org.cloudsimplus.heuristics

Class SimulatedAnnealing<S extends HeuristicSolution<?>>

java.lang.Object

org.cloudsimplus.heuristics.HeuristicAbstract<S>

org.cloudsimplus.heuristics.SimulatedAnnealing<S>

Type Parameters:

S - the class of solutions the heuristic will deal with, starting with a random solution and execute the solution search in order to achieve a satisfying solution (defined by a stop criteria)

All Implemented Interfaces:

Heuristic<S>

Direct Known Subclasses:

CloudletToVmMappingSimulatedAnnealing

public abstract class SimulatedAnnealing<S extends HeuristicSolution<?>>
extends HeuristicAbstract<S>

A base class for implementation of Simulated Annealing algorithms used to find a suboptimal solution for a problem defined by sub-classes of this one. The Simulated Annealing is a heuristic that starts with a random solution and iteratively generates a random neighbor solution that its fitness is assessed in order to reach a sub-optimal result. The algorithm try to avoid local maximums, randomly selecting worse solutions to get away from being stuck in these locals.

The algorithm basically works as follows:

1. Starts generating a random solution as you wish;
2. Computes its fitness using some function (defined by the developer implementing the heuristic);
3. Generates a neighbor random solution from the current solution and computes its fitness;
4. Assesses the neighbor and current solution (the conditions below are ensured by the getAcceptanceProbability() method):
   • if neighbor.getFitness() > current.getFitness() then move to the new solution;
   • if neighbor.getFitness() < current.getFitness() then randomly decide if move to the new solution;
5. Repeat steps 3 to 4 until an aceptable solution is found or some number of iterations or time is reached. These conditions are defined by the developer implementing the heuristic.

Since:

CloudSim Plus 1.0

Author:

Manoel Campos da Silva Filho

See Also:

[1] R. A. Rutenbar, “Simulated Annealing Algorithms: An overview,” IEEE Circuits Devices Mag., vol. 1, no. 5, pp. 19–26, 1989.

Field Summary

Fields inherited from interface org.cloudsimplus.heuristics.Heuristic

LOGGER, NULL

Method Summary

Modifier and Type	Method and Description
double	getAcceptanceProbability() Computes the acceptance probability to define if a neighbor solution has to be accepted or not, compared to the Heuristic.getBestSolutionSoFar().
double	getColdTemperature()
double	getCoolingRate()
double	getCurrentTemperature() Gets the current system temperature that represents the system state at the time of the method call.
boolean	isToStopSearch() Checks if the solution search can be stopped.
void	setColdTemperature(double coldTemperature) Sets the temperature that defines the system is cold enough and solution search may be stopped.
void	setCoolingRate(double coolingRate) Sets the percentage rate in which the system will be cooled, in scale from [0 to 1[.
protected void	setCurrentTemperature(double currentTemperature) Sets the current system temperature.
void	updateSystemState() Updates the state of the system in order to keep looking for a suboptimal solution.

Methods inherited from class org.cloudsimplus.heuristics.HeuristicAbstract

getBestSolutionSoFar, getNeighborhoodSearchesByIteration, getNeighborSolution, getRandom, getRandomValue, getSolveTime, setBestSolutionSoFar, setNeighborhoodSearchesByIteration, setNeighborSolution, setSolveTime, solve

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.cloudsimplus.heuristics.Heuristic

createNeighbor, getInitialSolution

Method Detail

getAcceptanceProbability

public double getAcceptanceProbability()

Computes the acceptance probability to define if a neighbor solution has to be accepted or not, compared to the Heuristic.getBestSolutionSoFar().

It is used the Boltzmann distribution to define the probability of a worse solution (considering its cost) to be accepted or not in order to avoid local minima. The computed Boltzmann factor also ensures that better solutions are always accepted. The Boltzmann Constant has different values depending of the used unit. In this case, it was used the natural unit of information.

Returns:

the acceptance probability, in scale from [0 to 1] where 0 is to maintain the current solution, 1 is to accept the neighbor solution, while intermediate values defines the probability that the neighbor solution will be randomly accepted.

See Also:

Boltzmann distribution, Boltzmann constant

isToStopSearch

public boolean isToStopSearch()

Checks if the solution search can be stopped.

Returns:

true if the system is cold enough and solution search can be stopped, false otherwise

updateSystemState

public void updateSystemState()

Updates the state of the system in order to keep looking for a suboptimal solution. Cools the system at a the defined cooling rate.

Specified by:

updateSystemState in class HeuristicAbstract<S extends HeuristicSolution<?>>

See Also:

()

getCurrentTemperature

public double getCurrentTemperature()

Gets the current system temperature that represents the system state at the time of the method call.

Returns:

the current system temperature

setCurrentTemperature

protected void setCurrentTemperature(double currentTemperature)

Sets the current system temperature.

Parameters:

currentTemperature - the temperature to set

getCoolingRate

public double getCoolingRate()

Returns:

percentage rate in which the system will be cooled, in scale from [0 to 1[.

setCoolingRate

public void setCoolingRate(double coolingRate)

Sets the percentage rate in which the system will be cooled, in scale from [0 to 1[.

Parameters:

coolingRate - the rate to set

getColdTemperature

public double getColdTemperature()

Returns:

the temperature that defines the system is cold enough and solution search may be stopped.

setColdTemperature

public void setColdTemperature(double coldTemperature)

Sets the temperature that defines the system is cold enough and solution search may be stopped.

Parameters:

coldTemperature - the cold temperature to set