limbo::solvers::SubGradientDescent< T > Class Template Reference

Update lagrangian multiplier with subgradient descent. More...

#include <MultiKnapsackLagRelax.h>

Inheritance diagram for limbo::solvers::SubGradientDescent< T >:

Public Types
typedef LagMultiplierUpdater< T >	base_type
	base type
typedef base_type::value_type	value_type
	value type
Public Types inherited from limbo::solvers::LagMultiplierUpdater< T >
typedef T	value_type
	value type

Public Member Functions
	SubGradientDescent (value_type alpha=1, value_type beta=1)
	constructor More...
	SubGradientDescent (SubGradientDescent const &rhs)
	copy constructor More...
SubGradientDescent &	operator= (SubGradientDescent const &rhs)
	assignment More...
	~SubGradientDescent ()
	destructor
value_type	operator() (unsigned int iter, value_type multiplier, value_type slackness)
	API to update lagrangian multiplier using subgradient descent. More...
void	operator() (unsigned int iter, unsigned int n, value_type const *vSlackness, value_type const *vLagMultiplier, value_type *vNewLagMultiplier)
	API to update lagrangian multiplier using subgradient descent. More...
Public Member Functions inherited from limbo::solvers::LagMultiplierUpdater< T >
	LagMultiplierUpdater ()
	constructor
virtual	~LagMultiplierUpdater ()
	destructor

Protected Member Functions
void	copy (SubGradientDescent const &rhs)
	copy object More...
void	computeScalingFactor (unsigned int iter)
	compute scaling factor More...

Protected Attributes
value_type	m_alpha
	power
value_type	m_beta
	constant
unsigned int	m_iter
	current iteration
value_type	m_scalingFactor
	scaling factor \( t_k = \beta \cdot k^{-\alpha} \)

Detailed Description

template<typename T>
class limbo::solvers::SubGradientDescent< T >

Update lagrangian multiplier with subgradient descent.

Template Parameters

T	coefficient value type

Definition at line 25 of file MultiKnapsackLagRelax.h.

Constructor & Destructor Documentation

template<typename T>

limbo::solvers::SubGradientDescent< T >::SubGradientDescent ( value_type  alpha = 1, value_type  beta = 1  )

inline

constructor

Parameters

alpha	the power term for scaling factor \( t_k = \beta \cdot k^{-\alpha} \)
beta	the constant

Definition at line 973 of file MultiKnapsackLagRelax.h.

template<typename T>

limbo::solvers::SubGradientDescent< T >::SubGradientDescent ( SubGradientDescent< T > const &  rhs )

inline

copy constructor

right hand side

Definition at line 983 of file MultiKnapsackLagRelax.h.

Member Function Documentation

template<typename T>

void limbo::solvers::SubGradientDescent< T >::computeScalingFactor ( unsigned int  iter )

inlineprotected

compute scaling factor

Parameters

iter	current iteration

Definition at line 1048 of file MultiKnapsackLagRelax.h.

template<typename T>

void limbo::solvers::SubGradientDescent< T >::copy ( SubGradientDescent< T > const &  rhs )

inlineprotected

copy object

right hand side

Definition at line 1039 of file MultiKnapsackLagRelax.h.

template<typename T>

value_type limbo::solvers::SubGradientDescent< T >::operator() ( unsigned int  iter, value_type  multiplier, value_type  slackness  )

inlinevirtual

API to update lagrangian multiplier using subgradient descent.

Parameters

iter	current iteration
multiplier	current multiplier value
slackness	current slackness value assuming the constraint is in \( Ax \le b \) and compute \( b-Ax \)

Returns

updated multiplier value

Implements limbo::solvers::LagMultiplierUpdater< T >.

Definition at line 1008 of file MultiKnapsackLagRelax.h.

template<typename T>

void limbo::solvers::SubGradientDescent< T >::operator() ( unsigned int  iter, unsigned int  n, value_type const *  vSlackness, value_type const *  vLagMultiplier, value_type *  vNewLagMultiplier  )

inlinevirtual

API to update lagrangian multiplier using subgradient descent.

Parameters

iter	current iteration
n	dimension
vSlackness	array of slackness
vLagMultiplier	array of lagrangian multipliers
vNewLagMultiplier	array of new lagrangian multipliers

Implements limbo::solvers::LagMultiplierUpdater< T >.

Definition at line 1021 of file MultiKnapsackLagRelax.h.

template<typename T>

SubGradientDescent& limbo::solvers::SubGradientDescent< T >::operator= ( SubGradientDescent< T > const &  rhs )

inline

assignment

right hand side

Definition at line 989 of file MultiKnapsackLagRelax.h.

---

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

• /Users/yibolin/Documents/Projects/Limbo/limbo/solvers/MultiKnapsackLagRelax.h