Opm::cmp Namespace Reference

Functions
template<typename T >
bool	scalar_equal (T value1, T value2, T abs_eps, T rel_eps)
template<typename T >
bool	scalar_equal (T value1, T value2)
template<typename T >
bool	vector_equal (const std::vector< T > &v1, const std::vector< T > &v2, T abs_eps, T rel_eps)
template<typename T >
bool	vector_equal (const std::vector< T > &v1, const std::vector< T > &v2)
template<typename T >
bool	array_equal (const T *p1, const T *p2, size_t num_elements, T abs_eps, T rel_eps)
template<typename T >
bool	array_equal (const T *p1, const T *p2, size_t num_elements)

Variables
const double	default_abs_epsilon = 1e-8
const double	default_rel_epsilon = 1e-5

Detailed Description

In the namespace cmp are implemented functions for approximate comparison of double values based on absolute and relative difference. There are three functions:

scalar_equal<T>() : Compare two <T> values.

ptr_equal<T>(): This compares all the element in the two T * pointers.

vector_equal<T>(): This compares all the elements in two std::vector<T> instances.

For both vector_equal<T>() and ptr_equal<T>() the actual comparison is based on the scalar_equal<T>() function. All functions exist as two overloads, one which takes explicit input values for the absolute and relative epsilon, and one which uses default values.

The comparison functions are implemented as templates, with the following caveats:

1. The static_assert() in scalar_equal<T> ensures that only floating point types can be used.
2. The default epsilon values are of type double - irrespective of the type of data being compared.

For more details of floating point comparison please consult this reference:

https://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/

Function Documentation

array_equal() [1/2]

template<typename T >

bool Opm::cmp::array_equal	(	const T *	p1,
		const T *	p2,
		size_t	num_elements
	)

References default_abs_epsilon, and default_rel_epsilon.

array_equal() [2/2]

template<typename T >

bool Opm::cmp::array_equal	(	const T *	p1,
		const T *	p2,
		size_t	num_elements,
		T	abs_eps,
		T	rel_eps
	)

References index.

scalar_equal() [1/2]

template<typename T >

bool Opm::cmp::scalar_equal	(	T	value1,
		T	value2
	)

References default_abs_epsilon, and default_rel_epsilon.

scalar_equal() [2/2]

template<typename T >

bool Opm::cmp::scalar_equal	(	T	value1,
		T	value2,
		T	abs_eps,
		T	rel_eps
	)

References exprtk::details::numeric::equal(), exprtk::details::numeric::max(), and exprtk::details::value().

vector_equal() [1/2]

template<typename T >

bool Opm::cmp::vector_equal	(	const std::vector< T > &	v1,
		const std::vector< T > &	v2
	)

References default_abs_epsilon, and default_rel_epsilon.

vector_equal() [2/2]

template<typename T >

bool Opm::cmp::vector_equal	(	const std::vector< T > &	v1,
		const std::vector< T > &	v2,
		T	abs_eps,
		T	rel_eps
	)

Variable Documentation

default_abs_epsilon

const double Opm::cmp::default_abs_epsilon = 1e-8

Referenced by array_equal(), scalar_equal(), and vector_equal().

default_rel_epsilon

const double Opm::cmp::default_rel_epsilon = 1e-5

Referenced by array_equal(), scalar_equal(), and vector_equal().