ABA_MASTER Class Reference

Class ABA_MASTER is the central object of the framework. The most important tasks of the class ABA_MASTER is the management of the implicit enumeration. Moreover, it provides already default implementations for constraints, cutting planes, and variables pools. More...

#include <master.h>

Inheritance diagram for ABA_MASTER:

List of all members.


Public Types
enum	STATUS { Optimal, Error, OutOfMemory, Unprocessed, Processing, Guaranteed, MaxLevel, MaxCpuTime, MaxCowTime, ExceptionFathom }
enum	OUTLEVEL { Silent, Statistics, Subproblem, LinearProgram, Full }
enum	ENUMSTRAT { BestFirst, BreadthFirst, DepthFirst, DiveAndBest }
enum	BRANCHINGSTRAT { CloseHalf, CloseHalfExpensive }
	This enumeration defines the two currently implemented branching variable selection strategies. More...
enum	PRIMALBOUNDMODE { NoPrimalBound, Optimum, OptimumOne }
	This enumeration provides various methods for the initialization of the primal bound. More...
enum	SKIPPINGMODE { SkipByNode, SkipByLevel }
enum	CONELIMMODE { NoConElim, NonBinding, Basic }
	This enumeration defines the ways for automatic constraint elimination during the cutting plane phase. More...
enum	VARELIMMODE { NoVarElim, ReducedCost }
	This enumeration defines the ways for automatic variable elimination during the column generation algorithm. More...
enum	VBCMODE { NoVbc, File, Pipe }
	This enumeration defines what kind of output can be generated for the VBCTOOL. More...
enum	OSISOLVER { Cbc, Clp, CPLEX, DyLP, FortMP, GLPK, MOSEK, OSL, SoPlex, SYMPHONY, Vol, XPRESS_MP }
	This enumeration defines which solvers can be used to solve theLP-relaxations. More...
Public Member Functions
	ABA_MASTER (const char *problemName, bool cutting, bool pricing, ABA_OPTSENSE::SENSE optSense=ABA_OPTSENSE::Unknown, double eps=1.0e-4, double machineEps=1.0e-7, double infinity=1.0e30, bool readParamFromFile=true)
virtual	~ABA_MASTER ()
	The destructor.
STATUS	optimize ()
ENUMSTRAT	enumerationStrategy () const
void	enumerationStrategy (ENUMSTRAT strat)
	This version of the function enumerationStrategy() changes the enumeration strategy.
virtual int	enumerationStrategy (const ABA_SUB s1, const ABA_SUB s2)
	Analyzes the enumeration strategy set in the parameter file { .abacus} and calls the corresponding comparison function for the subproblems s1 and s2. This function should be redefined for application specific enumeration strategies.
bool	guaranteed ()
	Can be used to check if the guarantee requirements are fulfilled, i.e., the difference between upper bound and the lower bound in respect to the lowerBound is less than this guarantee value in percent.
double	guarantee ()
void	printGuarantee ()
bool	check ()
	Can be used to control the correctness of the optimization if the value of the optimum solution has been loaded.
bool	knownOptimum (double &optVal)
	Opens the file specified with the parameter { OptimumFileName} in the configuration file { .abacus} and tries to find a line with the name of the problem instance (as specified in the constructor of ABA_MASTER) as first string.
virtual void	output ()
bool	cutting () const
bool	pricing () const
const ABA_OPTSENSE *	optSense () const
ABA_HISTORY *	history () const
ABA_OPENSUB *	openSub () const
ABA_STANDARDPOOL< ABA_CONSTRAINT, ABA_VARIABLE > *	conPool () const
ABA_STANDARDPOOL< ABA_CONSTRAINT, ABA_VARIABLE > *	cutPool () const
ABA_STANDARDPOOL< ABA_VARIABLE, ABA_CONSTRAINT > *	varPool () const
ABA_SUB *	root () const
ABA_SUB *	rRoot () const
STATUS	status () const
const ABA_STRING *	problemName () const
const ABA_COWTIMER *	totalCowTime () const
bool	solveApprox () const
const ABA_CPUTIMER *	totalTime () const
const ABA_CPUTIMER *	lpTime () const
const ABA_CPUTIMER *	lpSolverTime () const
const ABA_CPUTIMER *	separationTime () const
const ABA_CPUTIMER *	improveTime () const
const ABA_CPUTIMER *	pricingTime () const
const ABA_CPUTIMER *	branchingTime () const
int	nSub () const
int	nLp () const
int	highestLevel () const
int	nNewRoot () const
int	nSubSelected () const
void	printParameters ()
	Writes all parameters of the class ABA_MASTER together with their values to the global output stream.
BRANCHINGSTRAT	branchingStrategy () const
void	branchingStrategy (BRANCHINGSTRAT strat)
OSISOLVER	defaultLpSolver () const
void	defaultLpSolver (OSISOLVER osiSolver)
ABA_LPMASTEROSI *	lpMasterOsi () const
int	nBranchingVariableCandidates () const
void	nBranchingVariableCandidates (int n)
	This version of the function nbranchingVariableCandidates() sets the number of tested branching variable candidates.
double	requiredGuarantee () const
void	requiredGuarantee (double g)
	This version of the function requiredGuarantee() changes the guarantee specification.
int	maxLevel () const
void	maxLevel (int ml)
	This version of the function maxLevel() changes the maximal enumeration depth.
const ABA_STRING &	maxCpuTime () const
void	maxCpuTime (const ABA_STRING &t)
const ABA_STRING &	maxCowTime () const
	The function maxCowTime().
void	maxCowTime (const ABA_STRING &t)
	This version of the function maxCowTime() set the maximal wall-clock time for the optimization.
bool	objInteger () const
void	objInteger (bool b)
	This version of function objInteger() sets the assumption that the objective function values of all feasible solutions are integer.
int	tailOffNLp () const
	The function tailOffNLp().
void	tailOffNLp (int n)
double	tailOffPercent () const
	The function tailOffPercent().
void	tailOffPercent (double p)
	This version of the function tailOffPercent() sets the minimal change of the dual bound for the tailing off analysis.
OUTLEVEL	outLevel () const
void	outLevel (OUTLEVEL mode)
	The version of the function outLevel() sets the output mode.
OUTLEVEL	logLevel () const
void	logLevel (OUTLEVEL mode)
	This version of the function logLevel() sets the output mode for the log-file.
bool	delayedBranching (int nOpt_) const
void	dbThreshold (int threshold)
	Sets the number of optimizations of a subproblem until sons are created in ABA_SUB::branching().
int	dbThreshold () const
int	minDormantRounds () const
void	minDormantRounds (int nRounds)
PRIMALBOUNDMODE	pbMode () const
void	pbMode (PRIMALBOUNDMODE mode)
int	pricingFreq () const
void	pricingFreq (int f)
	This version of the function pricingFreq() sets the number of linear programs being solved between two additional pricing steps.
int	skipFactor () const
void	skipFactor (int f)
	This version of the function skipFactor() sets the frequency for constraint and variable generation.
void	skippingMode (SKIPPINGMODE mode)
	This version of the function skippingMode() sets the skipping strategy.
SKIPPINGMODE	skippingMode () const
CONELIMMODE	conElimMode () const
void	conElimMode (CONELIMMODE mode)
VARELIMMODE	varElimMode () const
void	varElimMode (VARELIMMODE mode)
double	conElimEps () const
void	conElimEps (double eps)
double	varElimEps () const
void	varElimEps (double eps)
int	varElimAge () const
void	varElimAge (int eps)
int	conElimAge () const
void	conElimAge (int eps)
bool	fixSetByRedCost () const
void	fixSetByRedCost (bool on)
bool	printLP () const
void	printLP (bool on)
int	maxConAdd () const
void	maxConAdd (int max)
	Sets the maximal number of constraints that are added in an iteration of the cutting plane algorithm.
int	maxConBuffered () const
void	maxConBuffered (int max)
	Changes the maximal number of constraints that are buffered in an iteration of the cutting plane algorithm.
int	maxVarAdd () const
void	maxVarAdd (int max)
	Changes the maximal number of variables that are added in an iteration of the subproblem optimization.
int	maxVarBuffered () const
void	maxVarBuffered (int max)
	Changes the maximal number of variables that are buffered in an iteration of the subproblem optimization.
int	maxIterations () const
void	maxIterations (int max)
	Changes the default value for the maximal number of iterations of the optimization of a subproblem.
bool	eliminateFixedSet () const
void	eliminateFixedSet (bool turnOn)
	This version of the function eliminateFixedSet() can be used to turn the elimination of fixed and set variables on or off.
bool	newRootReOptimize () const
void	newRootReOptimize (bool on)
const ABA_STRING &	optimumFileName () const
void	optimumFileName (const char *name)
bool	showAverageCutDistance () const
void	showAverageCutDistance (bool on)
	Turns the output of the average distance of the added cuts from the fractional solution on or off.
VBCMODE	vbcLog () const
void	vbcLog (VBCMODE mode)
virtual bool	setSolverParameters (OsiSolverInterface *interface, bool solverIsApprox)
bounds
In order to embed both minimization and maximization problems in this system we work internally with primal bounds, i.e., a value which is worse than the best known solution (often a value of a feasible solution), and dual bounds, i.e., a bound which is better than the best known solution. Primal and dual bounds are then interpreted as lower or upper bounds according to the sense of the optimization.
double	lowerBound () const
double	upperBound () const
double	primalBound () const
void	primalBound (double x)
	This version of the function primalBound() sets the primal bound to x and makes a new entry in the solution history. It is an error if the primal bound gets worse.
double	dualBound () const
void	dualBound (double x)
	This version of the function dualBound() sets the dual bound to x and makes a new entry in the solution history.
