numeric_limits<T>::min();
numeric_limits<T>::lowest();
What is the different between the value returned by both functions?
Paragraph 18.3.2.4 of the C++11 Standard specifies:
static constexpr T min() noexcept
;1 Minimum finite value.
2 For floating types with denormalization, returns the minimum positive normalized value.
3 Meaningful for all specializations
[...]
static constexpr T lowest() noexcept
;6 A finite value x such that there is no other finite value y where y < x.
7 Meaningful for all specializations in which is_bounded != false.
Footnote 197 then adds the relevant remark:
lowest()
is necessary because not all floating-point representations have a smallest (most negative) value that is the negative of the largest (most positive) finite value.