ThresholdScaleFitness {xegaSelectGene} | R Documentation |
Dispersion Ratio Based Threshold Fitness Scaling.
Description
Fitness is transformed by a power function with a scaling exponent. The choice of the scaling exponent depends on the ratio of the dispersion measures of the current and the previous population fitness.
Usage
ThresholdScaleFitness(fit, lF)
Arguments
fit |
Fitness vector. |
lF |
Local configuration. |
Details
The scaling exponent is selected by the following rule:
If
lF$RDM()>1+lF$ScalingThreshold()
then choose the scaling exponentlF$ScalingExp()
. The scaling exponent should be larger than 1 to increase the selection pressure.If
lF$RDM()<1+lF$SCalingThreshold
andlF$RDM()>1-lF$SCalingThreshold
, the fitness is not scaled.If
lF$RDM()<1-lF$SCalingThreshold
then choose the scaling exponentlF$ScalingExp2()
. The scaling exponent should be smaller than 1 to decrease the selection pressure.
Value
Scaled fitness vector.
See Also
Other Scaling:
ContinuousScaleFitness()
,
DispersionRatio()
,
ScaleFitness()
,
ScalingFitness()
Other Adaptive Parameter:
ContinuousScaleFitness()
Examples
lF<-list()
lF$Offset<-parm(0.0001)
lF$ScalingThreshold<-parm(0.05)
lF$RDM<-parm(1.0)
lF$ScalingExp<-parm(2.0)
lF$ScalingExp2<-parm(0.5)
fit<-sample(10, 20, replace=TRUE)
fit
ThresholdScaleFitness(fit, lF)
lF$RDM<-parm(1.2)
ThresholdScaleFitness(fit, lF)
lF$RDM<-parm(0.8)
ThresholdScaleFitness(fit, lF)