On 04/09/2017 12:23 PM, Marc Glisse wrote:
On Sun, 9 Apr 2017, Oliver Kullmann wrote:
Hello,
I thought that with the facilities from <random> we had good and
well-defined ones -- but alas, only recently I realised that only say
the Mersenne twister is well-defined in the standard, but roughly
everything else is up to the implementer.
Since reproducibility is crucial for us (as with most scientific
applications), thus we need to implement ourselves the algorithms for
1. std::seed_seq
2. std::uniform_int_distribution
3. std::bernoulli_distribution (only for default p=0.5).
Easiest would be to copy (and specialise) the gcc-implementations.
It's worth keeping in mind that copying libstdc++ code will have
licensing implications. A different caveat is that conforming C++
programs can define explicit specializations of standard library
templates only for user-defined types. I.e., it's not valid to
define one's own std::uniform_int_ditribution class template or
specialization of it on a fundamental type such as int.
I searched for it in the source-code, but couldn't find it:
Where do I find the implementations for <random> ?
Option 1)
First you find the file called "random" (/usr/include/c++/6/random
here), then you notice that it contains
#include <bits/random.h>
#include <bits/opt_random.h>
#include <bits/random.tcc>
so you look for those files (and recursively if needed)
Option 2)
You compile a file that contains #include <random> using the -E flag,
and look for bernoulli_distribution or whatever in the output.
Option 3)
Get the sources for gcc.
grep -r bernoulli_distribution /the/sources
Another option is to browse the sources online:
https://gcc.gnu.org/viewcvs/gcc/trunk/libstdc%2B%2B-v3/include/bits/random.h?view=markup
Martin
They all point to bits/random.h (with one function implemented in
bits/random.tcc), with older versions in tr1/.
