libcruft-util/geom/sample.hpp

/*
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 *
 * Copyright 2015-2018 Danny Robson <danny@nerdcruft.net>
 */

#ifndef __UTIL_GEOM_SAMPLE_HPP
#define __UTIL_GEOM_SAMPLE_HPP

#include "../coord/fwd.hpp"
#include "../extent.hpp"
#include "ops.hpp"

#include <cstddef>

#include <random>

///////////////////////////////////////////////////////////////////////////////
namespace util::geom {
    /// a function object that selects a uniformly random point inside a shape
    /// using a provided random generator. the point will lie within the shape,
    /// inclusive of boundaries.
    ///
    /// may be specialised for arbitrary shapes but uses rejection sampling
    /// as a safe default. this implies that execution may not take a constant
    /// time.
    ///
    /// \tparam S coordinate type dimension
    /// \tparam T value type for the shape/coordinate
    /// \tparam K the shape type to test
    /// \tparam G a UniformRandomBitGenerator, eg std::min19937
    template <
        size_t S,
        typename T,
        template <size_t,typename> class K,
        typename G
    >
    struct sampler {
        static point<S,T>
        fn (K<S,T> k, G &g)
        {
            auto b = bounds (k);

            while (true) {
                auto p = sample (b, g);
                if (intersects (k, p))
                    return p;
            }
        }
    };


    ///////////////////////////////////////////////////////////////////////////
    /// a convenience function that calls sample::fn to select a random point
    /// in a provided shape.
    template <
        size_t S,
        typename T,
        template <size_t,typename> class K,
        typename G  // random generator
    >
    point<S,T>
    sample (K<S,T> k, G &g)
    {
        return sampler<S,T,K,G>::fn (k, g);
    }


    std::vector<util::point2f>
    poisson_sample (util::extent2i, float distance, int samples);


    namespace surface {
        // a generator of samples that lie on the surface of a shape
        template <typename ShapeT>
        class sampler;

        template <typename ShapeT>
        sampler (ShapeT const&) -> sampler<std::decay_t<ShapeT>>;


        //---------------------------------------------------------------------
        template <size_t S, typename T>
        class sampler<util::extent<S,T>> {
        public:
            sampler (util::extent<S,T> _target):
                target (_target)
            { ; }

            template <typename GeneratorT>
            util::point<S,T>
            operator() (GeneratorT &&gen) const noexcept
            {
                util::point<S,T> p;

                for (size_t i = 0; i < S; ++i) {
                    if constexpr (std::is_floating_point_v<T>) {
                        p[i] = std::uniform_real_distribution<T> (0, target[i]) (gen);
                    } else {
                        CHECK_GE (target[i], T{1});
                        p[i] = std::uniform_int_distribution<T> (0, target[i] - 1) (gen);
                    }
                }

                return p;
            }

        private:
            util::extent<S,T> target;
        };

        /// approximate a poisson disc sampling through mitchell's best candidate.
        ///
        /// try to keep adding a new point to a set. each new point is the
        /// best of a set of candidates. the 'best' is the point that is
        /// furthest from all selected points.
        ///
        /// \return a vector of the computed points
        template <typename SamplerT, typename DistanceT, typename GeneratorT>
        auto
        poisson (SamplerT const &target,
                 GeneratorT &&gen,
                 DistanceT &&minimum_distance,
                 size_t candidates_count)

        {
            using point_type = decltype (target (gen));
            using value_type = typename point_type::value_type;

            std::vector<point_type> selected;
            std::vector<point_type> candidates;

            // prime the found elements list with an initial point we can
            // perform distance calculations on
            selected.push_back (target (gen));

            // keep trying to add one more new point
            while (1) {
                // generate a group of candidate points
                candidates.clear ();
                std::generate_n (
                    std::back_inserter (candidates),
                    candidates_count,
                    [&] (void) {
                        return target (gen);
                    }
                );

                // find the point whose minimum distance to the existing
                // points is the greatest (while also being greater than the
                // required minimum distance);
                auto best_distance2 = std::numeric_limits<value_type>::lowest ();
                size_t best_index;

                for (size_t i = 0; i < candidates.size (); ++i) {
                    auto const p = candidates[i];
                    auto d2 = std::numeric_limits<value_type>::max ();

                    // find the minimum distance from this candidate to the
                    // selected points
                    for (auto q: selected)
                        d2 = util::min (d2, util::distance2 (p, q));

                    // record if it's the furthest away
                    if (d2 > best_distance2 && d2 > util::pow (minimum_distance (p), 2)) {
                        best_distance2 = d2;
                        best_index = i;
                    }
                }

                // if we didn't find a suitable point then we give up and
                // return the points we found, otherwise record the best point
                if (best_distance2 <= 0)
                    break;

                selected.push_back (candidates[best_index]);
            }

            return selected;
        }
    }
}

#endif