SPOT (version 2.11.10)

optimES: Evolution Strategy

Description

This is an implementation of an Evolution Strategy.

Usage

optimES(x = NULL, fun, lower, upper, control = list(), ...)

Arguments

x

optional start point, not used

fun

objective function, which receives a matrix x and returns observations y

lower

is a vector that defines the lower boundary of search space (this also defines the dimensionality of the problem)

upper

is a vector that defines the upper boundary of search space (same length as lower)

control

list of control parameters. The control list can contain the following settings:

funEvals

number of function evaluations, stopping criterion, default is 500

mue

number of parents, default is 10

nu

selection pressure. That means, number of offspring (lambda) is mue multiplied with nu. Default is 10

mutation

string of mutation type, default is 1

sigmaInit

initial sigma value (step size), default is 1.0

nSigma

number of different sigmas, default is 1

tau0

number, default is 0.0. tau0 is the general multiplier.

tau

number, learning parameter for self adaption, i.e. the local multiplier for step sizes (for each dimension).default is 1.0

rho

number of parents involved in the procreation of an offspring (mixing number), default is "bi"

sel

number of selected individuals, default is 1

stratReco

Recombination operator for strategy variables. 1: none. 2: dominant/discrete (default). 3: intermediate. 4: variation of intermediate recombination.

objReco

Recombination operator for object variables. 1: none. 2: dominant/discrete (default). 3: intermediate. 4: variation of intermediate recombination.

maxGen

number of generations, stopping criterion, default is Inf

seed

number, random seed, default is 1

noise

number, value of noise added to fitness values, default is 0.0

verbosity

defines output verbosity of the ES, default is 0

plotResult

boolean, specifies if results are plotted, default is FALSE

logPlotResult

boolean, defines if plot results should be logarithmic, default is FALSE

sigmaRestart

number, value of sigma on restart, default is 0.1

preScanMult

initial population size is multiplied by this number for a pre-scan, default is 1

globalOpt

termination criterion on reaching a desired optimum value, default is rep(0,dimension)

...

additional parameters to be passed on to fun

Value

list, with elements

x

NULL, currently not used

y

NULL, currently not used

xbest

best solution

ybest

best observation

count

number of evaluations of fun

Examples

# NOT RUN {
cont <- list(funEvals=100)
optimES(fun=funSphere,lower=rep(0,2), upper=rep(1,2), control= cont) 

# }