前一天调试一个算法,被一个小bug折腾了半天,贴出来游行示众:
#include <iostream>
#include <cmath>
int main()
{
const double pi = 3.141592653589793238; // double 有效位数为 16
double d1 = 2.1212;
double d2 = -5.1515;
double d3 = 9.1919;
int k1 = d1/(2*pi);
int k2 = d2/(2*pi);
int k3 = d3/(2*pi);
double t1 = std::fmod(d3, pi); // (-pi, pi)
double t2 = std::remainder(d3, 2*pi); // (-pi,pi)
// 千万别在算法中将系数写成整数, 改成 1.0/2*d3 或者 1*d3/2 就ok了
double t3 = 1/2*d3; // 根据先后顺序得出 t3=0 !!!
std::cerr << "k_1: " << k1 << std::endl;
std::cerr << "k_2: " << k2 << std::endl;
std::cerr << "k_3: " << k3 << std::endl;
std::cerr << "t_1: " << t1 << std::endl;
std::cerr << "t_2: " << t2 << std::endl;
std::cerr << "t_3: " << t3 << std::endl;
}
Output:
k_1: 0
k_2: 0
k_3: 1
t_1: 2.90871
t_2: 2.90871
t_3: 0