run_epic_nm: This is a main function which iterates an EPIC algorithm...

In ingridas/EPICR: An Implementation of Efficient Pareto Iterative Classification Method

Description

This is a main function which iterates an EPIC algorithm function, saves data

Usage

run_epic_nm(maxEval, nobj, nvar, L, U, func, problem.name = NULL,
  local.search = TRUE, p.next = 0.6, p.pareto = 0.7, strategy = 1,
  repN = 500, absmin = NULL, absmax = NULL, con = FALSE,
  stopping = FALSE)

Arguments

maxEval	- maximum number of evaluations
nobj	- number of objectives
nvar	- number of decision variables
L, U	- row vectors of lower and upper bounds of the design space
func	- TRUE, if there is availble function, FALSE if we have only a table of evaluated data (experiments)
problem.name	- the string of problem corresponding to its function, NULL - if problem function is unavailable
local.search	- indicates what EPIC version will be run. TRUE - enhanced EPIC version,FALSE - the original one. DEFAULT = TRUE.
p.next	- the propbability value used to select a decision vector, defalut value is 0.6
p.pareto	- indicator used to assign points to one of the set: dominated or nondominated, default value is 0.7
strategy	- strategy id used to select nex vector to evaluate, default value is 1
repN	- a size of decision space representation sample
absmin	- the estimate of minimum values of objective function (if not provided, calculated minimum value is used)
absmax	- the estimate of maximum values of objective functions (if not provided we will use the max value of Y)
con	- constraints, an analytical function cheap to evaluate, in a form g_i(x)>=0, if available; otherwise it is equal to FALSE
stopping	- an indicator used to stop an algorithm at every iteration; TRUE - if algorithm is stopping, otherwise FALSE

Value

Pareto optimal solutions of evaluated vectors as well as all evaluated solutions.

ingridas/EPICR documentation built on May 18, 2019, 4:54 a.m.