mitsuba::BrentSolver Class Reference

Brent's method nonlinear zero finder. More...

#include <mitsuba/core/brent.h>

Classes
struct	Result
	Return value of BrentSolver::solve() More...

Public Member Functions
	BrentSolver (size_t maxIterations=100, Float absAccuracy=1e-6f, Float absAccuracyPos=1e-6f, Float relAccuracyPos=1e-6f)
	Create a new Brent-style solver with the specified accuracy requirements. More...
Result	solve (const boost::function< Float(Float)> &func, Float min, Float max) const
	Find a zero in the given interval. More...
Result	solve (const boost::function< Float(Float)> &func, Float min, Float max, Float initial) const
	Find a zero in the given interval with an initial guess. More...
Result	solve (const boost::function< Float(Float)> &func, Float x0, Float y0, Float x1, Float y1, Float x2, Float y2) const

Protected Attributes
size_t	m_maxIterations
Float	m_absAccuracy
Float	m_absAccuracyPos
Float	m_relAccuracyPos

Detailed Description

Brent's method nonlinear zero finder.

The implementation is transcribed from the Apache Commons Java implementation. The supplied function is required to be continuous, but not necessarily smooth.

Constructor & Destructor Documentation

mitsuba::BrentSolver::BrentSolver ( size_t  maxIterations = 100, Float  absAccuracy = 1e-6f, Float  absAccuracyPos = 1e-6f, Float  relAccuracyPos = 1e-6f  )

inline

Create a new Brent-style solver with the specified accuracy requirements.

Parameters

maxIterations	Max. number of successive iterations (default: 100)
absAccuracy	Absolute accuracy requirement – the iterations will stop when |f(x)| < absAccuracy.
absAccuracyPos	Absolute accuracy requirement of the position – the iterations will stop when |minX-maxX| < absAccuracyPos.
absAccuracyPos	Absolute accuracy requirement of the position – the iterations will stop when |minX-maxX|/minX < relAccuracyPos.

Member Function Documentation

Result mitsuba::BrentSolver::solve	(	const boost::function< Float(Float)> &	func,
		Float	min,
		Float	max
	)		const

Find a zero in the given interval.

Requires that the values of the function at the endpoints have opposite signs.

Parameters

min	the lower bound for the interval.
max	the upper bound for the interval.

Returns

the value where the function is zero

Result mitsuba::BrentSolver::solve	(	const boost::function< Float(Float)> &	func,
		Float	min,
		Float	max,
		Float	initial
	)		const

Find a zero in the given interval with an initial guess.

Requires that the values of the function at the endpoints have opposite signs (note that it is allowed to have endpoints with the same sign if the initial point has opposite sign function-wise).

Parameters

min	the lower bound for the interval.
max	the upper bound for the interval.
initial	the start value to use (must be set to min if no initial point is known)

Returns

the value where the function is zero

Result mitsuba::BrentSolver::solve	(	const boost::function< Float(Float)> &	func,
		Float	x0,
		Float	y0,
		Float	x1,
		Float	y1,
		Float	x2,
		Float	y2
	)		const

Find a zero starting search according to the three provided points.

Parameters

x0	old approximation for the root
y0	function value at the approximation for the root
x1	last calculated approximation for the root
y1	function value at the last calculated approximation for the root
x2	bracket point (must be set to x0 if no bracket point is known, this will force starting with linear interpolation)
y2	function value at the bracket point.

Returns

the value where the function is zero

Member Data Documentation

Float mitsuba::BrentSolver::m_absAccuracy

protected

Float mitsuba::BrentSolver::m_absAccuracyPos

protected

size_t mitsuba::BrentSolver::m_maxIterations

protected

Float mitsuba::BrentSolver::m_relAccuracyPos

protected

---

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

• include/mitsuba/core/brent.h