diff options
Diffstat (limited to 'src')
6 files changed, 115 insertions, 80 deletions
diff --git a/src/Subsampling/example/example_choose_n_farthest_points.cpp b/src/Subsampling/example/example_choose_n_farthest_points.cpp index 533aba74..fc9ea7a6 100644 --- a/src/Subsampling/example/example_choose_n_farthest_points.cpp +++ b/src/Subsampling/example/example_choose_n_farthest_points.cpp @@ -19,7 +19,7 @@ int main(void) { K k; std::vector<Point_d> results; - Gudhi::subsampling::choose_n_farthest_points(k, points, 100, std::back_inserter(results)); + Gudhi::subsampling::choose_n_farthest_points(k, points, 100, Gudhi::subsampling::random_starting_point, std::back_inserter(results)); std::cout << "Before sparsification: " << points.size() << " points.\n"; std::cout << "After sparsification: " << results.size() << " points.\n"; diff --git a/src/Subsampling/example/example_custom_kernel.cpp b/src/Subsampling/example/example_custom_kernel.cpp index 25b5bf6c..2719ee90 100644 --- a/src/Subsampling/example/example_custom_kernel.cpp +++ b/src/Subsampling/example/example_custom_kernel.cpp @@ -54,7 +54,7 @@ int main(void) { std::vector<Point_d> points = {0, 1, 2, 3}; std::vector<Point_d> results; - Gudhi::subsampling::choose_n_farthest_points(k, points, 2, std::back_inserter(results)); + Gudhi::subsampling::choose_n_farthest_points(k, points, 2, Gudhi::subsampling::random_starting_point, std::back_inserter(results)); std::cout << "Before sparsification: " << points.size() << " points.\n"; std::cout << "After sparsification: " << results.size() << " points.\n"; std::cout << "Result table: {" << results[0] << "," << results[1] << "}\n"; diff --git a/src/Subsampling/include/gudhi/choose_n_farthest_points.h b/src/Subsampling/include/gudhi/choose_n_farthest_points.h index 5e908090..50d3cf80 100644 --- a/src/Subsampling/include/gudhi/choose_n_farthest_points.h +++ b/src/Subsampling/include/gudhi/choose_n_farthest_points.h @@ -25,16 +25,9 @@ #include <boost/range.hpp> -#include <gudhi/Kd_tree_search.h> - -#include <gudhi/Clock.h> - -#include <CGAL/Search_traits.h> -#include <CGAL/Search_traits_adapter.h> -#include <CGAL/Fuzzy_sphere.h> +#include <gudhi/Null_output_iterator.h> #include <iterator> -#include <algorithm> // for sort #include <vector> #include <random> #include <limits> // for numeric_limits<> @@ -43,36 +36,51 @@ namespace Gudhi { namespace subsampling { +/** + * \ingroup subsampling + */ +enum : std::size_t { +/** + * Argument for `choose_n_farthest_points` to indicate that the starting point should be picked randomly. + */ + random_starting_point = std::size_t(-1) +}; + /** * \ingroup subsampling * \brief Subsample by a greedy strategy of iteratively adding the farthest point from the * current chosen point set to the subsampling. - * The iteration starts with the landmark `starting point`. + * The iteration starts with the landmark `starting point` or, if `starting point==random_starting_point`, with a random landmark. * \tparam Kernel must provide a type Kernel::Squared_distance_d which is a model of the * concept <a target="_blank" - * href="http://doc.cgal.org/latest/Kernel_d/classKernel__d_1_1Squared__distance__d.html">Kernel_d::Squared_distance_d</a> - * concept. - * It must also contain a public member 'squared_distance_d_object' of this type. + * href="http://doc.cgal.org/latest/Kernel_d/classKernel__d_1_1Squared__distance__d.html">Kernel_d::Squared_distance_d</a> (despite the name, taken from CGAL, this can be any kind of metric or proximity measure). + * It must also contain a public member `squared_distance_d_object()` that