c++ - 卡在使用蒙特卡洛积分法?
问题描述
我想使用蒙特卡罗集成方法,我的代码如下。如您所见,我确定了区间积分,但结果错误!这段代码有什么问题?
任何帮助将不胜感激 。
#include <iostream>
#include <math.h>
#include <stdlib.h>
#define N 500
using namespace std;
double Func(double x) { return pow(x, 2) + 1; }
double Monte_Carlo(double Func(double), double xmin, double xmax, double ymin,
double ymax)
{
int acc = 0;
int tot = 0;
for (int count = 0; count < N; count++)
{
double x0 = (double)rand() / 4 + (-2);
double y0 = (double)rand() / 4 + 0;
float x = x0 / (float)RAND_MAX;
float y = y0 / (float)RAND_MAX;
cout << x << endl;
if (y <= Func(x))
acc++;
tot++;
// cout << "Dgage" << tot << '\t' << acc << endl;
}
double Coeff = acc / N;
return (xmax - xmin) * (1.2 * Func(xmax)) * Coeff;
}
int main()
{
cout << "Integral value is: " << Monte_Carlo(Func, -2, 2, 0, 4) << endl;
system("pause");
return 0;
}
解决方案
Monte_Carlo 函数使事情变得比他们需要的更复杂。为了整合一维函数,我们所要做的就是在我们正在整合的区域内多次采样函数的值:
#include <random>
double Monte_Carlo(double Func(double), double xmin, double xmax, int N)
{
// This is the distribution we're using to generate inputs
auto x_dist = std::uniform_real_distribution<>(xmin, xmax);
// This is the random number generator itself
auto rng = std::default_random_engine();
// Calculate the total of N random samples
double total = 0.0;
for(int i = 0; i < N; i++) {
double x = x_dist(rng); // Generate a value
total += Func(x);
}
// Return the size of the interval times the total,
// divided by the number of samples
return (xmax - xmin) * total / N;
}
如果我们用 运行这段代码N = 1000
,我们会得到 的整数值9.20569
,它非常接近准确的答案 ( 9.33333...
)。
// It's much more efficent to use x*x instead of pow
double func(double x) { return x * x + 1; }
int main()
{
cout << "Integral value is: " << Monte_Carlo(func, -2, 2, 1000) << endl;
getchar(); // Pause until the user presses enter
return 0;
}
我们也可以尝试多个 , 的值N
,让程序显示它是如何收敛的。下面的程序计算积分,其中 N 是2
从0
到的幂30
#include <iostream>
#include <cmath>
#include <random>
using namespace std;
double func(double x) { return x*x + 1; }
double Monte_Carlo(double Func(double), double xmin, double xmax, int N) {
auto x_dist = std::uniform_real_distribution<>(xmin, xmax);
auto rng = std::default_random_engine();
double total = 0.0;
for(int i = 0; i < N; i++) {
double x = x_dist(rng); // Generate a value
total += Func(x);
}
return (xmax - xmin) * total / N;
}
int main() {
int N = 1;
for(int i = 0; i < 31; i++) {
std::cout << "N = " << N << "\t\tintegral = " << Monte_Carlo(func, -2, 2, N) << endl;
N *= 2; // Double N
}
}
输出显示蒙特卡罗方法确实收敛:
N = 1 integral = 12.6889
N = 2 integral = 8.39917
N = 4 integral = 7.97521
N = 8 integral = 9.24233
N = 16 integral = 9.75632
N = 32 integral = 9.87064
N = 64 integral = 9.46945
N = 128 integral = 9.27281
N = 256 integral = 9.27395
N = 512 integral = 9.17546
N = 1024 integral = 9.19097
N = 2048 integral = 9.26203
N = 4096 integral = 9.37979
N = 8192 integral = 9.36167
N = 16384 integral = 9.28918
N = 32768 integral = 9.29766
N = 65536 integral = 9.31101
N = 131072 integral = 9.3227
N = 262144 integral = 9.32588
N = 524288 integral = 9.32805
N = 1048576 integral = 9.32726
N = 2097152 integral = 9.32722
N = 4194304 integral = 9.331
N = 8388608 integral = 9.33082
N = 16777216 integral = 9.33174
N = 33554432 integral = 9.33164
N = 67108864 integral = 9.33303
N = 134217728 integral = 9.33283
N = 268435456 integral = 9.33327
N = 536870912 integral = 9.33325
N = 1073741824 integral = 9.33333
推荐阅读
- php - PHP in_array() 总是返回 false
- android - 项目包含多个android模块时不生成dagger测试组件
- android - Android Firebase 登录无效的 idToken
- java - 使用 Postman 通过 Socket 传递 JSON 时如何修复错误
- apache - 在此服务器上找不到请求的 URL /install/index
- angular - Angular 7 - 从数据库中的字节显示图像
- python - 检查字符串中连续单词的首字母是否匹配另一个字符串的首字母缩写词
- swift - 为什么单个展开 segue 具有不同的行为?
- python - 如何将 Micropython 上的电机导入 DFRobot Quad Motor Shield
- python - 如果每个 ec2 实例不存在,请在其上创建一个标签。如果存在,则使用新值更新标签