比较两个双精度浮点数或两个浮点数最有效的方法是什么?

简单地这样做是不正确的:

bool CompareDoubles1 (double A, double B)
{
   return A == B;
}

比如:

bool CompareDoubles2 (double A, double B) 
{
   diff = A - B;
   return (diff < EPSILON) && (-diff < EPSILON);
}

似乎是浪费加工。

有人知道更聪明的浮点比较器吗?


当前回答

Why not perform bitwise XOR? Two floating point numbers are equal if their corresponding bits are equal. I think, the decision to place the exponent bits before mantissa was made to speed up comparison of two floats. I think, many answers here are missing the point of epsilon comparison. Epsilon value only depends on to what precision floating point numbers are compared. For example, after doing some arithmetic with floats you get two numbers: 2.5642943554342 and 2.5642943554345. They are not equal, but for the solution only 3 decimal digits matter so then they are equal: 2.564 and 2.564. In this case you choose epsilon equal to 0.001. Epsilon comparison is also possible with bitwise XOR. Correct me if I am wrong.

其他回答

/// testing whether two doubles are almost equal. We consider two doubles
/// equal if the difference is within the range [0, epsilon).
///
/// epsilon: a positive number (supposed to be small)
///
/// if either x or y is 0, then we are comparing the absolute difference to
/// epsilon.
/// if both x and y are non-zero, then we are comparing the relative difference
/// to epsilon.
bool almost_equal(double x, double y, double epsilon)
{
    double diff = x - y;
    if (x != 0 && y != 0){
        diff = diff/y; 
    }

    if (diff < epsilon && -1.0*diff < epsilon){
        return true;
    }
    return false;
}

我在我的小项目中使用了这个函数,它是有效的,但注意以下几点:

双精度误差可以为你制造惊喜。假设epsilon = 1.0e-6,那么根据上面的代码,1.0和1.000001不应该被认为是相等的,但在我的机器上,函数认为它们是相等的,这是因为1.000001不能精确地转换为二进制格式,它可能是1.0000009xxx。我用1.0和1.0000011测试了它,这次我得到了预期的结果。

就数量的规模而言:

如果在某种物理意义上,ε是量的大小(即相对值)的一小部分,而A和B类型在同一意义上具有可比性,那么我认为,下面的观点是相当正确的:

#include <limits>
#include <iomanip>
#include <iostream>

#include <cmath>
#include <cstdlib>
#include <cassert>

template< typename A, typename B >
inline
bool close_enough(A const & a, B const & b,
                  typename std::common_type< A, B >::type const & epsilon)
{
    using std::isless;
    assert(isless(0, epsilon)); // epsilon is a part of the whole quantity
    assert(isless(epsilon, 1));
    using std::abs;
    auto const delta = abs(a - b);
    auto const x = abs(a);
    auto const y = abs(b);
    // comparable generally and |a - b| < eps * (|a| + |b|) / 2
    return isless(epsilon * y, x) && isless(epsilon * x, y) && isless((delta + delta) / (x + y), epsilon);
}

int main()
{
    std::cout << std::boolalpha << close_enough(0.9, 1.0, 0.1) << std::endl;
    std::cout << std::boolalpha << close_enough(1.0, 1.1, 0.1) << std::endl;
    std::cout << std::boolalpha << close_enough(1.1,    1.2,    0.01) << std::endl;
    std::cout << std::boolalpha << close_enough(1.0001, 1.0002, 0.01) << std::endl;
    std::cout << std::boolalpha << close_enough(1.0, 0.01, 0.1) << std::endl;
    return EXIT_SUCCESS;
}

这个怎么样?

template<typename T>
bool FloatingPointEqual( T a, T b ) { return !(a < b) && !(b < a); }

我见过各种方法,但从来没有见过这个,所以我也很好奇听到任何评论!

比较浮点数取决于上下文。因为即使改变操作的顺序也会产生不同的结果,所以知道你希望这些数字有多“相等”是很重要的。

在研究浮点数比较时,比较Bruce Dawson编写的浮点数是一个很好的开始。

以下定义来自Knuth的《The art of computer programming》:

bool approximatelyEqual(float a, float b, float epsilon)
{
    return fabs(a - b) <= ( (fabs(a) < fabs(b) ? fabs(b) : fabs(a)) * epsilon);
}

bool essentiallyEqual(float a, float b, float epsilon)
{
    return fabs(a - b) <= ( (fabs(a) > fabs(b) ? fabs(b) : fabs(a)) * epsilon);
}

bool definitelyGreaterThan(float a, float b, float epsilon)
{
    return (a - b) > ( (fabs(a) < fabs(b) ? fabs(b) : fabs(a)) * epsilon);
}

bool definitelyLessThan(float a, float b, float epsilon)
{
    return (b - a) > ( (fabs(a) < fabs(b) ? fabs(b) : fabs(a)) * epsilon);
}

当然,选择取决于上下文,并决定你想要的数字有多相等。

比较浮点数的另一种方法是查看数字的ULP(最后位置的单位)。虽然没有专门处理比较,但“每个计算机科学家都应该知道浮点数”这篇论文是了解浮点数如何工作以及陷阱是什么,包括什么是ULP的很好的资源。

以更一般的方式:

template <typename T>
bool compareNumber(const T& a, const T& b) {
    return std::abs(a - b) < std::numeric_limits<T>::epsilon();
}

注意: 正如@SirGuy所指出的,这种方法是有缺陷的。 我把这个答案留在这里,作为一个不遵循的例子。