nike::RelativeEpsilonEqualFunctor Class Reference

#include <RelativeEpsilonEqual.h>

Public Member Functions
bool	operator() (float lhs, float rhs) const

Friends
RelativeEpsilonEqualFunctor	relative_epsilon_equal (unsigned)

Detailed Description

Definition at line 23 of file RelativeEpsilonEqual.h.

Member Function Documentation

operator()()

bool nike::RelativeEpsilonEqualFunctor::operator()	(	float	lhs,
		float	rhs
	)		const

Definition at line 27 of file RelativeEpsilonEqual.cpp.

{
  if (std::isnan(lhs) && std::isnan(rhs))
  {
    return true;
  }
 
  // TODO How to handle sign difference?
  auto const delta = std::fabs(lhs - rhs);
  auto const max = std::max(std::fabs(lhs), std::fabs(rhs));
 
  return delta <= (max * FLT_EPSILON * _tolerance);
}

Friends And Related Symbol Documentation

relative_epsilon_equal

RelativeEpsilonEqualFunctor relative_epsilon_equal ( unsigned  tolerance )

friend

Note

RelativeEpsilonEqualFunctor uses its own rule for NaN values.

For example, "NAN == NAN" is false but "relative_epsilon_equal(1)(NAN, NAN)" is true.

Definition at line 41 of file RelativeEpsilonEqual.cpp.

{
  return RelativeEpsilonEqualFunctor{tolerance};
}

---

The documentation for this class was generated from the following files:

• compiler/nike/include/nike/RelativeEpsilonEqual.h
• compiler/nike/src/RelativeEpsilonEqual.cpp