bool	betterDual (double x) const
bool	primalViolated (double x) const
bool	betterPrimal (double x) const
bool	feasibleFound () const
	We use this function ,e.g., to adapt the enumeration strategy in the DiveAndBest-Strategy.
Static Public Attributes
static const char *	STATUS_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { STATUS[0]=="Optimal"}).
static const char *	OUTLEVEL_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { OUTLEVEL[0]=="Silent"}).
static const char *	ENUMSTRAT_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { ENUMSTRAT[0]=="BestFirst"}).
static const char *	BRANCHINGSTRAT_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { BRANCHINGSTRAT[0]=="CloseHalf"}).
static const char *	PRIMALBOUNDMODE_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { PRIMALBOUNDMODE[0]=="None"}).
static const char *	SKIPPINGMODE_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { SKIPPINGMODE[0]=="None"}).
static const char *	CONELIMMODE_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { CONELIMMODE[0]=="None"}).
static const char *	VARELIMMODE_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { VARELIMMODE[0]=="None"}).
static const char *	VBCMODE_ []
	Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { VBCMODE[0]=="None"}).
static const char *	OSISOLVER_ []
	Array for the literal values for possible Osi solvers.
Protected Member Functions
virtual void	initializePools (ABA_BUFFER< ABA_CONSTRAINT * > &constraints, ABA_BUFFER< ABA_VARIABLE * > &Variables, int varPoolSize, int cutPoolSize, bool dynamicCutPool=false)
virtual void	initializePools (ABA_BUFFER< ABA_CONSTRAINT * > &constraints, ABA_BUFFER< ABA_CONSTRAINT * > &cuts, ABA_BUFFER< ABA_VARIABLE * > &Variables, int varPoolSize, int cutPoolSize, bool dynamicCutPool=false)
	Is overloaded such that also a first set of cutting planes can be inserted into the cutting plane pool.
void	initializeOptSense (ABA_OPTSENSE::SENSE sense)
	Can be used to initialize the sense of the optimization in derived classes, if this has not been already performed when the constructor of ABA_MASTER has been called.
int	bestFirstSearch (const ABA_SUB s1, const ABA_SUB s2) const
virtual int	equalSubCompare (const ABA_SUB s1, const ABA_SUB s2) const
	Is called from the function bestFirstSearch() and from the function depthFirstSearch() if the subproblems s1 and s2 have the same priority.
int	depthFirstSearch (const ABA_SUB s1, const ABA_SUB s2) const
	Implements the depth first search enumeration strategy, i.e., the subproblem with maximum level is selected.
int	breadthFirstSearch (const ABA_SUB s1, const ABA_SUB s2) const
	Implements the breadth first search enumeration strategy, i.e., the subproblem with minimum level is selected.
int	diveAndBestFirstSearch (const ABA_SUB s1, const ABA_SUB s2) const
	Performs depth-first search until a feasible solution is found, then the search process is continued with best-first search.
virtual void	initializeParameters ()
	Is only a dummy. This function can be used to initialize parameters of derived classes and to overwrite parameters read from the file { .abacus} by the function ().
virtual ABA_SUB *	firstSub ()=0
virtual void	initializeOptimization ()
	The default implementation of initializeOptimization() does nothing.
virtual void	terminateOptimization ()
	The default implementation of terminateOptimization() does nothing.
Private Member Functions
void	_initializeParameters ()
	Reads the parameter-file { .abacus}, which is searched in the directory given by the environment variable ABACUS_DIR, and calls the virtual function initializeParameters() which can initialize parameters of derived classes and overwrite parameters of this class.
void	_createLpMasters ()
void	_deleteLpMasters ()
void	_initializeLpParameters ()
void	_setDefaultLpParameters ()
	Initializes the LP solver specific default Parameters if they are not read from the parameter-file { .abacus}.
void	_printLpParameters ()
void	_outputLpStatistics ()
ABA_SUB *	select ()
int	initLP ()
void	writeTreeInterface (const char *info, bool time=true) const
void	treeInterfaceNewNode (ABA_SUB *sub) const
	Adds the subproblem sub to the stream storing information for graphical output of the enumeration tree if this logging is turned on.
void	treeInterfacePaintNode (int id, int color) const
void	treeInterfaceLowerBound (double lb) const
void	treeInterfaceUpperBound (double ub) const
void	treeInterfaceNodeBounds (int id, double lb, double ub)
	Updates the node information in the node with number id by writing the lower bound lb and the upper bound ub to the node.
void	newSub (int level)
void	countLp ()
	Increments the counter for linear programs and should be called in each optimization call of the LP-relaxation.
void	newFixed (int n)
	Increments the counter of the number of fixed variables by n.
void	addCons (int n)
	Increments the counter for the total number of added constraints by n.
void	removeCons (int n)
	Increments the counter for the total number of removed constraints by n.
void	addVars (int n)
	Increments the counter for the total number of added variables by n.
void	removeVars (int n)
	Increments the counter for the total number of removed variables by n.
ABA_FIXCAND *	fixCand () const
void	rRoot (ABA_SUB *newRoot, bool reoptimize)
void	status (STATUS stat)
void	rootDualBound (double x)
void	theFuture ()
	ABA_MASTER (const ABA_MASTER &rhs)
const ABA_MASTER &	operator= (const ABA_MASTER &rhs)
Private Attributes
ABA_STRING	problemName_
bool	readParamFromFile_
ABA_OPTSENSE	optSense_
ABA_SUB *	root_
ABA_SUB *	rRoot_
ABA_OPENSUB *	openSub_
ABA_HISTORY *	history_
ENUMSTRAT	enumerationStrategy_
BRANCHINGSTRAT	branchingStrategy_
int	nBranchingVariableCandidates_
	The number of candidates that are evaluated for branching on variables.
OSISOLVER	defaultLpSolver_
ABA_LPMASTEROSI *	lpMasterOsi_
ABA_STANDARDPOOL< ABA_CONSTRAINT, ABA_VARIABLE > *	conPool_
ABA_STANDARDPOOL< ABA_CONSTRAINT, ABA_VARIABLE > *	cutPool_
ABA_STANDARDPOOL< ABA_VARIABLE, ABA_CONSTRAINT > *	varPool_
double	primalBound_
double	dualBound_
double	rootDualBound_
ABA_FIXCAND *	fixCand_
bool	cutting_
bool	pricing_
bool	solveApprox_
int	nSubSelected_
	The number of subproblems already selected from the list of open subproblems.
VBCMODE	VbcLog_
	Ouput for the Tree Interface is generated depending on the value of this variable.
ostream *	treeStream_
double	requiredGuarantee_
	The guarantee in percent which should be reached when the optimization stops.
int	maxLevel_
ABA_STRING	maxCpuTime_
ABA_STRING	maxCowTime_
bool	objInteger_
	true, if all objective function values of feasible solutions are assumed to be integer.
int	tailOffNLp_
double	tailOffPercent_
int	dbThreshold_
int	minDormantRounds_
	The minimal number of rounds, i.e., number of subproblem optimizations, a subproblem is dormant, i.e., it is not selected from the set of open subproblem if its status is Dormant, if possible.
OUTLEVEL	outLevel_
OUTLEVEL	logLevel_
PRIMALBOUNDMODE	pbMode_
int	pricingFreq_
int	skipFactor_
	The frequency constraints or variables are generated depending on the skipping mode.
SKIPPINGMODE	skippingMode_
	Either constraints are generated only every skipFactor_ subproblem (SkipByNode) only every skipFactor_ level (SkipByLevel).
bool	fixSetByRedCost_
bool	printLP_
int	maxConAdd_
	The maximal number of added constraints per iteration of the cutting plane algorithm.
int	maxConBuffered_
int	maxVarAdd_
	The maximal number of added variables per iteration of the column generation algorithm.
int	maxVarBuffered_
int	maxIterations_
	The maximal number of iterations of the cutting plane/column generation algorithm in the subproblem.
bool	eliminateFixedSet_
bool	newRootReOptimize_
	If true, then an already earlier processed node is reoptimized if it becomes the new root of the remaining \ tree.
ABA_STRING	optimumFileName_
	The name of a file storing a list of optimum solutions of problem instances.
bool	showAverageCutDistance_
	If true then the average distance of the added cutting planes is output every iteration of the cutting plane algorithm.
CONELIMMODE	conElimMode_
	The way constraints are automatically eliminated in the cutting plane algorithm.
VARELIMMODE	varElimMode_
	The way variables are automatically eliminated in the column generation algorithm.
double	conElimEps_
	The tolerance for the elimination of constraints by the mode NonBinding/.
double	varElimEps_
	The tolerance for the elimination of variables by the mode ReducedCost.
int	conElimAge_
	The number of iterations an elimination criterion must be satisfied until a constraint can be removed.
int	varElimAge_
	The number of iterations an elimination criterion must be satisfied until a variable can be removed.
STATUS	status_
ABA_COWTIMER	totalCowTime_
ABA_CPUTIMER	totalTime_
ABA_CPUTIMER	lpTime_
ABA_CPUTIMER	lpSolverTime_
ABA_CPUTIMER	separationTime_
ABA_CPUTIMER	improveTime_
	The timer for the cpu time spent in the heuristics for the computation of feasible solutions.
ABA_CPUTIMER	pricingTime_
ABA_CPUTIMER	branchingTime_
int	nSub_
int	nLp_
int	highestLevel_
int	nFixed_
int	nAddCons_
int	nRemCons_
int	nAddVars_
int	nRemVars_
int	nNewRoot_
Friends
class	ABA_SUB
class	ABA_FIXCAND

Detailed Description

Definition at line 76 of file master.h.

Member Enumeration Documentation

enum ABA_MASTER::STATUS

The various statuses of the optimization process.

