使用Boost查询点

您的样本在统计上不可靠。

让我们大大增加查询框：

double const delta = .05; // so boxes are .1x.1, i.e. 1/100th of total area
// and later
for (auto& p : querylist) {
    box query_box(point {p.x - delta, p.y - delta}, point {p.x + delta, p.y + delta});

结果正好达到预期的点击次数：

在Coliru上直播

Time for building R tree 2182054657 nanoseconds
Time for query 52749607 nanoseconds
Average intersections: 10000.3
Standard deviation:    98.9173
Expected mean:         10000

代码

票据

• 用Boost Random（uniform_real<>分布）代替rand()
• 修正了框不与区域边界重叠的问题（这样我们就不会得到扭曲的结果）

• 不需要每次都填充一个向量，只需测量查询结果的数量：

  auto n = std::distance(
          bgi::qbegin(rtree, bgi::intersects(query_box)),
          bgi::qend(rtree));
  

• 使用Boost Accumulator:进行一些适当的统计，而不是"轶事式"报告一些"交叉点"计数

  std::cout << "Average intersections: " << ba::mean(stats) << "n";
  std::cout << "Standard deviation:    " << sqrt(ba::variance(stats)) << "n";
  std::cout << "Expected mean:         " << (N * (2*delta) * (2*delta)) << "n";
  

完整列表

#include <iostream>
#include <vector>
#include <time.h>
#include <math.h>
#include <stdlib.h>
#include <iomanip>
#include <boost/geometry/geometry.hpp>
#include <boost/geometry/geometries/geometries.hpp>
#include <boost/geometry/geometries/point.hpp>
#include <boost/geometry/index/rtree.hpp>
#include <boost/random.hpp>
#include <boost/chrono.hpp>
#include <boost/accumulators/accumulators.hpp>
#include <boost/accumulators/statistics.hpp>
namespace bg  = boost::geometry;
namespace bgi = boost::geometry::index;
double const delta = .05; // so boxes are .1x.1, i.e. 1/100th of total area
typedef bg::model::point<double, 2, bg::cs::cartesian> point;
typedef bg::model::box<point> box;
struct particle {
    double x;
    double y;
};
namespace ba  = boost::accumulators;
namespace bat = ba::tag;
using hrc = boost::chrono::high_resolution_clock;
int main() {
    boost::mt19937 prng(std::random_device{}());
    size_t N = 1000000ul;
    std::vector<particle> myvec;
    myvec.reserve(N);
    {
        boost::uniform_real<double> dist(0,1);
        // Set vector values
        for (size_t i = 0; i<N; ++i)
            myvec.push_back({dist(prng), dist(prng)});
    }
    // Build R-tree
    auto start = hrc::now();
    bgi::rtree<point, bgi::quadratic<16> > rtree;
    for (auto& p : myvec)
        rtree.insert(point {p.x, p.y});
    auto stop = hrc::now();
    std::cout << "Time for building R tree " << (stop - start) << std::endl;
    // Build Query List
    std::vector<particle> querylist;
    querylist.reserve(100);
    {
        boost::uniform_real<double> dist(0+delta,1-delta);
        for (size_t i = 0; i<100; ++i)
            querylist.push_back({dist(prng), dist(prng)});
    }
    ba::accumulator_set<double, ba::stats<bat::mean, bat::variance>> stats;
    // Query R-tree
    start = hrc::now();
    for (auto& p : querylist) {
        box query_box(point {p.x - delta, p.y - delta}, point {p.x + delta, p.y + delta});
#if 0
        std::vector<point> queryresult;
        rtree.query(bgi::intersects(query_box), std::back_inserter(queryresult));
        //std::cout << "The intersection has " << queryresult.size() << " elements:n";
        stats(queryresult.size());
#else
        auto n = std::distance(
                bgi::qbegin(rtree, bgi::intersects(query_box)),
                bgi::qend(rtree));
        stats(n);
#endif
    }
    stop = hrc::now();
    std::cout << "Time for query " << (stop - start) << "n";
    std::cout << "Average intersections: " << ba::mean(stats) << "n";
    std::cout << "Standard deviation:    " << sqrt(ba::variance(stats)) << "n";
    std::cout << "Expected mean:         " << (N * (2*delta) * (2*delta)) << "n";
}

缩放

注意，结果等级：

double const delta = .005; // so boxes are .01x.01, i.e. 1/10000th of total area
Time for building R tree 2101529662 nanoseconds
Time for query 2982070 nanoseconds
Average intersections: 99.69
Standard deviation:    8.96069
Expected mean:         100

甚至

double const delta = .00005; // so boxes are .0001x.0001, i.e. 1/100000000th of total area
Time for building R tree 726361394 nanoseconds
Time for query 216690 nanoseconds
Average intersections: 0.02
Standard deviation:    0.14
Expected mean:         0.01