Detailed Description

Concept for real scalar types.

A real type is an ordered type that supports arithmetic operations. Moreover,

it must be constructible from integer and floating-points. It must also be constructible from a default constructor that takes no arguments.
it must implement the copy assignment operator =.
it must support the math operators abs(), sqrt(), pow(int,), log2(), ceil(), and floor(). Those functions must be callable from the namespace tlapack.
it must support the boolean functions isinf() and isnan(), which must be callable from the namespace tlapack.
it must implement the functions min(T,T) and max(T,T) to decide the minimum and maximum in a pair of values.
it must have a specialization of std::numeric_limits<>.

Template Parameters

T Type.

Note: Note that the functions min() and max() may not propagate NaNs. The implementation in the C++ standard library says nothing about NaN propagation in std::min() and std::max() and we follow the same rule. The C++ documentation shows possible implementations that do not propagate NaNs. See: https://en.cppreference.com/w/cpp/algorithm/max and https://en.cppreference.com/w/cpp/algorithm/min.

The documentation for this interface was generated from the following file: