ifs.solver
Class Solver

java.lang.Object
  ifs.solver.Solver

public class Solver
extends Object

IFS Solver.

The iterative forward search (IFS) algorithm is based on ideas of local search methods. However, in contrast to classical local search techniques, it operates over feasible, though not necessarily complete solutions. In such a solution, some variables can be left unassigned. Still all hard constraints on assigned variables must be satisfied. Similarly to backtracking based algorithms, this means that there are no violations of hard constraints.

This search works in iterations. During each step, an unassigned or assigned variable is initially selected. Typically an unassigned variable is chosen like in backtracking-based search. An assigned variable may be selected when all variables are assigned but the solution is not good enough (for example, when there are still many violations of soft constraints). Once a variable is selected, a value from its domain is chosen for assignment. Even if the best value is selected (whatever “best” means), its assignment to the selected variable may cause some hard conflicts with already assigned variables. Such conflicting variables are removed from the solution and become unassigned. Finally, the selected value is assigned to the selected variable.

Algorithm schema:

procedure ifs(initial) // initial solution is the parameter
iteration = 0; // iteration counter
current = initial; // current (partial) feasible solution
best = initial; // best solution
while canContinue(current, iteration) do
iteration = iteration + 1;
variable = selectVariable(current);
value = selectValue(current, variable);
UNASSIGN(current, CONFLICTING_VARIABLES(current, variable, value));
ASSIGN(current, variable, value);
if better(current, best) then best = current;
end while
return best;
end procedure


The algorithm attempts to move from one (partial) feasible solution to another via repetitive assignment of a selected value to a selected variable. During this search, the feasibility of all hard constraints in each iteration step is enforced by unassigning the conflicting variables. The search is terminated when the requested solution is found or when there is a timeout, expressed e.g., as a maximal number of iterations or available time being reached. The best solution found is then returned.

The above algorithm schema is parameterized by several functions, namely:

• the termination condition (function canContinue, see TerminationCondition),
• the solution comparator (function better, see SolutionComparator),
• the variable selection (function selectVariable, see VariableSelection) and
• the value selection (function selectValue, see ValueSelection).

Usage:

DataProperties cfg = ToolBox.loadProperties(inputCfg); //input configuration
Solver solver = new Solver(cfg);
solver.setInitalSolution(model); //sets initial solution

solver.start(); //server is executed in a thread

try { //wait untill the server finishes
solver.getSolverThread().join();
} catch (InterruptedException e) {}

Solution solution = solver.lastSolution(); //last solution
solution.restoreBest(); //restore best solution ever found


Solver's parameters:

Parameter	Type	Comment
General.SaveBestUnassigned	Integer	During the search, solution is saved when it is the best ever found solution and if the number of assigned variables is less or equal this parameter (if set to -1, the solution is always saved)
General.Seed	Long	If set, random number generator is initialized with this seed
General.SaveConfiguration	Boolean	If true, given configuration is stored into the output folder (during initialization of the solver, ${General.Output}/${General.ProblemName}.properties)
Solver.AutoConfigure	Boolean	If true, IFS Solver is configured according to the following parameters
Termination.Class	String	Fully qualified class name of the termination condition (see TerminationCondition, e.g. GeneralTerminationCondition)
Comparator.Class	String	Fully qualified class name of the solution comparator (see SolutionComparator, e.g. GeneralSolutionComparator)
Value.Class	String	Fully qualified class name of the value selection criterion (see ValueSelection, e.g. GeneralValueSelection)
Variable.Class	String	Fully qualified class name of the variable selection criterion (see VariableSelection, e.g. GeneralVariableSelection)
PerturbationCounter.Class	String	Fully qualified class name of the perturbation counter in case of solving minimal perturbation problem (see PerturbationsCounter, e.g. DefaultPerturbationsCounter)
Extensions.Classes	String	Semi-colon separated list of fully qualified class names of IFS extensions (see Extension, e.g. ConflictStatistics or MacPropagation)

Version:

1.0

Author:

Tomáš Müller

See Also:

SolverListener, Model, Solution, TerminationCondition, SolutionComparator, PerturbationsCounter, VariableSelection, ValueSelection, Extension

Nested Class Summary

protected class

Solver.SolverThread Solver thread

Field Summary

protected Solution	iCurrentSolution current solution
protected boolean	iDone solver status: done
protected Solution	iLastSolution last solution (after IFS Solver finishes)
protected boolean	iRunning solver status: running
protected Solver.SolverThread	iSolverThread solver thread
protected boolean	iStop solver status: stopped
protected static Logger	sLogger log

Constructor Summary

Solver(DataProperties properties)
Constructor.

