std::numeric_limits<T>::epsilon

 

C++

编译器支持

独立和托管

语言

标准库

标准库头文件

命名要求

功能测试宏 (C++20)

语言支持库

概念库 (C++20)

元编程库 (C++11)

诊断库

通用实用工具库

字符串库

容器库

迭代器库

范围库 (C++20)

算法库

数值库

本地化库

输入/输出库

文件系统库 (C++17)

正则表达式库 (C++11)

并发支持库 (C++11)

执行支持库 (C++26)

技术规范

符号索引

外部库

 

实用工具库

语言支持 类型支持 (基本类型，RTTI) 库功能测试宏 (C++20) 动态内存管理 程序实用工具 协程支持 (C++20) 可变参数函数 is_constant_evaluated (C++20) is_within_lifetime (C++26) initializer_list (C++11) source_location (C++20) 调试支持 is_debugger_present (C++26) breakpoint_if_debugging (C++26) breakpoint (C++26) 三向比较 three_way_comparablethree_way_comparable_with (C++20)(C++20) strong_ordering (C++20) weak_ordering (C++20) partial_ordering (C++20) common_comparison_category (C++20) compare_three_way_result (C++20) compare_three_way (C++20) strong_order (C++20) weak_order (C++20) partial_order (C++20) compare_strong_order_fallback (C++20) compare_weak_order_fallback (C++20) compare_partial_order_fallback (C++20)    is_eqis_ltis_lteq (C++20)(C++20)(C++20) is_neqis_gtis_gteq (C++20)(C++20)(C++20)

通用实用工具 日期和时间 函数对象 格式化库 (C++20) bitset hash (C++11) 关系运算符 (C++20 中已弃用) rel_ops::operator!=rel_ops::operator>     rel_ops::operator<=rel_ops::operator>= 整数比较函数 cmp_equalcmp_lesscmp_less_than (C++20)(C++20)(C++20)    cmp_not_equalcmp_greatercmp_greater_than (C++20)(C++20)(C++20) in_range (C++20) 交换 和 类型操作 swap ranges::swap (C++20) exchange (C++14) declval (C++11) to_underlying (C++23) forward (C++11) forward_like (C++23) move (C++11) move_if_noexcept (C++11) as_const (C++17) 通用词汇类型 pair tuple (C++11) optional (C++17) any (C++17) variant (C++17) tuple_size (C++11) tuple_element (C++11) apply (C++17) make_from_tuple (C++17) expected (C++23) 基本字符串转换 to_chars (C++17) from_chars (C++17) chars_format (C++17) to_chars_result (C++17) from_chars_result (C++17)

 

类型支持

基本类型
固定宽度整数类型 (C++11)
固定宽度浮点数类型 (C++23)
ptrdiff_t size_t max_align_t (C++11)     byte (C++17)	nullptr_t (C++11) offsetof NULL
数值限制
numeric_limits
C 数值限制接口
运行时类型信息
type_info type_index (C++11)	bad_typeid bad_cast

 

std::numeric_limits

静态常量

numeric_limits::is_specialized

numeric_limits::is_signed

numeric_limits::is_integer

numeric_limits::is_exact

numeric_limits::has_infinity

numeric_limits::has_quiet_NaN

numeric_limits::has_signaling_NaN

numeric_limits::has_denorm

numeric_limits::has_denorm_loss

numeric_limits::round_style

numeric_limits::is_iec559

numeric_limits::is_bounded

numeric_limits::is_modulo

numeric_limits::digits

numeric_limits::digits10

numeric_limits::max_digits10 (C++11)

numeric_limits::radix

numeric_limits::min_exponent

numeric_limits::min_exponent10

numeric_limits::max_exponent

numeric_limits::max_exponent10

numeric_limits::traps

numeric_limits::tinyness_before

静态成员函数

numeric_limits::min

numeric_limits::lowest (C++11)

numeric_limits::max

numeric_limits::epsilon

numeric_limits::round_error

numeric_limits::infinity

numeric_limits::quiet_NaN

numeric_limits::signaling_NaN

numeric_limits::denorm_min

辅助类型

float_round_style

float_denorm_style

 

static T epsilon() throw();		(直到 C++11)
static constexpr T epsilon() noexcept;		(自 C++11 起)

返回机器 epsilon，即 1.0 与浮点数类型 T 可表示的下一个值之间的差。它仅在 std::numeric_limits<T>::is_integer == false 时有意义。

返回值

T	std::numeric_limits<T>::epsilon()
/* 非特化 */	T()
bool	false
char	​0​
signed char	​0​
unsigned char	​0​
wchar_t	​0​
char8_t (自 C++20 起)	​0​
char16_t (自 C++11 起)	​0​
char32_t (自 C++11 起)	​0​
short	​0​
unsigned short	​0​
int	​0​
unsigned int	​0​
long	​0​
unsigned long	​0​
long long (自 C++11 起)	​0​
unsigned long long(自 C++11 起)	​0​
float	FLT_EPSILON
double	DBL_EPSILON
long double	LDBL_EPSILON

示例

演示使用机器 epsilon 来比较浮点数的值以判断是否相等

运行此代码

#include <algorithm>
#include <cmath>
#include <cstddef>
#include <iomanip>
#include <iostream>
#include <limits>
#include <type_traits>
 
template <class T>
std::enable_if_t<not std::numeric_limits<T>::is_integer, bool>
equal_within_ulps(T x, T y, std::size_t n)
{
    // Since `epsilon()` is the gap size (ULP, unit in the last place)
    // of floating-point numbers in interval [1, 2), we can scale it to
    // the gap size in interval [2^e, 2^{e+1}), where `e` is the exponent
    // of `x` and `y`.
 
    // If `x` and `y` have different gap sizes (which means they have
    // different exponents), we take the smaller one. Taking the bigger
    // one is also reasonable, I guess.
    const T m = std::min(std::fabs(x), std::fabs(y));
 
    // Subnormal numbers have fixed exponent, which is `min_exponent - 1`.
    const int exp = m < std::numeric_limits<T>::min()
                  ? std::numeric_limits<T>::min_exponent - 1
                  : std::ilogb(m);
 
    // We consider `x` and `y` equal if the difference between them is
    // within `n` ULPs.
    return std::fabs(x - y) <= n * std::ldexp(std::numeric_limits<T>::epsilon(), exp);
}
 
int main()
{
    double x = 0.3;
    double y = 0.1 + 0.2;
    std::cout << std::hexfloat;
    std::cout << "x = " << x << '\n';
    std::cout << "y = " << y << '\n';
    std::cout << (x == y ? "x == y" : "x != y") << '\n';
    for (std::size_t n = 0; n <= 10; ++n)
        if (equal_within_ulps(x, y, n))
        {
            std::cout << "x equals y within " << n << " ulps" << '\n';
            break;
        }
}

输出

x = 0x1.3333333333333p-2
y = 0x1.3333333333334p-2
x != y
x equals y within 1 ulps

另请参见

nextafternextafterfnextafterlnexttowardnexttowardfnexttowardl (C++11)(C++11)(C++11)(C++11)(C++11)(C++11)

下一个可表示的浮点数，朝给定值的方向 (函数)