boost discrete uniform distribution

Question

I have a variety of distributions from which I draw samples, pdf and cdf. For polymorphic reasons I am using a uniform_distribution instead of uniform_int_distribution.

The following returns me a floating value between 10 and 20 instead of integral values:

 typedef uniform_distribution<double,
                policy<discrete_quantile<integer_round_outwards>>> uniform_round_outwards;

 uniform_round_outwards  _uniformObject(10,20);

 x = quantile(_uniformObject, p); 

Is the policy being applied at all?

Answer 1

For polymorphic reasons I am using a uniform_distribution instead of uniform_int_distribution.

The Math library doesn't have a uniform_int_distribution, making that a weird statement.

The following returns me a floating value between 10 and 20 instead of integral values:

That distribution is a continuous distribution. The docs only fleetingly refer to Wikipedia to say "There is also a discrete uniform distribution.", but they fail to explicitly state whether or not such a thing is implemented.

Looking at the source shows that the policy is not being used:

  template <class RealType, class Policy>
  inline RealType quantile(const uniform_distribution<RealType, Policy>& dist, const RealType& p)
  {
    RealType lower = dist.lower();
    RealType upper = dist.upper();
    RealType result = 0; // of checks
    if(false == detail::check_uniform("boost::math::quantile(const uniform_distribution<%1%>&, %1%)",lower, upper, &result, Policy()))
    {
      return result;
    }
    if(false == detail::check_probability("boost::math::quantile(const uniform_distribution<%1%>&, %1%)", p, &result, Policy()))
    {
      return result;
    }
    if(p == 0)
    {
      return lower;
    }
    if(p == 1)
    {
      return upper;
    }
    return p * (upper - lower) + lower;
  }

My conclusion is that the discrete uniform distribution is not implemented in this class.