我如何使用 Boost Random
我需要使用 Boost Random 生成随机数.
I need to generate random number with Boost Random.
我尝试遵循一般指南.
我提取了图书馆的文件.那么如果我想使用库的类和对象j我应该怎么做?
I extracted the files of the library. So if I want to use the classes and objectj of the library how I should do?
首先我知道在程序中包含库.然后我必须编译库和program.cpp本身?(并且都使用相同的编译器 - 我使用的是 g++).
First I know including the library in the program. Then I have to compile the library and the program.cpp itself? (And both with the same compiler - I'm using g++).
我正在使用 ubuntu 的虚拟盒子.我是第一次使用图书馆，所以我真的不知道.
I am using a virtual box of ubuntu. It is first time that I am using library so I really don't know.
推荐答案
我的情况下的随机数必须是 double 而不仅仅是整数...
the random number for my case must be double not just integer...
因此，您使用的是实数分布.
So, you use a real number distribution.
我不是这种入门"最适合 StackOverflow，但我会给你这个快速提示:
I'm not this kind of "getting started" is the best fit for StackOverflow, but I'll give you this quick hints:
在您的 Ubuntu 虚拟机中:
In your Ubuntu virtual box:
sudo apt-get install libboost-all-dev
mkdir -pv ~/myproject
cd ~/myproject
使用您喜欢的编辑器创建一个文件.如果你没有，gedit main.cpp 或 nano main.cpp 是一个开始:
Create a file using your favourite editor. If you have none, gedit main.cpp or nano main.cpp is a start:
#include <boost/random.hpp>
#include <iostream>
int main() {
boost::random::mt19937 rng;
boost::random::uniform_real_distribution<double> gen(0.0, 1.0);
for (int i = 0; i < 10; ++i) {
std::cout << gen(rng) << "
";
}
}
现在使用
g++ -O2 -Wall -Wextra -pedantic main.cpp -o demo
程序现在可以运行了:Live On Coliru
./demo
打印
0.814724
0.135477
0.905792
0.835009
0.126987
0.968868
0.913376
0.221034
0.632359
0.308167
播种&&非头文件库
上述工作是因为 Boost Random 库主要是仅标头.如果您想使用 random_device 实现来为随机生成器提供种子怎么办?
Seeding && Non-Header Only Libraries
The above works because the Boost Random library is mostly header only. What if you wanted to use the random_device implementation to seed the random generator?
生活在 Coliru
#include <boost/random.hpp>
#include <boost/random/random_device.hpp>
#include <iostream>
int main() {
boost::random::random_device seeder;
boost::random::mt19937 rng(seeder());
boost::random::uniform_real_distribution<double> gen(0.0, 1.0);
for (int i = 0; i < 10; ++i) {
std::cout << gen(rng) << "
";
}
}
现在您还必须链接:使用编译
Now you'll have to link as well: Compiling with
g++ -O2 -Wall -Wextra -pedantic main.cpp -o demo -lboost_random
现在每次运行的输出都会不同.
Now the output will be different each run.
这里根本不需要 Boost:
You don't need Boost here at all:
生活在 Coliru
#include <random>
#include <iostream>
int main() {
std::random_device seeder;
std::mt19937 rng(seeder());
std::uniform_real_distribution<double> gen(0.0, 1.0);
for (int i = 0; i < 10; ++i) {
std::cout << gen(rng) << "
";
}
}
编译
g++ -std=c++11 -O2 -Wall -Wextra -pedantic main.cpp -o demo
然后用 ./demo
显示均值=0 和标准偏差=1 的整个分布范围:
Showing a whole gamut of distributions that have mean=0 and stddev=1:
生活在 Coliru
#include <random>
#include <iostream>
#include <iomanip>
#include <chrono>
#include <boost/serialization/array_wrapper.hpp>
#include <boost/accumulators/accumulators.hpp>
#include <boost/accumulators/statistics.hpp>
namespace ba = boost::accumulators;
using Accum = ba::accumulator_set<double, ba::stats<ba::tag::variance, ba::tag::mean> >;
using Clock = std::chrono::high_resolution_clock;
using namespace std::chrono_literals;
static double identity(double d) { return d; }
template <typename Prng, typename Dist, typename F = double(double), size_t N = (1ull << 22)>
void test(Prng& rng, Dist dist, F f = &identity) {
Accum accum;
auto s = Clock::now();
for (size_t i = 0; i<N; ++i)
accum(f(dist(rng)));
std::cout
<< std::setw(34) << typeid(Dist).name()
<< ": " << ba::mean(accum)
<< " stddev: " << sqrt(ba::variance(accum))
<< " N=" << N
<< " in " << ((Clock::now()-s)/1.s) << "s"
<< std::endl;
}
int main() {
std::mt19937 rng(std::random_device{}());
auto shift = [](double shift) { return [=](double v) { return v + shift; }; };
auto scale = [](double scale) { return [=](double v) { return v * scale; }; };
std::cout << std::fixed << std::showpos;
test(rng, std::uniform_real_distribution<double>(-sqrt(3), sqrt(3)));
test(rng, std::weibull_distribution<double>(), shift(-1));
test(rng, std::exponential_distribution<double>(), shift(-1));
test(rng, std::normal_distribution<double>());
test(rng, std::lognormal_distribution<double>(0, log(0.5)), shift(-exp(pow(log(0.5),2)/2)));
test(rng, std::chi_squared_distribution<double>(0.5), shift(-0.5));
{
auto sigma = sqrt(6)/M_PI;
static constexpr double ec = 0.57721566490153286060;
test(rng, std::extreme_value_distribution<double>(-sigma*ec, sigma));
}
test(rng, std::fisher_f_distribution<double>(48, 8), shift(-(8.0/6.0)));
test(rng, std::student_t_distribution<double>(4), scale(sqrt(0.5)));
test(rng, std::student_t_distribution<double>(4), scale(sqrt(0.5)));
}
印刷品
St25uniform_real_distributionIdE: +0.000375 stddev: +1.000056 N=4194304 in +0.169681s
St20weibull_distributionIdE: +0.001030 stddev: +1.000518 N=4194304 in +0.385036s
St24exponential_distributionIdE: -0.000360 stddev: +1.000343 N=4194304 in +0.389443s
St19normal_distributionIdE: -0.000133 stddev: +1.000330 N=4194304 in +0.390235s
St22lognormal_distributionIdE: +0.000887 stddev: +1.000372 N=4194304 in +0.521975s
St24chi_squared_distributionIdE: -0.000092 stddev: +0.999695 N=4194304 in +1.233835s
St26extreme_value_distributionIdE: -0.000381 stddev: +1.000242 N=4194304 in +0.611973s
St21fisher_f_distributionIdE: -0.000073 stddev: +1.001588 N=4194304 in +1.326189s
St22student_t_distributionIdE: +0.000957 stddev: +0.998087 N=4194304 in +1.080468s
St22student_t_distributionIdE: +0.000677 stddev: +0.998786 N=4194304 in +1.079066s
相关文章