npoptimal <- function(r,theta,n,B,rseed){
	set.seed(rseed)
    select=rep(0,length(r))
    j=1
    while(j<=B){
      latent=runif(n,0,1)
      tox=matrix(nrow=n,ncol=length(r))
      i=1
      while(i<=n){
        tox[i,]<-as.numeric(latent[i]<=r)
        i=i+1
      }
      loss=abs(colMeans(tox)-theta)
      sugglev=which(loss==min(loss))
      mtd=ifelse(length(sugglev)==1,sugglev,sample(sugglev,1))
      select[mtd]=select[mtd]+1
      j=j+1
    }
    cat("True DLT probability: ", round(r,3), sep="    ",  "\n");
    cat("Selection percentage: ", formatC((select/B)*100, digits=1, format="f"), sep="    ",  "\n");
    cat("Accuracy Index:       ", round(1-length(r)*(sum(abs(r-theta)*(select/B))/sum(abs(r-theta))),4), sep="    ",  "\n");
  }


##Assumed true toxicity scenario
truth=c(0.05,0.07,0.20,0.35,0.55,0.70)

####################################
#
#	Input:
#
#	r = assumed true toxicity probabilities 
#	n = sample size
#	theta = target toxicity rate
#	B = number of simulated trials
# 
####################################
theta=0.20
n=25
B=2000
r=truth
rseed=580
npoptimal(r,theta,n,B,rseed)