tests/bdstest.R

# rm(list=ls())
##  author: John C. Nash
# fname<-paste(format(Sys.time(), "%Y%m%d%H%M"),"-btnvm.out",sep='')
# sink(fname, append=TRUE, split=TRUE)
require("optimx")
# Following is used when starting from opx21 directory
# source("optimx/tests/simplefun.R")
# Following is for use in package testing
# Simple Test Function 1:
simfun.f = function(x) { 
     fun <- sum(x^2 )
#	print(c(x = x, fun = fun))
     fun
}
simfun.g = function(x) { 
     grad<-2.0*x
     grad
}
simfun.h = function(x) { 
     n<-length(x)
     t<-rep(2.0,n)
     hess<-diag(t)
}
sessionInfo()
#####################

# This test script illustrates the use of bounds in optimr() with the
# optimizers nvm and L-BFGS-B, as well as a Kuhn Karush Tucker check 
# on the final parameters from the second optimization.
# Masks are tested at the very end for the two methods for which they are
# available. Note that they must be called via the opm() function.


n<-4
lower<-rep(0,n)
upper<-lower # to get arrays set
bdmsk<-rep(1,n)
# bdmsk[(trunc(n/2)+1)]<-0
for (i in 1:n) { 
    lower[i]<-1.0*(i-1)*(n-1)/n
    upper[i]<-1.0*i*(n+1)/n
}
xx<-0.5*(lower+upper)

cat("lower bounds:")
print(lower)
cat("start:       ")
print(xx)
cat("upper bounds:")
print(upper)

cat("nvm\n") # changed from Rvmmin 2023-10-22

abtrvm <- optimr(xx, simfun.f, simfun.g, lower=lower, upper=upper, 
        method="nvm", control=list(trace=0))
# Note: use lower=lower etc. because there is a missing hess= argument
proptimr(abtrvm)

cat("Axial search")
axabtrvm <- axsearch(abtrvm$par, fn=simfun.f, fmin=abtrvm$value, lower, upper, bdmsk=NULL)
print(axabtrvm)

cat("Now force an early stop\n")
abtrvm1 <- optimr(xx, simfun.f, simfun.g, lower=lower, upper=upper, method="nvm", 
                  control=list(maxit=1, trace=0))
print(abtrvm1)
cat("Axial search")
axabtrvm1 <- axsearch(abtrvm1$par, fn=simfun.f, fmin=abtrvm1$value, lower, upper, bdmsk=NULL)
print(axabtrvm1)


cat("Maximization test\n")
mabtrvm <- optimr(xx, simfun.f, simfun.g, lower=lower, upper=upper, method="nvm", 
                 control=list(trace=1, maximize=TRUE))
# Note: use lower=lower etc. because there is a missing hess= argument
print(mabtrvm)
cat("Do NOT try axsearch() with maximize\n")
cat("KKT condition check\n")
akktm <- kktchk(mabtrvm$par, simfun.f, simfun.g, hess=NULL, upper=upper, lower=lower,  
              maximize=TRUE, control=list(trace=0))
print(akktm)

alb<-optimr(xx,simfun.f, simfun.g, lower=lower, upper=upper, method="L-BFGS-B", 
            control=list(trace=0))
print(alb)

cat("KKT condition check\n")
alkkt <- kktchk(alb$par, simfun.f, simfun.g, hess=NULL, upper=upper, lower=lower,  maximize=FALSE, control=list(trace=0))
print(alkkt)

alhn<-optimr(xx, simfun.f, lower=lower, upper=upper, method="hjn", 
             control=list(trace=0))
print(alhn)

#sink()
cat("All bounded methods attempt with opm\n") # ?? should give errors
allbds <- opm(xx, simfun.f, simfun.g, lower=lower, upper=upper, method="MOST", control=list(trace=0))
print(summary(allbds, order=value))

cat("Now force a mask upper=lower for parameter 3 and see what happens\n")
lower[3] <- upper[3]
xx[3] <- lower[3] # MUST reset parameter also


ncgbdm <- optimr(xx, simfun.f, simfun.g, lower=lower, upper=upper, method="ncg", 
                 control=list(trace=1, watch=TRUE))
proptimr(ncgbdm)

## lbfmsk <- optim(xx, simfun.f, simfun.g, lower=lower, upper=upper, method="L-BFGS-B", 
##                  control=list(trace=1, watch=TRUE))
## proptimr(ncgbdm)

allbdm <- try(opm(xx, simfun.f, simfun.g, lower=lower, upper=upper, method="MOST", 
               control=list(trace=2)))
print(summary(allbdm, order=value))

mmth <- ctrldefault(2)$maskmeth
allmsk <- try(opm(xx, simfun.f, simfun.g, lower=lower, upper=upper, method=mmth, 
                  control=list(trace=2)))
print(summary(allmsk, order=value))

# Check unsuitable method trap
try(optimr(xx, simfun.f, simfun.g, method="ucminf", lower=lower, upper=upper))

Try the optimx package in your browser

Any scripts or data that you put into this service are public.

optimx documentation built on Oct. 24, 2023, 5:06 p.m.