Parameters:

	Optimal	The optimization terminated with an error and without reaching one of the resource limits. If there is a feasible solution then the optimal solution has been computed.
	Error	An error occurred during the optimization process.
	Unprocessed	The initial status, before the optimization starts.
	Processing	The status while the optimization is performed.
	Guaranteed	If not the optimal solution is determined, but the required guarantee is reached, then the status is Guaranteed.
	MaxLevel	The status, if subproblems are ignored since the maximum enumeration level is exceeded.
	MaxCpuTime	The status, if the optimization terminates since the maximum cpu time is exceeded.
	MaxCowTime	The status, if the optimization terminates since the maximum wall-clock time is exceeded.
	ExceptionFathom	The status, if at least one subproblem has been fathomed according to a problem specific criteria determined in the function ABA_SUB::exceptionFathom().

Enumerator:

Optimal
Error
OutOfMemory
Unprocessed
Processing
Guaranteed
MaxLevel
MaxCpuTime
MaxCowTime
ExceptionFathom

Definition at line 109 of file master.h.

enum ABA_MASTER::OUTLEVEL

This enumeration defines the different output levels:

Parameters:

	Silent	No output at all.
	Statistics	No output during the optimization, but output of final statistics.
	Subproblem	In addition to the previous level also a single line of output after every subproblem optimization.
	LinearProgram	In addition to the previous level also a single line of output after every solved linear program.
	Full	Tons of output.

Enumerator:

Silent
Statistics
Subproblem
LinearProgram
Full

Definition at line 131 of file master.h.

enum ABA_MASTER::ENUMSTRAT

Enumerator:

BestFirst
BreadthFirst
DepthFirst
DiveAndBest

Definition at line 158 of file master.h.

enum ABA_MASTER::BRANCHINGSTRAT

This enumeration defines the two currently implemented branching variable selection strategies.

Parameters:

	CloseHalf	Selects the variable with fractional part closest to .
	CloseHalfExpensive	Selects the variable with fractional part close to (within some interval around ) and has highest absolute objective function coefficient.

Enumerator:

CloseHalf
CloseHalfExpensive

Definition at line 175 of file master.h.

enum ABA_MASTER::PRIMALBOUNDMODE

This enumeration provides various methods for the initialization of the primal bound.

The modes OptimalPrimalBound and OptimalOnePrimalBound can be useful in the testing phase. For these modes the value of an optimum solution must stored in the file given by the parameter { OptimumFileName} in the parameter file.

Parameters:

	NoPrimalBound	The primal bound is initialized with $-\infty$ for maximization problems and $\infty$ for minimization problems, respectively.
	OptimalPrimalBound	The primal bound is initialized with the value of the optimum solution.
	OptimalOnePrimalBound	The primal bound is initialized with the value of optimum solution minus 1 for maximization problems and with the value of the optimum solution plus one for minimization problems, respectively.

Enumerator:

NoPrimalBound
Optimum
OptimumOne

Definition at line 202 of file master.h.

enum ABA_MASTER::SKIPPINGMODE

The way nodes are skipped for the generation of cuts.

Parameters:

	SkipByNode	Cuts are only generated in every { SkipFactor} subproblem, where { SkipFactor} can be controlled with the parameter file { .abacus}.
	SkipByLevel	Cuts are only generated in every { SkipFactor} level of the enumeration tree.

Enumerator:

SkipByNode
SkipByLevel

Definition at line 218 of file master.h.

enum ABA_MASTER::CONELIMMODE

This enumeration defines the ways for automatic constraint elimination during the cutting plane phase.

Parameters:

	NoConElim	No constraints are eliminated.
	NonBinding	Nonbinding constraints are eliminated.
	Basic	Constraints with basic slack variable are eliminated.

Enumerator:

NoConElim
NonBinding
Basic

Definition at line 233 of file master.h.

enum ABA_MASTER::VARELIMMODE

This enumeration defines the ways for automatic variable elimination during the column generation algorithm.

Parameters:

	NoVarElim	No variables are eliminated.
	ReducedCost	Variables with high absolute reduced costs are eliminated.

Enumerator:

NoVarElim
ReducedCost

Definition at line 249 of file master.h.

enum ABA_MASTER::VBCMODE

This enumeration defines what kind of output can be generated for the VBCTOOL.

Parameters:

	None	No output for the tree interface.
	File	Output for the tree interface is written to a file.
	Pipe	Output for the tree interface is pipe to the standard output.

Enumerator:

NoVbc
File
Pipe

Definition at line 266 of file master.h.

enum ABA_MASTER::OSISOLVER

This enumeration defines which solvers can be used to solve theLP-relaxations.

Enumerator:

Cbc
Clp
CPLEX
DyLP
FortMP
GLPK
MOSEK
OSL
SoPlex
SYMPHONY
Vol
XPRESS_MP

Definition at line 280 of file master.h.

Constructor & Destructor Documentation

ABA_MASTER::ABA_MASTER	(	const char *	problemName,
		bool	cutting,
		bool	pricing,
		ABA_OPTSENSE::SENSE	optSense = `ABA_OPTSENSE::Unknown`,
		double	eps = `1.0e-4`,
		double	machineEps = `1.0e-7`,
		double	infinity = `1.0e30`,
		bool	readParamFromFile = `true`
	)

The constructor.

Parameters:

	problemName	The name of the problem being solved. Must not be a 0-pointer.
	cutting	If true, then cutting planes can be generated if the function ABA_SUB::separate() is redefined.
	pricing	If true, then inactive variables are priced in, if the function ABA_SUB::pricing() is redefined.
	optSense	The sense of the optimization. The default value is ABA_OPTSENSE::Unknown. If the sense is unknown when this constructor is called, e.g., if it is read from a file in the constructor of the derived class, then it must be initialized in the constructor of the derived class.
	eps	The zero-tolerance used within all member functions of objects which have a pointer to this master (default value 1.0e-4).
	machineEps	The machine dependent zero tolerance (default value 1.0e-7).
	infinity	All values greater than infinity are regarded as ``infinite big'', all values less than -infinity are regarded as ``infinite small'' (default value 1.0e30).
	readParamFromFile	If true, then the parameter file .abacus is read, otherwise the parameters are initialized with default values (default true).

The members primalBound_ and dualBound_ stay uninitialized since this can only be done when the sense of optimization (minimization or maximization) is known. The initialization is performed automatically in the function optimize().

virtual ABA_MASTER::~ABA_MASTER ( ) [virtual]

The destructor.

ABA_MASTER::ABA_MASTER ( const ABA_MASTER & rhs ) [private]

Member Function Documentation

STATUS ABA_MASTER::optimize ( )

Performs the optimization by .

The status of the optimization.

double ABA_MASTER::lowerBound ( ) const [inline]

Returns:: The value of the global lower bound.

Definition at line 1885 of file master.h.

double ABA_MASTER::upperBound ( ) const [inline]

Returns:: The value of the global upper bound.

Definition at line 1891 of file master.h.

double ABA_MASTER::primalBound ( ) const [inline]

Returns:: The value of the primal bound, i.e., the lowerBound() for a maximization problem and the upperBound() for a minimization problem, respectively.

Definition at line 1897 of file master.h.

void ABA_MASTER::primalBound ( double x )

This version of the function primalBound() sets the primal bound to x and makes a new entry in the solution history. It is an error if the primal bound gets worse.

Parameters:

The new value of the primal bound.

double ABA_MASTER::dualBound ( ) const [inline]

Returns:: The value of the dual bound, i.e., the upperBound() for a maximization problem and the lowerBound() for a minimization problem, respectively.

Definition at line 1902 of file master.h.

void ABA_MASTER::dualBound ( double x )

This version of the function dualBound() sets the dual bound to x and makes a new entry in the solution history.

It is an error if the dual bound gets worse.

Parameters:

The new value of the dual bound.

bool ABA_MASTER::betterDual ( double x ) const

Returns:: true If x is better than the best known dual bound.
false otherwise.

Parameters:

The value being compared with the best know dual bound.

bool ABA_MASTER::primalViolated ( double x ) const

Can be used to compare a value with the one of the best known primal bound.

If the objective function values of all feasible solutions are integer, then we do not have to be so carefully.

Returns:: true If x is not better than the best known primal bound,
false otherwise.

Parameters:

The value being compared with the primal bound.

bool ABA_MASTER::betterPrimal ( double x ) const

Can be used to check if a value is better than the best know primal bound.

Returns:: true If x is better than the best known primal bound,
false otherwise.

Parameters:

The value compared with the primal bound.

bool ABA_MASTER::feasibleFound ( ) const

We use this function ,e.g., to adapt the enumeration strategy in the DiveAndBest-Strategy.

This function is only correct if any primal bound better than plus/minus infinity corresponds to a feasible solution.

Returns:: true If a feasible solution of the optimization problem has been found.
false otherwise.

ABA_MASTER::ENUMSTRAT ABA_MASTER::enumerationStrategy ( ) const [inline]

Returns:: The enumeration strategy.

Definition at line 2251 of file master.h.

void ABA_MASTER::enumerationStrategy ( ENUMSTRAT strat ) [inline]

This version of the function enumerationStrategy() changes the enumeration strategy.

Parameters:

strat

The new enumeration strategy.

Definition at line 2256 of file master.h.

virtual int ABA_MASTER::enumerationStrategy	(	const ABA_SUB *	s1,
		const ABA_SUB *	s2
	)			`[virtual]`

Analyzes the enumeration strategy set in the parameter file { .abacus} and calls the corresponding comparison function for the subproblems s1 and s2. This function should be redefined for application specific enumeration strategies.

