dev.nm.solver.multivariate.unconstrained.annealing.acceptanceprobabilityfunction

Class MetropolisAcceptanceProbabilityFunction

java.lang.Object

dev.nm.solver.multivariate.unconstrained.annealing.acceptanceprobabilityfunction.MetropolisAcceptanceProbabilityFunction

All Implemented Interfaces:

TemperedAcceptanceProbabilityFunction

public class MetropolisAcceptanceProbabilityFunction
extends Object
implements TemperedAcceptanceProbabilityFunction

Uses the classic Metropolis rule, f_{t+1}/f_t. The density function is given by:

 f(currentState) = currentEnergy
 f(previousState) = previousEnergy
 exp(-f(proposedState) / temperature) = exp(-proposedEnergy / temperature)
 

The reference has it: \(\pi_t(\mathbf{x}) = exp(-h(\mathbf{x})/T_t))\) This is the default acceptance probability function for 'normal' annealing.

See Also:

"Jun S. Liu. "p.209," Monte Carlo Strategies in Scientific Computing. 2002."

Constructor Summary

Constructors

Constructor and Description
MetropolisAcceptanceProbabilityFunction()

Method Summary

Modifier and Type	Method and Description
double	acceptanceProbability(Vector currentState, double currentEnergy, Vector proposedState, double proposedEnergy, double temperature) Computes the probability that the next state transition will be accepted.

Methods inherited from class java.lang.Object

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

Constructor Detail

MetropolisAcceptanceProbabilityFunction

public MetropolisAcceptanceProbabilityFunction()

Method Detail

acceptanceProbability

public double acceptanceProbability(Vector currentState,
                                    double currentEnergy,
                                    Vector proposedState,
                                    double proposedEnergy,
                                    double temperature)

Description copied from interface: TemperedAcceptanceProbabilityFunction

Computes the probability that the next state transition will be accepted.

Specified by:

acceptanceProbability in interface TemperedAcceptanceProbabilityFunction

Parameters:

currentState - the current state of the system

currentEnergy - the energy in the current state

proposedState - the proposed next state of the system

proposedEnergy - the energy in the proposed state

temperature - the current temperature

Returns:

the acceptance probability