returns an object of this type. * \tparam Point_range Range whose value type is Kernel::Point_d. It must provide random-access * via `operator[]` and the points should be stored contiguously in memory. - * \tparam OutputIterator Output iterator whose value type is Kernel::Point_d. - * \details It chooses `final_size` points from a random access range `input_pts` and - * outputs it in the output iterator `output_it`. + * \tparam PointOutputIterator Output iterator whose value type is Kernel::Point_d. + * \tparam DistanceOutputIterator Output iterator for distances. + * \details It chooses `final_size` points from a random access range + * `input_pts` and outputs them in the output iterator `output_it`. It also + * outputs the distance from each of those points to the set of previous + * points in `dist_it`. * @param[in] k A kernel object. * @param[in] input_pts Const reference to the input points. * @param[in] final_size The size of the subsample to compute. * @param[in] starting_point The seed in the farthest point algorithm. - * @param[out] output_it The output iterator. + * @param[out] output_it The output iterator for points. + * @param[out] dist_it The optional output iterator for distances. * */ template < typename Kernel, typename Point_range, -typename OutputIterator> +typename PointOutputIterator, +typename DistanceOutputIterator=Null_output_iterator> void choose_n_farthest_points(Kernel const &k, Point_range const &input_pts, std::size_t final_size, std::size_t starting_point, - OutputIterator output_it) { + PointOutputIterator output_it, + DistanceOutputIterator dist_it={}) { std::size_t nb_points = boost::size(input_pts); if (final_size > nb_points) final_size = nb_points; @@ -81,6 +89,14 @@ void choose_n_farthest_points(Kernel const &k, if (final_size < 1) return; + if (starting_point == random_starting_point) { + // Choose randomly the first landmark + std::random_device rd; + std::mt19937 gen(rd()); + std::uniform_int_distribution<std::size_t> dis(0, (input_pts.size() - 1)); + starting_point = dis(gen); + } + typename Kernel::Squared_distance_d sqdist = k.squared_distance_d_object(); std::size_t current_number_of_landmarks = 0; // counter for landmarks @@ -92,6 +108,7 @@ void choose_n_farthest_points(Kernel const &k, for (current_number_of_landmarks = 0; current_number_of_landmarks != final_size; current_number_of_landmarks++) { // curr_max_w at this point is the next landmark *output_it++ = input_pts[curr_max_w]; + *dist_it++ = dist_to_L[curr_max_w]; std::size_t i = 0; for (auto& p : input_pts) { double curr_dist = sqdist(p, *(std::begin(input_pts) + curr_max_w)); @@ -109,47 +126,6 @@ void choose_n_farthest_points(Kernel const &k, } } -/** - * \ingroup subsampling - * \brief Subsample by a greedy strategy of iteratively adding the farthest point from the - * current chosen point set to the subsampling. - * The iteration starts with a random landmark. - * \tparam Kernel must provide a type Kernel::Squared_distance_d which is a model of the - * concept <a target="_blank" - * href="http://doc.cgal.org/latest/Kernel_d/classKernel__d_1_1Squared__distance__d.html">Kernel_d::Squared_distance_d</a> - * concept. - * It must also contain a public member 'squared_distance_d_object' of this type. - * \tparam Point_range Range whose value type is Kernel::Point_d. It must provide random-access - * via `operator[]` and the points should be stored contiguously in memory. - * \tparam OutputIterator Output iterator whose value type is Kernel::Point_d. - * \details It chooses `final_size` points from a random access range `input_pts` and - * outputs it in the output iterator `output_it`. - * @param[in] k A kernel object. - * @param[in] input_pts Const reference to the input points. - * @param[in] final_size The size of the subsample to compute. - * @param[out] output_it The output iterator. - * - */ -template < typename Kernel, -typename Point_range, -typename OutputIterator> -void choose_n_farthest_points(Kernel const& k, - Point_range const &input_pts, - unsigned final_size, - OutputIterator output_it) { - // Tests to the limit - if ((final_size < 1) || (input_pts.size() == 0)) - return; - - // Choose randomly the first landmark - std::random_device rd; - std::mt19937 gen(rd()); - std::uniform_int_distribution<> dis(0, (input_pts.size() - 1)); - std::size_t starting_point = dis(gen); - - choose_n_farthest_points(k, input_pts, final_size, starting_point, output_it); -} - } // namespace subsampling } // namespace Gudhi diff --git a/src/Subsampling/test/test_choose_n_farthest_points.cpp b/src/Subsampling/test/test_choose_n_farthest_points.cpp index 0bc0dff4..ee9d4c77 100644 --- a/src/Subsampling/test/test_choose_n_farthest_points.cpp +++ b/src/Subsampling/test/test_choose_n_farthest_points.cpp @@ -25,7 +25,7 @@ // #endif #define BOOST_TEST_DYN_LINK -#define BOOST_TEST_MODULE "witness_complex_points" +#define BOOST_TEST_MODULE Subsampling - test choose_n_farthest_points #include <boost/test/unit_test.hpp> #include <boost/mpl/list.hpp> @@ -56,7 +56,7 @@ BOOST_AUTO_TEST_CASE_TEMPLATE(test_choose_farthest_point, Kernel, list_of_tested landmarks.clear(); Kernel k; - Gudhi::subsampling::choose_n_farthest_points(k, points, 100, std::back_inserter(landmarks)); + Gudhi::subsampling::choose_n_farthest_points(k, points, 100, Gudhi::subsampling::random_starting_point, std::back_inserter(landmarks)); BOOST_CHECK(landmarks.size() == 100); for (auto landmark : landmarks) @@ -70,34 +70,45 @@ BOOST_AUTO_TEST_CASE_TEMPLATE(test_choose_farthest_point_limits, Kernel, list_of typedef typename Kernel::FT FT; typedef typename Kernel::Point_d Point_d; std::vector< Point_d > points, landmarks; + std::vector< FT > distances; landmarks.clear(); Kernel k; // Choose -1 farthest points in an empty point cloud - Gudhi::subsampling::choose_n_farthest_points(k, points, -1, std::back_inserter(landmarks)); + Gudhi::subsampling::choose_n_farthest_points(k, points, -1, -1, std::back_inserter(landmarks), std::back_inserter(distances)); BOOST_CHECK(landmarks.size() == 0); - landmarks.clear(); + landmarks.clear(); distances.clear(); // Choose 0 farthest points in an empty point cloud - Gudhi::subsampling::choose_n_farthest_points(k, points, 0, std::back_inserter(landmarks)); + Gudhi::subsampling::choose_n_farthest_points(k, points, 0, -1, std::back_inserter(landmarks), std::back_inserter(distances)); BOOST_CHECK(landmarks.size() == 0); - landmarks.clear(); + landmarks.clear(); distances.clear(); // Choose 1 farthest points in an empty point cloud - Gudhi::subsampling::choose_n_farthest_points(k, points, 1, std::back_inserter(landmarks)); + Gudhi::subsampling::choose_n_farthest_points(k, points, 1, -1, std::back_inserter(landmarks), std::back_inserter(distances)); BOOST_CHECK(landmarks.size() == 0); - landmarks.clear(); + landmarks.clear(); distances.clear(); std::vector<FT> point({0.0, 0.0, 0.0, 0.0}); points.push_back(Point_d(point.begin(), point.end())); - // Choose -1 farthest points in an empty point cloud - Gudhi::subsampling::choose_n_farthest_points(k, points, -1, std::back_inserter(landmarks)); - BOOST_CHECK(landmarks.size() == 1); - landmarks.clear(); + // Choose -1 farthest points in a one point