Returns:: 1 If s1 has higher priority than s2
0 if s2 has higher priority it returns , and if both subproblems have equal priority

Parameters:

	s1	A pointer to subproblem.
	s2	A pointer to subproblem.

bool ABA_MASTER::guaranteed ( )

Can be used to check if the guarantee requirements are fulfilled, i.e., the difference between upper bound and the lower bound in respect to the lowerBound is less than this guarantee value in percent.

If the lower bound is zero, but the upper bound is nonzero, we cannot give any guarantee.

Warning:: A guarantee for a solution can only be given if the pricing problem is solved exactly or no column generation is performed at all.

Returns:: true If the guarantee requirements are fulfilled,
false otherwise.

double ABA_MASTER::guarantee ( )

Can be used to access the guarantee which can be given for the best known feasible solution.

It is an error to call this function if the lower bound is zero, but the upper bound is nonzero.

Returns:: The guarantee for best known feasible solution in percent.

void ABA_MASTER::printGuarantee ( )

Writes the guarantee nicely formated on the output stream associated with this object.

If no bounds are available, or the lower bound is zero, but the upper bound is nonzero, then we cannot give any guarantee.

bool ABA_MASTER::check ( )

Can be used to control the correctness of the optimization if the value of the optimum solution has been loaded.

This is done, if a file storing the optimum value is specified with the parameter { OptimumFileName} in the configuration file { .abacus}.

Returns:: true If the optimum solution of the problem is known and equals the primal bound,
false otherwise.

bool ABA_MASTER::knownOptimum ( double & optVal )

Opens the file specified with the parameter { OptimumFileName} in the configuration file { .abacus} and tries to find a line with the name of the problem instance (as specified in the constructor of ABA_MASTER) as first string.

Returns:: true If a line with problemName_ has been found,
false otherwise.

Parameters:

optVal

If the return value is true, then optVal holds the optimum value found in the line with the name of the problem instance as first string. Otherwise, optVal is undefined.

virtual void ABA_MASTER::output ( ) [virtual]

Does nothing but can be redefined in derived classes for output before the timing statistics.

bool ABA_MASTER::cutting ( ) const [inline]

Returns:: true If cutting has been set to true in the call of the constructor of the class ABA_MASTER, i.e., if cutting planes should be generated in the subproblem optimization.
false otherwise.

Definition at line 1952 of file master.h.

bool ABA_MASTER::pricing ( ) const [inline]

Returns:: true If pricing has been set to true in the call of the constructor of the class ABA_MASTER, i.e., if a columns should be generated in the subproblem optimization.
false otherwise.

Definition at line 1957 of file master.h.

const ABA_OPTSENSE * ABA_MASTER::optSense ( ) const [inline]

Returns:: A pointer to the object holding the optimization sense of the problem.

Definition at line 1917 of file master.h.

ABA_HISTORY * ABA_MASTER::history ( ) const [inline]

Returns:: A pointer to the object storing the solution history of this branch and cut problem.

Definition at line 1922 of file master.h.

ABA_OPENSUB * ABA_MASTER::openSub ( ) const [inline]

Returns:: A pointer to the set of open subproblems.

Definition at line 1927 of file master.h.

ABA_STANDARDPOOL< ABA_CONSTRAINT, ABA_VARIABLE > * ABA_MASTER::conPool ( ) const [inline]

Returns:: A pointer to the default pool storing the constraints of the problem formulation.

Definition at line 1937 of file master.h.

ABA_STANDARDPOOL< ABA_CONSTRAINT, ABA_VARIABLE > * ABA_MASTER::cutPool ( ) const [inline]

Returns:: A pointer to the default pool for the generated cutting planes.

Definition at line 1942 of file master.h.

ABA_STANDARDPOOL< ABA_VARIABLE, ABA_CONSTRAINT > * ABA_MASTER::varPool ( ) const [inline]

Returns:: A pointer to the default pool storing the variables.

Definition at line 1947 of file master.h.

ABA_SUB * ABA_MASTER::root ( ) const [inline]

Can be used to access the root node of the \ tree.

Returns:: A pointer to the root node of the enumeration tree.

Definition at line 1907 of file master.h.

ABA_SUB * ABA_MASTER::rRoot ( ) const [inline]

Returns:: A pointer to the root of the remaining \ tree, i.e., the subproblem which is an ancestor of all open subproblems and has highest level in the tree.

Definition at line 1912 of file master.h.

ABA_MASTER::STATUS ABA_MASTER::status ( ) const [inline]

Returns:: The status of the ABA_MASTER.

Definition at line 2141 of file master.h.

const ABA_STRING* ABA_MASTER::problemName ( ) const

Returns:: A pointer to the name of the instance being optimized (as specified in the constructor of this class).

const ABA_COWTIMER * ABA_MASTER::totalCowTime ( ) const [inline]

Returns:: A pointer to the timer measuring the total wall clock time.

Definition at line 1966 of file master.h.

bool ABA_MASTER::solveApprox ( ) const [inline]

True, if an approximative solver should be used

Definition at line 1961 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::totalTime ( ) const [inline]

Returns:: A pointer to the timer measuring the total cpu time for the optimization.

Definition at line 1971 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::lpTime ( ) const [inline]

Returns:: A pointer to the timer measuring the cpu time spent in members of the LP-interface.

Definition at line 1976 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::lpSolverTime ( ) const [inline]

Returns:: A pointer to the timer measuring the cpu time required by the LP solver.

Definition at line 1981 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::separationTime ( ) const [inline]

Returns:: A pointer to the timer measuring the cpu time spent in the separation of cutting planes.

Definition at line 1986 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::improveTime ( ) const [inline]

Returns:: A pointer to the timer measuring the cpu time spent in the heuristics for the computation of feasible solutions.

Definition at line 1991 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::pricingTime ( ) const [inline]

Returns:: A pointer to the timer measuring the cpu time spent in pricing.

Definition at line 1996 of file master.h.

const ABA_CPUTIMER * ABA_MASTER::branchingTime ( ) const [inline]

Returns:: A pointer to the timer measuring the cpu time spent in finding and selecting the branching rules.

Definition at line 2001 of file master.h.

int ABA_MASTER::nSub ( ) const [inline]

Returns:: The number of generated subproblems.

Definition at line 2036 of file master.h.

int ABA_MASTER::nLp ( ) const [inline]

Returns:: The number of optimized linear programs (only LP-relaxations).

Definition at line 2041 of file master.h.

int ABA_MASTER::highestLevel ( ) const [inline]

Returns:: The highest level in the tree which has been reached during the implicit enumeration.

Definition at line 2046 of file master.h.

int ABA_MASTER::nNewRoot ( ) const [inline]

Returns:: The number of root changes of the remaining \ tree.

Definition at line 2051 of file master.h.

int ABA_MASTER::nSubSelected ( ) const [inline]

Returns:: The number of subproblems which have already been selected from the set of open subproblems.

Definition at line 2056 of file master.h.

void ABA_MASTER::printParameters ( )

Writes all parameters of the class ABA_MASTER together with their values to the global output stream.

ABA_MASTER::BRANCHINGSTRAT ABA_MASTER::branchingStrategy ( ) const [inline]

Returns:: The branching strategy.

Definition at line 2261 of file master.h.

void ABA_MASTER::branchingStrategy ( BRANCHINGSTRAT strat ) [inline]

Changes the branching strategy.

Parameters:

strat

The new branching strategy.

Definition at line 2266 of file master.h.

ABA_MASTER::OSISOLVER ABA_MASTER::defaultLpSolver ( ) const [inline]

Returns:: The Lp Solver.

Definition at line 2271 of file master.h.

void ABA_MASTER::defaultLpSolver ( OSISOLVER osiSolver ) [inline]

Changes the default Lp solver.

Parameters:

osiSolver

The new solver.

Definition at line 2276 of file master.h.

ABA_LPMASTEROSI* ABA_MASTER::lpMasterOsi ( ) const [inline]

Definition at line 739 of file master.h.

int ABA_MASTER::nBranchingVariableCandidates ( ) const [inline]

Returns:: The number of variables that should be tested for the selection of the branching variable.

Definition at line 2281 of file master.h.

void ABA_MASTER::nBranchingVariableCandidates ( int n )

This version of the function nbranchingVariableCandidates() sets the number of tested branching variable candidates.

Parameters:

The new value of the number of tested variables for becoming branching variable.

double ABA_MASTER::requiredGuarantee ( ) const [inline]

The guarantee specification for the optimization.

Definition at line 2286 of file master.h.

void ABA_MASTER::requiredGuarantee ( double g )

This version of the function requiredGuarantee() changes the guarantee specification.

Parameters:

The new guarantee specification (in percent). This must be a nonnative value. Note, if the guarantee specification is changed after a single node of the enumeration tree has been fathomed, then the overall guarantee might differ from the new value.

int ABA_MASTER::maxLevel ( ) const [inline]

Returns:: The maximal depth up to which the enumeration should be performed. By default the maximal enumeration depth is INT .

Definition at line 2291 of file master.h.

void ABA_MASTER::maxLevel ( int ml )

This version of the function maxLevel() changes the maximal enumeration depth.

If it is set to 1 the \ algorithm becomes a pure cutting plane algorithm.

Parameters:

max

The new value of the maximal enumeration level.

const ABA_STRING & ABA_MASTER::maxCpuTime ( ) const [inline]

Returns:: The maximal cpu time which can be used by the optimization.

Definition at line 2296 of file master.h.

void ABA_MASTER::maxCpuTime ( const ABA_STRING & t ) [inline]

Sets the maximal usable cpu time for the optimization.

Parameters:

The new value of the maximal cpu time in the form { "hh:mm:ss"}.

Definition at line 2301 of file master.h.

const ABA_STRING & ABA_MASTER::maxCowTime ( ) const [inline]

The function maxCowTime().

