Initialise: This function returns the default controls and other useful...

In LorenzoGentile/SCGA: Genetic Algorithm for for structured and variable-sized problems

Description

This function returns the default controls and other useful variables. Control is a list of the settings:

Usage

Initialise(control = list(), ...)

Arguments

convergence	Stopping criterion: absolute difference between the current best and the known minimum
cpus	numeric. indicatig number of cores over which parallelise
creatCandFun	function. See createCandidate
createMutFun	function. See createMutFun
crossFun	function. See crossFun
dontChangeCross	numeric vector. Feature number that not undergo to Crossover
dontChangeMut	numeric vector. Feature number that not undergo to Mutation
elitism	numeric. Number of candidates to preserve to the next population. Default is size / 10
evaluatePopDF	function. See evaluatePopDF
feature	list or function that creates the list. See feature
fitnessFN	function. Receives the observations of the objective functions and returns
Fun	function. Objective function a vector of the same length repesententing the fitness. Default is Ranking fitness.
maxStallGenerations	numeric. Maximum number of iterations without improvements. If overcomen, the population is reinitialised.
keep	vector of characters. Additional columns in the matrix representing the candidate.
localOptGenerations	numeric. Maximum number of iterations without improvements. If overcomen, a local optimisation on the numeric variables starts from the best solution found freezing the remaining genes. Then, the population is reinitialised.
localOptimiser	function. Function that performs the local optimisation. Default si optim. function. LocalOptimisationMatlab is also an option. It starts connection with matlab and uses fmnincon.
maxEvaluations	numeric. Stopping criterion. Maximum number of evaluations allowed. If more stopping criterion are given, the more strict will be used.
maxGenerations	numeric. Stopping criterion. Maximum number of generations allowed. If more stopping criterion are given, the more strict will be used.
multiPopulation	Boolean. Use or not multiPopulation strategy. controls ar specified in multiPopControl
maxRelaxation	numeric. Value in [0,1]. Indicates the fraction of constraint relaxation at the beginning of the optimisation.
multiPopControl	list. controls are :...
mutRate	numeric. Value in [0,1]. Probablity to mutate a candidate
parallel	Boolean. Indicates wheter to create a cluster with MakeCluster command using the number of cores indicated by @param cpus.
percCross	numeric. Value in [0,1]. Indicates the maximum percentage of genes to swap during crossover.
percMut	numeric. Value in [0,1]. Indicates the maximum percentage of genes to mutate
plotEvolution	Boolean. plot the evoluation of the best found solution.
plotEvolutionLimit	numeric. Upper limit for the plotEvolution plot. Helps the visualisation when the initial best is far from final best.
plotFitness	Boolean. If there are constraints, it produces a plot that shows the fitness in respect of the objective function value and feasibility.
plotPopulation	Boolean. Plot an historgram for every gene. The histograms show the count of the values assumed in the current population.
plotSigma	Boolean. Plot an historgram for every sigma The histograms show the count of the values assumed in the current population.
plotInterval	integer. Create the plots every plotInterval generations.
popCreateFun	function. It creates new candidates. Default is createCandidate
printIter	Boolean. Print on screen the evolution of the optimisation.
printSigma	Boolean. Print on screen the mean values of sigma.
printXMin	Boolean. Print on screen the current xbest.
printPlot	Boolean. Save plots in a dedicated folder: currentDirectory/runResults/control$job$algo.name/control$seed
probability	vector. It specifies the probability of every gene to be selected by the operators. Default is all 1.
pureFeasibility	numeric. Value in [0,1]. Fraction of the available budget to be spent without constraint relaxation.
repairFun	function. Repair function used to repair the possible corrupted candidates.
resume	Boolean. Restart the optimisation loading a backup RData names as @param resumeFrom.
resumeFrom	character. Name for a possible backup RData
saveAll	Boolean. save all the x at each iteration
seed	integer. Seed to use for repetitivity .
selection	function. selection method function. See selectpoolTournament
size	integer. Population size
target	numeric. Knwown minimum value achievable. Stopping criterion. If reached wihin the specified tolerance @param convergence
tournamentSize	integer. tournament size for selectpoolTournament
updateSigma	Boolean. To use adaptive step size mutation
useCrossover	Boolean. To crossover as operator.
vectorOnly	Boolean. Pass to the objective function the candidate as vector.
vectorized	Boolean. Pass to the objective function the entire population.

LorenzoGentile/SCGA documentation built on June 29, 2021, 4:15 p.m.