So does your code solve this problem?
The current mapping applies a scaling factor to boost::gamma_distribution and apparently the resulting empirical cdf converges to the true cdf...
I've looked at your code but I fail to create a test case. I would like to compare the result of your code with other tools like R,
Sure, double checking is a good idea. Among your included directories you should have: #include <boost/statistics/detail/distribution_toolkit/distributions/gamma/include.hpp> OR #include <boost/statistics/detail/distribution_toolkit/distributions/gamma/random.hpp> These have to be in the search path. /sandbox/statistics/non_parametric/ /sandbox/statistics/distribution_common/ /sandbox/statistics/distribution_toolkit/ /sandbox/statistics/random/ /usr/local/boost_1_41_0/ but I have just removed a deprecated dependency so please check out again. To be certain, here's a small subset that I just compiled: #include <boost/typeof/typeof.hpp> #include <boost/random/mersenne_twister.hpp> #include <boost/statistics/detail/distribution_toolkit/distributions/gamma/include.hpp> #include <boost/statistics/detail/distribution_common/meta/random/generator.hpp> void example_gamma(){ namespace stat = boost::statistics::detail; namespace dist = stat::distribution; typedef double val_; typedef boost::mt19937 urng_; typedef boost::math::gamma_distribution<val_> dist_; const val_ shape = 2.0; const val_ scale = 3.0; dist_ d( shape, scale ); urng_ urng; BOOST_AUTO( vg, dist::make_random_generator(urng,d) ); } Please let me know if you have any other issues.