c