Returns:: The maximal wall-clock time for the optimization.

Definition at line 2306 of file master.h.

void ABA_MASTER::maxCowTime ( const ABA_STRING & t ) [inline]

This version of the function maxCowTime() set the maximal wall-clock time for the optimization.

Parameters:

The new value of the maximal wall-clock time in the form { hh:mm:ss}.

Definition at line 2311 of file master.h.

bool ABA_MASTER::objInteger ( ) const [inline]

Returns:: true Then we assume that all feasible solutions have integral objective function values,
false otherwise.

Definition at line 2316 of file master.h.

void ABA_MASTER::objInteger ( bool b ) [inline]

This version of function objInteger() sets the assumption that the objective function values of all feasible solutions are integer.

Parameters:

The new value of the assumption.

Definition at line 2321 of file master.h.

int ABA_MASTER::tailOffNLp ( ) const [inline]

The function tailOffNLp().

Returns:: The number of linear programs considered in the tailing off analysis.

Definition at line 2326 of file master.h.

void ABA_MASTER::tailOffNLp ( int n ) [inline]

Sets the number of linear programs considered in the tailing off analysis.

This new value is only relevant for subproblems activated { after} the change of this value.

Parameters:

The new number of LPs for the tailing off analysis.

Definition at line 2331 of file master.h.

double ABA_MASTER::tailOffPercent ( ) const [inline]

The function tailOffPercent().

Returns:: The minimal change of the dual bound for the tailing off analysis in percent.

Definition at line 2336 of file master.h.

void ABA_MASTER::tailOffPercent ( double p )

This version of the function tailOffPercent() sets the minimal change of the dual bound for the tailing off analysis.

This change is only relevant for subproblems activated { after} calling this function.

Parameters:

The new value for the tailing off analysis.

ABA_MASTER::OUTLEVEL ABA_MASTER::outLevel ( ) const [inline]

Returns:: The output mode.

Definition at line 2341 of file master.h.

void ABA_MASTER::outLevel ( OUTLEVEL mode ) [inline]

The version of the function outLevel() sets the output mode.

Parameters:

mode

The new value of the output mode.

Definition at line 2346 of file master.h.

ABA_MASTER::OUTLEVEL ABA_MASTER::logLevel ( ) const [inline]

Returns:: The output mode for the log-file.

Definition at line 2351 of file master.h.

void ABA_MASTER::logLevel ( OUTLEVEL mode ) [inline]

This version of the function logLevel() sets the output mode for the log-file.

Parameters:

mode

The new value of the output mode.

Definition at line 2356 of file master.h.

bool ABA_MASTER::delayedBranching ( int nOpt_ ) const

Returns:: true If the number of optimizations nOpt of a subproblem exceeds the delayed branching threshold,
false otherwise.

Parameters:

nOpt

The number of optimizations of a subproblem.

void ABA_MASTER::dbThreshold ( int threshold ) [inline]

Sets the number of optimizations of a subproblem until sons are created in ABA_SUB::branching().

If this value is 0, then a branching step is performed at the end of the subproblem optimization as usually if the subproblem can be fathomed. Otherwise, if this value is strictly positive, the subproblem is put back for a later optimization. This can be advantageous if in the meantime good cutting planes or primal bounds can be generated. The number of times the subproblem is put back without branching is indicated by this value.

Parameters:

threshold

The new value of the delayed branching threshold.

Definition at line 2361 of file master.h.

int ABA_MASTER::dbThreshold ( ) const [inline]

Returns:: The number of optimizations of a subproblem until sons are created. For further detatails we refer to dbThreshold(int).

Definition at line 2366 of file master.h.

int ABA_MASTER::minDormantRounds ( ) const [inline]

Returns:: The maximal number of rounds, i.e., number of subproblem optimizations, a subproblem is dormant, i.e., it is not selected from the set of open subproblem if its status is Dormant, if possible.

Definition at line 2371 of file master.h.

void ABA_MASTER::minDormantRounds ( int nRounds ) [inline]

Sets the number of rounds a subproblem should stay dormant.

Parameters:

nRounds

The new minimal number of dormant rounds.

Definition at line 2376 of file master.h.

ABA_MASTER::PRIMALBOUNDMODE ABA_MASTER::pbMode ( ) const [inline]

Returns:: The mode of the primal bound initialization.

Definition at line 2381 of file master.h.

void ABA_MASTER::pbMode ( PRIMALBOUNDMODE mode ) [inline]

Sets the mode of the primal bound initialization.

Parameters:

mode

The new mode of the primal bound initialization.

Definition at line 2386 of file master.h.

int ABA_MASTER::pricingFreq ( ) const [inline]

Returns:: The number of linear programs being solved between two additional pricing steps. If no additional pricing steps should be executed this parameter has to be set to 0. The default value of the pricing frequency is 0. This parameter does not influence the execution of pricing steps which are required for the correctness of the algorithm.

Definition at line 2391 of file master.h.

void ABA_MASTER::pricingFreq ( int f )

This version of the function pricingFreq() sets the number of linear programs being solved between two additional pricing steps.

Parameters:

The pricing frequency.

int ABA_MASTER::skipFactor ( ) const [inline]

Returns:: The frequency of subproblems in which constraints or variables should be generated.

Definition at line 2396 of file master.h.

void ABA_MASTER::skipFactor ( int f )

This version of the function skipFactor() sets the frequency for constraint and variable generation.

Parameters:

The new value of the frequency.

void ABA_MASTER::skippingMode ( SKIPPINGMODE mode ) [inline]

This version of the function skippingMode() sets the skipping strategy.

Parameters:

mode

The new skipping strategy.

Definition at line 2406 of file master.h.

ABA_MASTER::SKIPPINGMODE ABA_MASTER::skippingMode ( ) const [inline]

Returns:: The skipping strategy.

Definition at line 2401 of file master.h.

ABA_MASTER::CONELIMMODE ABA_MASTER::conElimMode ( ) const [inline]

Returns:: The mode for the elimination of constraints.

Definition at line 2191 of file master.h.

void ABA_MASTER::conElimMode ( CONELIMMODE mode ) [inline]

Changes the constraint elimination mode.

Parameters:

mode

The new constraint elimination mode.

Definition at line 2196 of file master.h.

ABA_MASTER::VARELIMMODE ABA_MASTER::varElimMode ( ) const [inline]

Returns:: The mode for the elimination of variables.

Definition at line 2201 of file master.h.

void ABA_MASTER::varElimMode ( VARELIMMODE mode ) [inline]

Changes the variable elimination mode.

Parameters:

mode

The new variable elimination mode.

Definition at line 2206 of file master.h.

double ABA_MASTER::conElimEps ( ) const [inline]

Returns:: The zero tolerance for the elimination of constraints by the slack criterion.

Definition at line 2211 of file master.h.

void ABA_MASTER::conElimEps ( double eps ) [inline]

Changes the tolerance for the elimination of constraints by the slack criterion.

Parameters:

eps

The new tolerance.

Definition at line 2216 of file master.h.

double ABA_MASTER::varElimEps ( ) const [inline]

Returns:: The zero tolerance for the elimination of variables by the reduced cost criterion.

Definition at line 2221 of file master.h.

void ABA_MASTER::varElimEps ( double eps ) [inline]

Changes the tolerance for the elimination of variables by the reduced cost criterion.

Parameters:

eps

The new tolerance.

Definition at line 2226 of file master.h.

int ABA_MASTER::varElimAge ( ) const [inline]

Returns:: The age for the elimination of variables by the reduced cost criterion.

Definition at line 2231 of file master.h.

void ABA_MASTER::varElimAge ( int eps ) [inline]

Changes the age for the elimination of variables by the reduced cost criterion.

Parameters:

eps

The new age.

Definition at line 2236 of file master.h.

int ABA_MASTER::conElimAge ( ) const [inline]

Returns:: The age for the elimination of constraints.

Definition at line 2241 of file master.h.

void ABA_MASTER::conElimAge ( int eps ) [inline]

Changes the age for the elimination of constraints.

Parameters:

eps

The new age.

Definition at line 2246 of file master.h.

bool ABA_MASTER::fixSetByRedCost ( ) const [inline]

Returns:: true Then variables are fixed and set by reduced cost criteria.
false Then no variables are fixed or set by reduced cost criteria.

Definition at line 2061 of file master.h.

void ABA_MASTER::fixSetByRedCost ( bool on ) [inline]

Turns fixing and setting variables by reduced cost on or off.

Parameters:

If true, then variable fixing and setting by reduced cost is turned on. Otherwise it is turned of.

Definition at line 2066 of file master.h.

bool ABA_MASTER::printLP ( ) const [inline]

Returns:: true Then the linear program is output every iteration of the subproblem optimization.
false The linear program is not output.

Definition at line 2071 of file master.h.

void ABA_MASTER::printLP ( bool on ) [inline]

Turns the output of the linear program in every iteration on or off.

Parameters:

If true, then the linear program is output, otherwise it is not output.

Definition at line 2076 of file master.h.

int ABA_MASTER::maxConAdd ( ) const [inline]

Returns:: The maximal number of constraints which should be added in every iteration of the cutting plane algorithm.

Definition at line 2081 of file master.h.

void ABA_MASTER::maxConAdd ( int max ) [inline]

Sets the maximal number of constraints that are added in an iteration of the cutting plane algorithm.

Parameters:

max

The maximal number of constraints.

Definition at line 2086 of file master.h.

int ABA_MASTER::maxConBuffered ( ) const [inline]

Returns:: The size of the buffer for generated constraints in the cutting plane algorithm.

Definition at line 2091 of file master.h.

void ABA_MASTER::maxConBuffered ( int max ) [inline]

Changes the maximal number of constraints that are buffered in an iteration of the cutting plane algorithm.

