[Phys-L] RNG "quality"?

[Bernard Cleyet <bernard@cleyet.org>]

using  (JD suggestion)


#include <random>

using namespace std;



 random_device seed;
    mt19937 gen(seed());
    

in loop:  


 normal_distribution<double> distribution(0.00,0.20);
      
      noise = distribution(gen);
  

for 1e6 iterations:  

mean:  ~ -9 E-5 and 
SD:      ~ 0.19990  

While 1e4 is

0.0024  and 

  2.0244  

Others observe the expected changes.  


Is this gud?  

bc;  now incorporating into pendulum simulation, and discovered OS X uses clang not GCC