Method Summary

void	addExtension(Extension extension) Add an IFS extension
void	addSolverListener(SolverListener listener) Adds a solver listener
protected void	autoConfigure() Automatic configuratin of the solver -- when Solver.AutoConfigure is true
void	clearBest() Clears best solution
Solution	currentSolution() Current solution (during the search)
Vector	getExtensions() Returns list of all used extensions
PerturbationsCounter	getPerturbationsCounter() Returns perturbation counter (minimal perturbation problem)
DataProperties	getProperties() Returns configuration
SolutionComparator	getSolutionComparator() Returns solution comparator
Thread	getSolverThread() Returns solver's thread
TerminationCondition	getTerminationCondition() Returns termination condition
ValueSelection	getValueSelection() Returns values selection criterion
VariableSelection	getVariableSelection() Returns variable selection criterion
void	init() Initialization
boolean	isDone() Solver status: solver is done
boolean	isRunning() Solver status: solver is running
boolean	isStopped() Solver status: solver is stopped
Solution	lastSolution() Last solution (when solver finishes)
void	removeSolverListener(SolverListener listener) Removes a solver listener
void	setInitalSolution(Model model) Sets initial solution
void	setInitalSolution(Solution solution) Sets initial solution
void	setPerturbationsCounter(PerturbationsCounter perturbationsCounter) Sets perturbation counter (minimal perturbation problem)
void	setSolutionComparator(SolutionComparator solutionComparator) Sets solution comparator
void	setTerminalCondition(TerminationCondition terminationCondition) Sets termination condition
void	setValueSelection(ValueSelection valueSelection) Sets value selection criterion
void	setVariableSelection(VariableSelection variableSelection) Sets variable selection criterion
void	start() Starts solver
void	stopSolver() Stops the running solver

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

sLogger

protected static Logger sLogger

log

iCurrentSolution

protected Solution iCurrentSolution

current solution

iLastSolution

protected Solution iLastSolution

last solution (after IFS Solver finishes)

iDone

protected boolean iDone

solver status: done

iRunning

protected boolean iRunning

solver status: running

iStop

protected boolean iStop

solver status: stopped

iSolverThread

protected Solver.SolverThread iSolverThread

solver thread

Constructor Detail

Solver

public Solver(DataProperties properties)

Constructor.

Parameters:

properties - input configuration

Method Detail

setTerminalCondition

public void setTerminalCondition(TerminationCondition terminationCondition)

Sets termination condition

setSolutionComparator

public void setSolutionComparator(SolutionComparator solutionComparator)

Sets solution comparator

setValueSelection

public void setValueSelection(ValueSelection valueSelection)

Sets value selection criterion

setVariableSelection

public void setVariableSelection(VariableSelection variableSelection)

Sets variable selection criterion

setPerturbationsCounter

public void setPerturbationsCounter(PerturbationsCounter perturbationsCounter)

Sets perturbation counter (minimal perturbation problem)

addExtension

public void addExtension(Extension extension)

Add an IFS extension

getTerminationCondition

public TerminationCondition getTerminationCondition()

Returns termination condition

getSolutionComparator

public SolutionComparator getSolutionComparator()

Returns solution comparator

getValueSelection

public ValueSelection getValueSelection()

Returns values selection criterion

getVariableSelection

public VariableSelection getVariableSelection()

Returns variable selection criterion

getPerturbationsCounter

public PerturbationsCounter getPerturbationsCounter()

Returns perturbation counter (minimal perturbation problem)

getExtensions

public Vector getExtensions()

Returns list of all used extensions

addSolverListener

public void addSolverListener(SolverListener listener)

Adds a solver listener

removeSolverListener

public void removeSolverListener(SolverListener listener)

Removes a solver listener

getProperties

public DataProperties getProperties()

Returns configuration

autoConfigure

protected void autoConfigure()

Automatic configuratin of the solver -- when Solver.AutoConfigure is true

clearBest

public void clearBest()

Clears best solution

setInitalSolution

public void setInitalSolution(Solution solution)

Sets initial solution

setInitalSolution

public void setInitalSolution(Model model)

Sets initial solution

start

public void start()

Starts solver

getSolverThread

public Thread getSolverThread()

Returns solver's thread

init

public void init()

Initialization

lastSolution

public Solution lastSolution()

Last solution (when solver finishes)

currentSolution

public Solution currentSolution()

Current solution (during the search)

isDone

public boolean isDone()

Solver status: solver is done

isRunning

public boolean isRunning()

Solver status: solver is running

stopSolver

public void stopSolver()

Stops the running solver

isStopped

public boolean isStopped()

Solver status: solver is stopped