Minotaur::MaxInfBrancher Class Reference

A class to select a variable for branching using maximum-inference branching. More...

#include <MaxInfBrancher.h>

Inheritance diagram for Minotaur::MaxInfBrancher:

Collaboration diagram for Minotaur::MaxInfBrancher:

Public Member Functions
	MaxInfBrancher (EnvPtr env, HandlerVector &handlers)
	Construct using an environment pointer and initialize.
	~MaxInfBrancher ()
	Destroy.
Branches	findBranches (RelaxationPtr rel, NodePtr node, ConstSolutionPtr sol, SolutionPoolPtr s_pool, BrancherStatus &br_status, ModVector &mods)
	Find a branching candidate.
void	updateAfterSolve (NodePtr node, ConstSolutionPtr sol)
	Update pseudo-costs after LP is solved.
void	writeStats (std::ostream &out) const
	Write statistics to the given out stream.
bool	getTrustCutoff ()
	Return value of trustCutoff parameter.
UInt	getIterLim ()
	Get iteration limit of engine.
std::string	getName () const
	Return the name of this brancher.
UInt	getThresh () const
	Return the threshhold value.
void	initialize (RelaxationPtr rel)
	Initialize data structures.
void	setTrustCutoff (bool val)
	Set value of trustCutoff parameter.
void	setEngine (EnginePtr engine)
	Set engine.
void	setIterLim (UInt k)
	Set iteration limit of engine.
void	setMaxDepth (UInt k)
	Set the depth at which we stop strong branching.
void	setMinNodeDist (UInt k)
	Don't do strong branching on a cand if we did it 'k' nodes or less ago.
void	setThresh (UInt k)
	Set reliability threshhold.
void	printMaxInferenceScores () const
void	infScore ()
void	statsTable () const
double	gFunction (double x) const
Public Member Functions inherited from Minotaur::Brancher
	Brancher ()
	Default constructor.
virtual	~Brancher ()
	Destroy.

Public Attributes
std::vector< double >	upScores_
std::vector< double >	downScores_

Additional Inherited Members
Protected Attributes inherited from Minotaur::Brancher
LoggerPtr	logger_
	Log manager.

Detailed Description

A class to select a variable for branching using maximum-inference branching.

Member Function Documentation

findBranches()

Branches MaxInfBrancher::findBranches ( RelaxationPtr  rel, NodePtr  node, ConstSolutionPtr  sol, SolutionPoolPtr  s_pool, BrancherStatus &  br_status, ModVector &  mods  )

virtual

Find a branching candidate.

Returns

NULL if x does not have any fractional values for integer constrained variables or if no branching candidates are found (e.g. when we realize that problem is infeasible).

Parameters

[in]	rel	Relaxation at the current node.
[in]	node	The current node.
[in]	sol	The solution at the current node.
[in]	s_pool	Solution pool containing known feasible solutions.
[out]	br_status	Status returned by this brancher.
[out]	mods	Modification returned by this brancher. NULL if none found.

Implements Minotaur::Brancher.

getName()

std::string MaxInfBrancher::getName ( ) const

virtual

Return the name of this brancher.

Implements Minotaur::Brancher.

initialize()

void MaxInfBrancher::initialize

(

RelaxationPtr

rel

)

Initialize data structures.

Parameters

[in]

rel

Relaxation for which this brancher is used.

setEngine()

void MaxInfBrancher::setEngine

(

EnginePtr

engine

)

Set engine.

Parameters

[in]

engine

Pointer to the engine used in strong branching.

setIterLim()

void MaxInfBrancher::setIterLim

(

UInt

k

)

Set iteration limit of engine.

Parameters

[in]

k

The new iteration limit.

setMaxDepth()

void MaxInfBrancher::setMaxDepth

(

UInt

k

)

Set the depth at which we stop strong branching.

Parameters

[in]

k

The new depth value.

setMinNodeDist()

void MaxInfBrancher::setMinNodeDist

(

UInt

k

)

Don't do strong branching on a cand if we did it 'k' nodes or less ago.

Parameters

[in]

k

The new minimum value.

setThresh()

void MaxInfBrancher::setThresh

(

UInt

k

)

Set reliability threshhold.

The pseudo cost is considered reliable if it has been updated k times.

Parameters

[in]

k

The new threshhold value.

updateAfterSolve()

void MaxInfBrancher::updateAfterSolve ( NodePtr  node, ConstSolutionPtr  sol  )

virtual

Update pseudo-costs after LP is solved.

Parameters

[in]	node	The node for which relaxation is solved. The pseudo cost of branching candidate used in node->parent is updated.
[in]	sol	The solution of the relaxation at this node (not the parent).

Reimplemented from Minotaur::Brancher.

writeStats()

void MaxInfBrancher::writeStats ( std::ostream &  ) const

virtual

Write statistics to the given out stream.

Reimplemented from Minotaur::Brancher.

---

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

• /home/amahajan/tmp/minotaur-test/src/base/MaxInfBrancher.h
• /home/amahajan/tmp/minotaur-test/src/base/MaxInfBrancher.cpp