caRamel is a multiobjective evolutionary algorithm combining the MEAS algorithm and the NGSA-II algorithm.
Download the package from CRAN or GitHub and then install and load it.
It is possible to compute the first order derivatives of the Pareto front with caRamel by setting the logical parameter sensitivity to TRUE.
library(caRamel)
Schaffer test function has two objectives with one variable.
schaffer <- function(i) {
s1 <- x[i,1] * x[i,1]
s2 <- (x[i,1] - 2) * (x[i,1] - 2)
return(c(s1, s2))
}
Note that :
For instance, the variable will lie in the range [-10, 10]:
nvar <- 1 # number of variables
bounds <- matrix(data = 1, nrow = nvar, ncol = 2) # upper and lower bounds
bounds[, 1] <- -10 * bounds[, 1]
bounds[, 2] <- 10 * bounds[, 2]
Both functions are to be minimized:
nobj <- 2 # number of objectives
minmax <- c(FALSE, FALSE) # min and min
Before calling caRamel in order to optimize the Schaffer’s problem, some algorithmic parameters need to be set:
popsize <- 100 # size of the genetic population
archsize <- 100 # size of the archive for the Pareto front
maxrun <- 1000 # maximum number of calls
prec <- matrix(1.e-3, nrow = 1, ncol = nobj) # accuracy for the convergence phase
Then the minimization problem can be launched with a sensitivity analysis:
results <-
caRamel(nobj,
nvar,
minmax,
bounds,
schaffer,
popsize,
archsize,
maxrun,
prec,
carallel=FALSE,
sensitivity=TRUE) # sensitivity required
## Beginning of caRamel optimization <-- Wed Feb 2 07:29:58 2022
## Number of variables : 1
## Number of functions : 2
## Computing the sensitivity of the Pareto front...
## Done in 7.53024888038635 secs --> Wed Feb 2 07:30:05 2022
## Size of the Pareto front : 99
## Number of calls : 1119
Test if the convergence is successful:
print(results$success==TRUE)
## [1] TRUE
Plot the Pareto front:
plot(results$objectives[,1], results$objectives[,2], main="Schaffer Pareto front", xlab="Objective #1", ylab="Objective #2")
plot(results$parameters, main="Corresponding values for X", xlab="Element of the archive", ylab="X Variable")
The sensitivity of the Pareto front is evalutated by computing first order derivatives. For each of the objective, one Jacobian matrix is computed:
names(results$derivatives)
## [1] "Jacobian_1" "Jacobian_2"
Plot the sensitivity for the first objective:
plot(results$parameters, results$derivatives$Jacobian_1, main="Sensitivitiy for the first objective", ylab="Sensitivity values", xlab="X values")
Plot the histogram for the second objective:
hist(results$derivatives$Jacobian_2, main="Sensitivitiy for the second objective", xlab="Sensitivity values", ylab="Distribution of the Pareto front")
Plot the sensitivity of the Pareto front for the two objectives:
plot(results$derivatives$Jacobian_1, results$derivatives$Jacobian_2, main="Sensitivitiy for both objectives", ylab="Sensitivity values #2", xlab="Sensitivity values #1")