Public Member Functions
	BackTrackingLineSearch (int nMaxIters=20, TScalar tau=TScalar(0.5), TScalar c=TScalar(1e-4), TScalar alpha=TScalar(1))
	Construct a new Back Tracking Line Search object.

template<class FObjective, class TDerivedG, class TDerivedDX, class TDerivedX>
TScalar	Solve (FObjective f, Eigen::MatrixBase< TDerivedG > const &g, Eigen::MatrixBase< TDerivedDX > const &dx, Eigen::MatrixBase< TDerivedX > &xk) const
	Perform a backtracking line search.

Public Attributes
int	nMaxIters {20}
	Maximum number of iterations for the line search.

TScalar	tau {0.5}
	Step size decrease factor.

TScalar	c {1e-4}
	Armijo slope scale.

TScalar	alpha {1}
	Initial step size.

Constructor & Destructor Documentation

◆ BackTrackingLineSearch()

template<class TScalar>

pbat::math::optimization::BackTrackingLineSearch< TScalar >::BackTrackingLineSearch	(	int	nMaxIters = 20,
		TScalar	tau = TScalar(0.5),
		TScalar	c = TScalar(1e-4),
		TScalar	alpha = TScalar(1) )

inline

Construct a new Back Tracking Line Search object.

Parameters

nMaxIters	Maximum number of iterations for the line search
tau	Step size decrease factor
c	Armijo slope scale
alpha	Initial step size

template<class TScalar>

template<class FObjective, class TDerivedG, class TDerivedDX, class TDerivedX>

TScalar pbat::math::optimization::BackTrackingLineSearch< TScalar >::Solve	(	FObjective	f,
		Eigen::MatrixBase< TDerivedG > const &	g,
		Eigen::MatrixBase< TDerivedDX > const &	dx,
		Eigen::MatrixBase< TDerivedX > &	xk ) const

inline

Perform a backtracking line search.

Template Parameters

FObjective	Callable type for the objective function with signature `f(xk) -> fk`
TDerivedG	Derived type for the gradient
TDerivedDX	Derived type for the step direction
TDerivedX	Derived type for the current iterate

Parameters

f	Objective function
g	Gradient at the initial iterate
dx	Step direction
xk	Current iterate

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