#####################################################################
### If you want the duplicate values removed, use 
### identical = rep(0,length(z))
######################################################################
dataclean2 <- function(z, d, identical = rep(1, length(z)) ) {
       if(length(d) <= 1) {sortz <- z; sortd <- d; weight <- length(d)}
       else {
       niceorder <- order(z, - d)    # order them as z increase. When z
       sortz <- z[niceorder]         # has ties, use d=2, 1, 0 to order
       sortd <- d[niceorder]         # the tied z values.
       dupsortz <- duplicated(sortz)  #see if there is dup in z. But even if
       argdiff <- c(1, diff(sortd))  #z's tie, if d is diff, it's not a tie.
       dupsortz[argdiff != 0] <- F    #seems I need not dupsortz ==T &
       dupsortz[identical != 0] <- F  #also, do not collaps if identical !=0

       sortz <- sortz[dupsortz != T]   # get the unique values of sortz and
       sortd <- sortd[dupsortz != T]   # sortd. 

                        # Now the weight or duplicate index: w[i] 
       count <- (1:length(dupsortz))[dupsortz != T]
       weight <- diff( c(count, (1+length(dupsortz)) ) )
       } 
       list(z = sortz, d = sortd, w = weight)
}
### a possible improvement: already have weight(i) in the input