hermes::EFloat Class Reference

Represents a value with error bounds. More...

Public Member Functions
HERMES_DEVICE_CALLABLE	EFloat ()
	Default constructor.
HERMES_DEVICE_CALLABLE	EFloat (f32 v, f32 e=0.f)
	Constructs from value and error size. More...
HERMES_DEVICE_CALLABLE	operator float () const
	Casts to 32 bit floating point. More...
HERMES_DEVICE_CALLABLE	operator double () const
	Casts to 64 bit floating point. More...
HERMES_DEVICE_CALLABLE EFloat	operator+ (EFloat f) const
HERMES_DEVICE_CALLABLE EFloat	operator- (EFloat f) const
HERMES_DEVICE_CALLABLE EFloat	operator* (EFloat f) const
HERMES_DEVICE_CALLABLE EFloat	operator/ (EFloat f) const
HERMES_DEVICE_CALLABLE bool	operator== (EFloat f) const
HERMES_DEVICE_CALLABLE f32	absoluteError () const
	Computes the absolute error carried by this number. More...
HERMES_DEVICE_CALLABLE f32	upperBound () const
	Computes the upper error bound carried by this number. More...
HERMES_DEVICE_CALLABLE f32	lowerBound () const
	Computes the lower error bound carried by this number. More...

Friends
HERMES_DEVICE_CALLABLE friend EFloat	operator* (f32 f, EFloat fe)
HERMES_DEVICE_CALLABLE friend EFloat	operator/ (f32 f, EFloat fe)
HERMES_DEVICE_CALLABLE friend EFloat	operator+ (f32 f, EFloat fe)
HERMES_DEVICE_CALLABLE friend EFloat	operator- (f32 f, EFloat fe)

Detailed Description

Represents a value with error bounds.

Implements the running error analysis by carrying the error bounds accumulated by a floating point value. It keeps track of the interval of uncertainty of the computed value.

Constructor & Destructor Documentation

EFloat()

HERMES_DEVICE_CALLABLE hermes::EFloat::EFloat ( f32  v, f32  e = 0.f  )

inlineexplicit

Constructs from value and error size.

Parameters

v	floating point value
e	absolute error bound

Member Function Documentation

absoluteError()

HERMES_DEVICE_CALLABLE float hermes::EFloat::absoluteError ( ) const

inline

Computes the absolute error carried by this number.

Returns

a bound for the absolute error

lowerBound()

HERMES_DEVICE_CALLABLE float hermes::EFloat::lowerBound ( ) const

inline

Computes the lower error bound carried by this number.

Returns

upper error interval bound

operator double()

HERMES_DEVICE_CALLABLE hermes::EFloat::operator double ( ) const

inlineexplicit

Casts to 64 bit floating point.

Returns

operator float()

HERMES_DEVICE_CALLABLE hermes::EFloat::operator float ( ) const

inlineexplicit

Casts to 32 bit floating point.

Returns

upperBound()

HERMES_DEVICE_CALLABLE float hermes::EFloat::upperBound ( ) const

inline

Computes the upper error bound carried by this number.

Returns

lower error interval bound

---

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

• hermes/numeric/e_float.h
• hermes/numeric/e_float.cpp