cleanProfileResults: Clean a profile extrema object

In profExtrema: Compute and Visualize Profile Extrema Functions

Description

The function cleanProfileResults cleans a profile extrema object to partially redo some computations.

Usage

cleanProfileResults(object, level = 1)

Arguments

object	a list containing profile extrema results.
level	an integer 1-4 denoting how much it should be removed from object. See Value for details.

Value

returns object with the deleted parts as selected by level. In particular

• 1:keep only profMean_full.

• 2:keep profMean_full and profMean_approx. Remove all UQ results.

• 3:keep profMean_full and profMean_approx and the pilot points. Remove all UQ simulations.

• 4:Remove only the bound computations.

Author(s)

Dario Azzimonti

Examples

if (!requireNamespace("DiceKriging", quietly = TRUE)) {
stop("DiceKriging needed for this example to work. Please install it.",
     call. = FALSE)
}
# Compute a kriging model from 50 evaluations of the Branin function
# Define the function
g=function(x){
  return(-branin(x))
}
gp_des<-lhs::maximinLHS(20,2)
reals<-apply(gp_des,1,g)
kmModel<-km(design = gp_des,response = reals,covtype = "matern3_2")

threshold=-10

# Compute coordinate profiles on the posterior mean
# Increase multistart and size of designs for more precise results
options_full<-list(multistart=2,heavyReturn=TRUE, Design = replicate(2,seq(0,1,,50)))
init_des<-lhs::maximinLHS(12,2)
options_approx<- list(multistart=2,heavyReturn=TRUE,initDesign=init_des,fullDesignSize=50)
cProfilesMean<-coordinateProfiles(object=kmModel,threshold=threshold,options_full=options_full,
                                  options_approx=options_approx,uq_computations=FALSE,
                                  plot_level=3,plot_options=NULL,CI_const=NULL,return_level=2)

# If we want to run again the computation of approximate coordinate profiles
# we delete that result and run again the coordinate profiles function
cProfiles_full <- cleanProfileResults(cProfilesMean,level=1)
## Not run: 
# Coordinate profiles with UQ with approximate profiles
plot_options<-list(save=FALSE, titleProf = "Coordinate profiles",
                   title2d = "Posterior mean",qq_fill=TRUE)
cProfilesUQ<-coordinateProfiles(object=cProfilesMean,threshold=threshold,options_full=options_full,
                                  options_approx=options_approx,uq_computations=TRUE,
                                  plot_level=3,plot_options=NULL,CI_const=NULL,return_level=2)
# If we would like to remove all UQ results
cProf_noUQ <- cleanProfileResults(cProfilesUQ,level=2)

# If we would like to remove the simulations but keep the pilot points
cProf_noSims <- cleanProfileResults(cProfilesUQ,level=3)
# the line above is useful, for example, if we need a more accurate UQ. In that case
# we obtain more simulations with the same pilot points and then combine the results.


## End(Not run)

profExtrema documentation built on March 22, 2020, 1:07 a.m.