cloud + Gudhi::subsampling::choose_n_farthest_points(k, points, -1, -1, std::back_inserter(landmarks), std::back_inserter(distances)); + BOOST_CHECK(landmarks.size() == 1 && distances.size() == 1); + BOOST_CHECK(distances[0] == std::numeric_limits<FT>::infinity()); + landmarks.clear(); distances.clear(); // Choose 0 farthest points in a one point cloud - Gudhi::subsampling::choose_n_farthest_points(k, points, 0, std::back_inserter(landmarks)); - BOOST_CHECK(landmarks.size() == 0); - landmarks.clear(); + Gudhi::subsampling::choose_n_farthest_points(k, points, 0, -1, std::back_inserter(landmarks), std::back_inserter(distances)); + BOOST_CHECK(landmarks.size() == 0 && distances.size() == 0); + landmarks.clear(); distances.clear(); // Choose 1 farthest points in a one point cloud - Gudhi::subsampling::choose_n_farthest_points(k, points, 1, std::back_inserter(landmarks)); - BOOST_CHECK(landmarks.size() == 1); - landmarks.clear(); + Gudhi::subsampling::choose_n_farthest_points(k, points, 1, -1, std::back_inserter(landmarks), std::back_inserter(distances)); + BOOST_CHECK(landmarks.size() == 1 && distances.size() == 1); + BOOST_CHECK(distances[0] == std::numeric_limits<FT>::infinity()); + landmarks.clear(); distances.clear(); + std::vector<FT> point2({1.0, 0.0, 0.0, 0.0}); + points.push_back(Point_d(point2.begin(), point2.end())); + // Choose all farthest points in a one point cloud + Gudhi::subsampling::choose_n_farthest_points(k, points, -1, -1, std::back_inserter(landmarks), std::back_inserter(distances)); + BOOST_CHECK(landmarks.size() == 2 && distances.size() == 2); + BOOST_CHECK(distances[0] == std::numeric_limits<FT>::infinity()); + BOOST_CHECK(distances[1] == 1); + landmarks.clear(); distances.clear(); } diff --git a/src/common/include/gudhi/Null_output_iterator.h b/src/common/include/gudhi/Null_output_iterator.h new file mode 100644 index 00000000..0cf043cd --- /dev/null +++ b/src/common/include/gudhi/Null_output_iterator.h @@ -0,0 +1,48 @@ +/* This file is part of the Gudhi Library. The Gudhi library + * (Geometric Understanding in Higher Dimensions) is a generic C++ + * library for computational topology. + * + * Author(s): Marc Glisse + * + * Copyright (C) 2017 INRIA + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see <http://www.gnu.org/licenses/>. + */ + +#ifndef NULL_OUTPUT_ITERATOR_H_ +#define NULL_OUTPUT_ITERATOR_H_ + +#include <iterator> + +namespace Gudhi { + +/** An output iterator that ignores whatever it is given. */ +struct Null_output_iterator { + typedef std::output_iterator_tag iterator_category; + typedef void value_type; + typedef void difference_type; + typedef void pointer; + typedef void reference; + + Null_output_iterator& operator++(){return *this;} + Null_output_iterator operator++(int){return *this;} + struct proxy { + template<class T> + proxy& operator=(T&&){return *this;} + }; + proxy operator*()const{return {};} +}; +} // namespace Gudhi + +#endif // NULL_OUTPUT_ITERATOR_H_ diff --git a/src/cython/include/Subsampling_interface.h b/src/cython/include/Subsampling_interface.h index 1c6032c0..b0f4a50a 100644 --- a/src/cython/include/Subsampling_interface.h +++ b/src/cython/include/Subsampling_interface.h @@ -46,7 +46,7 @@ std::vector<std::vector<double>> subsampling_n_farthest_points(const std::vector unsigned nb_points) { std::vector<std::vector<double>> landmarks; Subsampling_dynamic_kernel k; - choose_n_farthest_points(k, points, nb_points, std::back_inserter(landmarks)); + choose_n_farthest_points(k, points, nb_points, random_starting_point, std::back_inserter(landmarks)); return landmarks; } |