Note:: This function changes only the default value for subproblems that are activated after its call.

Parameters:

max

The new maximal number of buffered constraints.

Definition at line 2096 of file master.h.

int ABA_MASTER::maxVarAdd ( ) const [inline]

Returns:: The maximal number of variables which should be added in the column generation algorithm.

Definition at line 2101 of file master.h.

void ABA_MASTER::maxVarAdd ( int max ) [inline]

Changes the maximal number of variables that are added in an iteration of the subproblem optimization.

Parameters:

max

The new maximal number of added variables.

Definition at line 2106 of file master.h.

int ABA_MASTER::maxVarBuffered ( ) const [inline]

Returns:: The size of the buffer for the variables generated in the column generation algorithm.

Definition at line 2111 of file master.h.

void ABA_MASTER::maxVarBuffered ( int max ) [inline]

Changes the maximal number of variables that are buffered in an iteration of the subproblem optimization.

Note:: This function changes only the default value for subproblems that are activated after its call.

Parameters:

max

The new maximal number of buffered variables.

Definition at line 2116 of file master.h.

int ABA_MASTER::maxIterations ( ) const [inline]

Returns:: The maximal number of iterations per subproblem optimization (-1 means no iteration limit).

Definition at line 2121 of file master.h.

void ABA_MASTER::maxIterations ( int max ) [inline]

Changes the default value for the maximal number of iterations of the optimization of a subproblem.

Note:: This function changes only this value for subproblems that are constructed after this function call. For already constructed objects the value can be changed with the function ABA_SUB::maxIterations().

Parameters:

max

The new maximal number of iterations of the subproblem optimization (-1 means no limit).

Definition at line 2126 of file master.h.

bool ABA_MASTER::eliminateFixedSet ( ) const [inline]

Returns:: true Then we try to eliminate fixed and set variables from the linear program.
false Fixed or set variables are not eliminated.

Definition at line 2151 of file master.h.

void ABA_MASTER::eliminateFixedSet ( bool turnOn ) [inline]

This version of the function eliminateFixedSet() can be used to turn the elimination of fixed and set variables on or off.

Parameters:

turnOn

The elimination is turned on if turnOn is true, otherwise it is turned off.

Definition at line 2156 of file master.h.

bool ABA_MASTER::newRootReOptimize ( ) const [inline]

Returns:: true Then a new root of the remaining \ tree is reoptimized such that the associated reduced costs can be used for the fixing of variables.
false A new root is not reoptimized.

Definition at line 2161 of file master.h.

void ABA_MASTER::newRootReOptimize ( bool on ) [inline]

Turns the reoptimization of new root nodes of the remaining branch and bound tree on or off.

Parameters:

If true, new root nodes are reoptimized.

Definition at line 2166 of file master.h.

const ABA_STRING & ABA_MASTER::optimumFileName ( ) const [inline]

Returns:: The name of the file that stores the optimum solutions.

Definition at line 2131 of file master.h.

void ABA_MASTER::optimumFileName ( const char * name ) [inline]

Changes the name of the file in which the value of the optimum solution is searched.

Parameters:

name

The new name of the file.

Definition at line 2136 of file master.h.

bool ABA_MASTER::showAverageCutDistance ( ) const [inline]

Returns:: true Then the average distance of the fractional solution from all added cutting planes is output every iteration of the subproblem optimization.
false The average cut distance is not output.

Definition at line 2171 of file master.h.

void ABA_MASTER::showAverageCutDistance ( bool on ) [inline]

Turns the output of the average distance of the added cuts from the fractional solution on or off.

Parameters:

If true the output is turned on, otherwise it is turned off.

Definition at line 2176 of file master.h.

ABA_MASTER::VBCMODE ABA_MASTER::vbcLog ( ) const [inline]

Returns:: The mode of output for the Vbc-Tool.

Definition at line 2181 of file master.h.

void ABA_MASTER::vbcLog ( VBCMODE mode ) [inline]

Changes the mode of output for the Vbc-Tool.

This function should only be called before the optimization is started with the function ABA_MASTER::optimize().

Parameters:

mode

The new mode.

Definition at line 2186 of file master.h.

virtual bool ABA_MASTER::setSolverParameters	(	OsiSolverInterface *	interface,
		bool	solverIsApprox
	)			`[virtual]`

Set solver specific parameters. The default does nothing.

Returns:: true if an error has occured otherwise

virtual void ABA_MASTER::initializePools	(	ABA_BUFFER< ABA_CONSTRAINT * > &	constraints,
		ABA_BUFFER< ABA_VARIABLE * > &	Variables,
		int	varPoolSize,
		int	cutPoolSize,
		bool	dynamicCutPool = `false`
	)			`[protected, virtual]`

Sets up the default pools for variables, constraints, and cutting planes.

Parameters:

	constraints	The constraints of the problem formulation are inserted in the constraint pool. The size of the constraint pool equals the number of constraints.
	variables	The variables of the problem formulation are inserted in the variable pool.
	varPoolSize	The size of the pool for the variables. If more variables are added the variable pool is automatically reallocated.
	cutPoolSize	The size of the pool for cutting planes.
	dynamicCutPool	If this argument is true, then the cut is automatically reallocated if more constraints are inserted than cutPoolSize. Otherwise, non-active constraints are removed if the pool becomes full. The default value is false.

virtual void ABA_MASTER::initializePools	(	ABA_BUFFER< ABA_CONSTRAINT * > &	constraints,
		ABA_BUFFER< ABA_CONSTRAINT * > &	cuts,
		ABA_BUFFER< ABA_VARIABLE * > &	Variables,
		int	varPoolSize,
		int	cutPoolSize,
		bool	dynamicCutPool = `false`
	)			`[protected, virtual]`

Is overloaded such that also a first set of cutting planes can be inserted into the cutting plane pool.

Parameters:

	constraints	The constraints of the problem formulation are inserted in the constraint pool. The size of the constraint pool equals the number of constraints.
	cuts	The constraints that are inserted in the cutting plane pool. The number of constraints in the buffer must be less or equal than the size of the cutting plane pool cutPoolSize.
	variables	The variables of the problem formulation are inserted in the variable pool.
	varPoolSize	The size of the pool for the variables. If more variables are added the variable pool is automatically reallocated.
	cutPoolSize	The size of the pool for cutting planes.
	dynamicCutPool	If this argument is true, then the cut is automatically reallocated if more constraints are inserted than cutPoolSize. Otherwise, non-active constraints are removed if the pool becomes full. The default value is false.

void ABA_MASTER::initializeOptSense ( ABA_OPTSENSE::SENSE sense ) [protected]

Can be used to initialize the sense of the optimization in derived classes, if this has not been already performed when the constructor of ABA_MASTER has been called.

Parameters:

sense

The sense of the optimization (ABA_OPTSENSE::Min or ABA_OPTSENSE::Max).

int ABA_MASTER::bestFirstSearch	(	const ABA_SUB *	s1,
		const ABA_SUB *	s2
	)			const `[protected]`

Implements the best first search enumeration.

If the bounds of both subproblems are equal, then the subproblems are compared with the function equalSubCompare().

Returns:

-1 If subproblem s1 has a worse dual bound than s2, i.e., if it has a smaller dual bound for minimization or a larger dual bound for maximization problems.

1 If subproblem s2 has a worse dual bound than s1.

0 If both subproblems have the same priority in the enumeration strategy.

Parameters:

	s1	A subproblem.
	s2	A subproblem.

virtual int ABA_MASTER::equalSubCompare	(	const ABA_SUB *	s1,
		const ABA_SUB *	s2
	)			const `[protected, virtual]`

Is called from the function bestFirstSearch() and from the function depthFirstSearch() if the subproblems s1 and s2 have the same priority.

If both subproblems were generated by setting a binary variable, then that subproblem has higher priority of which the branching variable is set to upper bound.

This function can be redefined to resolve equal subproblems according to problem specific criteria. As the root node is compared with itself and has no branching rule, we have to insert the first line of this function.

Parameters:

	s1	A subproblem.
	s2	A subproblem.

Returns:

0 If both subproblems were not generated by setting a variable, or the branching variable of both subproblems is set to the same bound.

1 If the branching variable of the first subproblem ist set to the upper bound.

-1 If the branching variable of the second subproblem ist set to the upper bound.

int ABA_MASTER::depthFirstSearch	(	const ABA_SUB *	s1,
		const ABA_SUB *	s2
	)			const `[protected]`

Implements the depth first search enumeration strategy, i.e., the subproblem with maximum level is selected.

If the level of both subproblems are equal, then the subproblems are compared with the function equalSubCompare().

Returns:

-1 If subproblem s1 has higher priority,

0 if both subproblems have equal priority,

1 otherwise.

Parameters:

	s1	The first subproblem.
	s2	The second subproblem.

int ABA_MASTER::breadthFirstSearch	(	const ABA_SUB *	s1,
		const ABA_SUB *	s2
	)			const `[protected]`

Implements the breadth first search enumeration strategy, i.e., the subproblem with minimum level is selected.

If both subproblems have the same level, the smaller one is the one which has been generated earlier, i.e., the one with the smaller id.

Returns:

-1 If subproblem s1 has higher priority,

0 if both subproblems have equal priority,

1 otherwise.

Parameters:

	s1	The first subproblem.
	s2	The second subproblem.

int ABA_MASTER::diveAndBestFirstSearch	(	const ABA_SUB *	s1,
		const ABA_SUB *	s2
	)			const `[protected]`

Performs depth-first search until a feasible solution is found, then the search process is continued with best-first search.

Returns:

-1 If subproblem s1 has higher priority,

0 if both subproblems have equal priority,

1 otherwise.

Parameters:

	s1	The first subproblem.
	s2	The second subproblem.

virtual void ABA_MASTER::initializeParameters ( ) [protected, virtual]

Is only a dummy. This function can be used to initialize parameters of derived classes and to overwrite parameters read from the file { .abacus} by the function ().

virtual ABA_SUB* ABA_MASTER::firstSub ( ) [protected, pure virtual]

Returns:: Should return a pointer to the first subproblem of the optimization, i.e., the root node of the enumeration tree. This is a pure virtual function since a pointer to a problem specific subproblem should be returned, which is derived from the class ABA_SUB.

virtual void ABA_MASTER::initializeOptimization ( ) [protected, virtual]

The default implementation of initializeOptimization() does nothing.

This virtual function can be used as an entrance point to perform some initializations after optimize() is called.

virtual void ABA_MASTER::terminateOptimization ( ) [protected, virtual]

The default implementation of terminateOptimization() does nothing.

This virtual function can be used as an entrance point after the optimization process is finished.

void ABA_MASTER::_initializeParameters ( ) [private]

Reads the parameter-file { .abacus}, which is searched in the directory given by the environment variable ABACUS_DIR, and calls the virtual function initializeParameters() which can initialize parameters of derived classes and overwrite parameters of this class.

All parameters are first inserted together with their values in a parameter table in the function readParameters(). If the virtual dummy function initializeParameters() is redefined in a derived class and also reads a parameter file with the function readParameters(), then already inserted parameters can be overwritten.

: After all parameters are input we extract with the function assignParameter() all parameters. Problem specific parameters should be extracted in a redefined version of initializeParameters(). extracted from this table

void ABA_MASTER::_createLpMasters ( ) [private]

void ABA_MASTER::_deleteLpMasters ( ) [private]

void ABA_MASTER::_initializeLpParameters ( ) [private]

void ABA_MASTER::_setDefaultLpParameters ( ) [private]

Initializes the LP solver specific default Parameters if they are not read from the parameter-file { .abacus}.

This function is implemented in the file lpif.cc.

void ABA_MASTER::_printLpParameters ( ) [private]

Prints the LP solver specific parameters.

This function is implemented in the file lpif.cc.

void ABA_MASTER::_outputLpStatistics ( ) [private]

Prints the LP solver specific statistics.

This function is implemented in the file lpif.cc.

ABA_SUB* ABA_MASTER::select ( ) [private]

Returns a pointer to an open subproblem for further processing.

If the set of open subproblems is empty or one of the criteria for early termination of the optimization (maximal cpu time, maximal elapsed time, guarantee) is fulfilled 0 is returned.

int ABA_MASTER::initLP ( ) [private]

void ABA_MASTER::writeTreeInterface	(	const char *	info,
		bool	time = `true`
	)			const `[private]`

Writes the string info to the stream associated with the Tree Interface.

A $ is preceded if the output is written to standard out for further pipelining. If time is true a time string is written in front of the information. The default value of time is true.

void ABA_MASTER::treeInterfaceNewNode ( ABA_SUB * sub ) const [private]

Adds the subproblem sub to the stream storing information for graphical output of the enumeration tree if this logging is turned on.

void ABA_MASTER::treeInterfacePaintNode	(	int	id,
		int	color
	)			const `[private]`

Assigns the color to the subproblem sub in the Tree Interface.

void ABA_MASTER::treeInterfaceLowerBound ( double lb ) const [private]

Passes the new lower bound lb to the Tree Interface.

void ABA_MASTER::treeInterfaceUpperBound ( double ub ) const [private]

Passes the new upper bound ub to the Tree Interface.

void ABA_MASTER::treeInterfaceNodeBounds	(	int	id,
		double	lb,
		double	ub
	)			`[private]`

Updates the node information in the node with number id by writing the lower bound lb and the upper bound ub to the node.

void ABA_MASTER::newSub ( int level ) [private]

Registers a new subproblem which is on level level in enumeration tree.

It is called each time a new subproblem is generated.

void ABA_MASTER::countLp ( ) [inline, private]

Increments the counter for linear programs and should be called in each optimization call of the LP-relaxation.

Definition at line 2006 of file master.h.

void ABA_MASTER::newFixed ( int n ) [inline, private]

Increments the counter of the number of fixed variables by n.

Definition at line 2011 of file master.h.

void ABA_MASTER::addCons ( int n ) [inline, private]

Increments the counter for the total number of added constraints by n.

Definition at line 2016 of file master.h.

void ABA_MASTER::removeCons ( int n ) [inline, private]

Increments the counter for the total number of removed constraints by n.

Definition at line 2021 of file master.h.

void ABA_MASTER::addVars ( int n ) [inline, private]

Increments the counter for the total number of added variables by n.

Definition at line 2026 of file master.h.

void ABA_MASTER::removeVars ( int n ) [inline, private]

Increments the counter for the total number of removed variables by n.

Definition at line 2031 of file master.h.

ABA_FIXCAND * ABA_MASTER::fixCand ( ) const [inline, private]

returns a pointer to the object storing the variables which are candidates for being fixed.

Definition at line 1932 of file master.h.

void ABA_MASTER::rRoot	(	ABA_SUB *	newRoot,
		bool	reoptimize
	)			`[private]`

Sets the root of the remaining \ tree to newRoot.

If reoptimize is true a reoptimization of the subproblem *newRoot is performed. This is controlled via a function argument since it might not be desirable when we find a new rRoot_ during the fathoming of a complete subtree ABA_SUB::FathomTheSubtree().

void ABA_MASTER::status ( STATUS stat ) [inline, private]

This version of the function status() sets the status of the ABA_MASTER.

Definition at line 2146 of file master.h.

void ABA_MASTER::rootDualBound ( double x ) [private]

Updates the final dual bound of the root node.

This function should be only called at the end of the root node optimization.

void ABA_MASTER::theFuture ( ) [private]

const ABA_MASTER& ABA_MASTER::operator= ( const ABA_MASTER & rhs ) [private]

Friends And Related Function Documentation

friend class ABA_SUB [friend]

Definition at line 77 of file master.h.

friend class ABA_FIXCAND [friend]

Definition at line 78 of file master.h.

Member Data Documentation

const char* ABA_MASTER::STATUS_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { STATUS[0]=="Optimal"}).

Definition at line 117 of file master.h.

const char* ABA_MASTER::OUTLEVEL_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { OUTLEVEL[0]=="Silent"}).

Definition at line 138 of file master.h.

const char* ABA_MASTER::ENUMSTRAT_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { ENUMSTRAT[0]=="BestFirst"}).

Definition at line 163 of file master.h.

const char* ABA_MASTER::BRANCHINGSTRAT_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { BRANCHINGSTRAT[0]=="CloseHalf"}).

Definition at line 181 of file master.h.

const char* ABA_MASTER::PRIMALBOUNDMODE_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { PRIMALBOUNDMODE[0]=="None"}).

Definition at line 208 of file master.h.

const char* ABA_MASTER::SKIPPINGMODE_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { SKIPPINGMODE[0]=="None"}).

Definition at line 224 of file master.h.

const char* ABA_MASTER::CONELIMMODE_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { CONELIMMODE[0]=="None"}).

Definition at line 240 of file master.h.

const char* ABA_MASTER::VARELIMMODE_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { VARELIMMODE[0]=="None"}).

Definition at line 255 of file master.h.

const char* ABA_MASTER::VBCMODE_[] [static]

Literal values for the enumerators of the corresponding enumeration type. The order of the enumerators is preserved. (e.g., { VBCMODE[0]=="None"}).

Definition at line 272 of file master.h.

const char* ABA_MASTER::OSISOLVER_[] [static]

Array for the literal values for possible Osi solvers.

Definition at line 284 of file master.h.

ABA_STRING ABA_MASTER::problemName_ [private]

The name of the optimized problem.

Definition at line 1563 of file master.h.

bool ABA_MASTER::readParamFromFile_ [private]

Definition at line 1564 of file master.h.

ABA_OPTSENSE ABA_MASTER::optSense_ [private]

The sense of the objective function.

Definition at line 1568 of file master.h.

ABA_SUB* ABA_MASTER::root_ [private]

The root node of the enumeration tree.

Definition at line 1572 of file master.h.

ABA_SUB* ABA_MASTER::rRoot_ [private]

The root node of the remaining enumeration tree.

Definition at line 1576 of file master.h.

ABA_OPENSUB* ABA_MASTER::openSub_ [private]

The set of open subproblems.

Definition at line 1580 of file master.h.

ABA_HISTORY* ABA_MASTER::history_ [private]

The solution history.

Definition at line 1584 of file master.h.

ENUMSTRAT ABA_MASTER::enumerationStrategy_ [private]

The enumeration strategy.

Definition at line 1588 of file master.h.

BRANCHINGSTRAT ABA_MASTER::branchingStrategy_ [private]

The branching strategy.

Definition at line 1592 of file master.h.

int ABA_MASTER::nBranchingVariableCandidates_ [private]

The number of candidates that are evaluated for branching on variables.

Definition at line 1597 of file master.h.

OSISOLVER ABA_MASTER::defaultLpSolver_ [private]

The default LP-Solver.

Definition at line 1601 of file master.h.

ABA_LPMASTEROSI* ABA_MASTER::lpMasterOsi_ [private]

Definition at line 1603 of file master.h.

ABA_STANDARDPOOL<ABA_CONSTRAINT, ABA_VARIABLE>* ABA_MASTER::conPool_ [private]

The default pool with the constraints of the problem formulation.

Definition at line 1607 of file master.h.

ABA_STANDARDPOOL<ABA_CONSTRAINT, ABA_VARIABLE>* ABA_MASTER::cutPool_ [private]

The default pool of dynamically generated constraints.

Definition at line 1612 of file master.h.

ABA_STANDARDPOOL<ABA_VARIABLE, ABA_CONSTRAINT>* ABA_MASTER::varPool_ [private]

The default pool with the variables of the problem formulation.

Definition at line 1616 of file master.h.

double ABA_MASTER::primalBound_ [private]

The best known primal bound.

Definition at line 1620 of file master.h.

double ABA_MASTER::dualBound_ [private]

The best known dual bound.

Definition at line 1624 of file master.h.

double ABA_MASTER::rootDualBound_ [private]

The best known dual bound at the end of the optimization of the root node.

Definition at line 1628 of file master.h.

ABA_FIXCAND* ABA_MASTER::fixCand_ [private]

The variables which are candidates for being fixed.

Definition at line 1632 of file master.h.

bool ABA_MASTER::cutting_ [private]

If true, then constraints are generated in the optimization.

Definition at line 1636 of file master.h.

bool ABA_MASTER::pricing_ [private]

If true, then variables are generated in the optimization.

Definition at line 1640 of file master.h.

bool ABA_MASTER::solveApprox_ [private]

If true, then an approximative solver is used to solve linear programs

Definition at line 1645 of file master.h.

int ABA_MASTER::nSubSelected_ [private]

The number of subproblems already selected from the list of open subproblems.

Definition at line 1650 of file master.h.

VBCMODE ABA_MASTER::VbcLog_ [private]

Ouput for the Tree Interface is generated depending on the value of this variable.

Definition at line 1655 of file master.h.

ostream* ABA_MASTER::treeStream_ [private]

A pointer to the log stream for the VBC-Tool.

Definition at line 1659 of file master.h.

double ABA_MASTER::requiredGuarantee_ [private]

The guarantee in percent which should be reached when the optimization stops.

If this value is $0.0$ , then the optimum solution is determined.

Definition at line 1666 of file master.h.

int ABA_MASTER::maxLevel_ [private]

The maximal level in enumeration tree.

Up to this level subproblems are considered in the enumeration.

Definition at line 1672 of file master.h.

ABA_STRING ABA_MASTER::maxCpuTime_ [private]

The maximal available cpu time.

Definition at line 1676 of file master.h.

ABA_STRING ABA_MASTER::maxCowTime_ [private]

The maximal available wall-clock time.

Definition at line 1680 of file master.h.

bool ABA_MASTER::objInteger_ [private]

true, if all objective function values of feasible solutions are assumed to be integer.

Definition at line 1685 of file master.h.

int ABA_MASTER::tailOffNLp_ [private]

The number of LP-iterations for the tailing off analysis.

Definition at line 1689 of file master.h.

double ABA_MASTER::tailOffPercent_ [private]

The minimal change of the LP-value on the tailing off analysis.

Definition at line 1693 of file master.h.

int ABA_MASTER::dbThreshold_ [private]

The number of optimizations of an ABA_SUB until branching is performed.

Definition at line 1697 of file master.h.

int ABA_MASTER::minDormantRounds_ [private]

The minimal number of rounds, i.e., number of subproblem optimizations, a subproblem is dormant, i.e., it is not selected from the set of open subproblem if its status is Dormant, if possible.

Definition at line 1703 of file master.h.

OUTLEVEL ABA_MASTER::outLevel_ [private]

The output mode.

Definition at line 1707 of file master.h.

OUTLEVEL ABA_MASTER::logLevel_ [private]

The amount of output written to the log file.

Definition at line 1711 of file master.h.

PRIMALBOUNDMODE ABA_MASTER::pbMode_ [private]

The mode of the primal bound initialization.

Definition at line 1715 of file master.h.

int ABA_MASTER::pricingFreq_ [private]

The number of solved LPs between two additional pricing steps.

Definition at line 1719 of file master.h.

int ABA_MASTER::skipFactor_ [private]

The frequency constraints or variables are generated depending on the skipping mode.

Definition at line 1724 of file master.h.

SKIPPINGMODE ABA_MASTER::skippingMode_ [private]

Either constraints are generated only every skipFactor_ subproblem (SkipByNode) only every skipFactor_ level (SkipByLevel).

Definition at line 1729 of file master.h.

bool ABA_MASTER::fixSetByRedCost_ [private]

If true, then variables are fixed and set by reduced cost criteria.

Definition at line 1733 of file master.h.

bool ABA_MASTER::printLP_ [private]

If true, then the linear program is output every iteration.

Definition at line 1737 of file master.h.

int ABA_MASTER::maxConAdd_ [private]

The maximal number of added constraints per iteration of the cutting plane algorithm.

Definition at line 1742 of file master.h.

int ABA_MASTER::maxConBuffered_ [private]

The size of the buffer for generated cutting planes.

Definition at line 1746 of file master.h.

int ABA_MASTER::maxVarAdd_ [private]

The maximal number of added variables per iteration of the column generation algorithm.

Definition at line 1751 of file master.h.

int ABA_MASTER::maxVarBuffered_ [private]

The size of the buffer for generated variables.

Definition at line 1755 of file master.h.

int ABA_MASTER::maxIterations_ [private]

The maximal number of iterations of the cutting plane/column generation algorithm in the subproblem.

Definition at line 1760 of file master.h.

bool ABA_MASTER::eliminateFixedSet_ [private]

If true, then nonbasic fixed and set variables are eliminated.

Definition at line 1764 of file master.h.

bool ABA_MASTER::newRootReOptimize_ [private]

If true, then an already earlier processed node is reoptimized if it becomes the new root of the remaining \ tree.

Definition at line 1769 of file master.h.

ABA_STRING ABA_MASTER::optimumFileName_ [private]

The name of a file storing a list of optimum solutions of problem instances.

Definition at line 1774 of file master.h.

bool ABA_MASTER::showAverageCutDistance_ [private]

If true then the average distance of the added cutting planes is output every iteration of the cutting plane algorithm.

Definition at line 1779 of file master.h.

CONELIMMODE ABA_MASTER::conElimMode_ [private]

The way constraints are automatically eliminated in the cutting plane algorithm.

Definition at line 1784 of file master.h.

VARELIMMODE ABA_MASTER::varElimMode_ [private]

The way variables are automatically eliminated in the column generation algorithm.

Definition at line 1789 of file master.h.

double ABA_MASTER::conElimEps_ [private]

The tolerance for the elimination of constraints by the mode NonBinding/.

Definition at line 1794 of file master.h.

double ABA_MASTER::varElimEps_ [private]

The tolerance for the elimination of variables by the mode ReducedCost.

Definition at line 1799 of file master.h.

int ABA_MASTER::conElimAge_ [private]

The number of iterations an elimination criterion must be satisfied until a constraint can be removed.

Definition at line 1804 of file master.h.

int ABA_MASTER::varElimAge_ [private]

The number of iterations an elimination criterion must be satisfied until a variable can be removed.

Definition at line 1809 of file master.h.

STATUS ABA_MASTER::status_ [private]

The current status of the optimization.

Definition at line 1813 of file master.h.

ABA_COWTIMER ABA_MASTER::totalCowTime_ [private]

The timer for the total elapsed time.

Definition at line 1817 of file master.h.

ABA_CPUTIMER ABA_MASTER::totalTime_ [private]

The timer for the total cpu time for the optimization.

Definition at line 1821 of file master.h.

ABA_CPUTIMER ABA_MASTER::lpTime_ [private]

The timer for the cpu time spent in the LP-interface.

Definition at line 1825 of file master.h.

ABA_CPUTIMER ABA_MASTER::lpSolverTime_ [private]

Definition at line 1826 of file master.h.

ABA_CPUTIMER ABA_MASTER::separationTime_ [private]

The timer for the cpu time spent in the separation

Definition at line 1830 of file master.h.

ABA_CPUTIMER ABA_MASTER::improveTime_ [private]

The timer for the cpu time spent in the heuristics for the computation of feasible solutions.

Definition at line 1835 of file master.h.

ABA_CPUTIMER ABA_MASTER::pricingTime_ [private]

The timer for the cpu time spent in pricing.

Definition at line 1839 of file master.h.

ABA_CPUTIMER ABA_MASTER::branchingTime_ [private]

The timer for the cpu time spent in determining the branching rules.

Definition at line 1843 of file master.h.

int ABA_MASTER::nSub_ [private]

The number of generated subproblems.

Definition at line 1847 of file master.h.

int ABA_MASTER::nLp_ [private]

The number of solved LPs.

Definition at line 1851 of file master.h.

int ABA_MASTER::highestLevel_ [private]

The highest level which has been reached in the enumeration tree.

Definition at line 1855 of file master.h.

int ABA_MASTER::nFixed_ [private]

The total number of fixed variables.

Definition at line 1859 of file master.h.

int ABA_MASTER::nAddCons_ [private]

The total number of added constraints.

Definition at line 1863 of file master.h.

int ABA_MASTER::nRemCons_ [private]

The total number of removed constraints.

Definition at line 1867 of file master.h.

int ABA_MASTER::nAddVars_ [private]

The total number of added variables.

Definition at line 1871 of file master.h.

int ABA_MASTER::nRemVars_ [private]

The total number of removed variables.

Definition at line 1875 of file master.h.

int ABA_MASTER::nNewRoot_ [private]

The number of changes of the root of the remaining \ tree.

Definition at line 1879 of file master.h.

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

Include/abacus/master.h

Generated on Tue Aug 14 18:09:58 2007 for ABACUS by

1.5.1

ABA_MASTER Class Reference

Public Types

Public Member Functions

Static Public Attributes

Protected Member Functions

Private Member Functions

Private Attributes

Friends

Detailed Description

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation

Friends And Related Function Documentation

Member Data Documentation