Generate list of vectors with limited overlap. Useful for randomly assigning individuals from a pool to subsets (such as experimental blocks) with limited similarity between subsets..

setCompositionRandomizer.R

#generate random number sets with limited overlap
#3/28/2018
#with em-bellis

####generate random integers from a pool####
#from 1-60, sample 10
sample.int(60, 10) 
#[1] 34 42 45 54  3  5 20 25 53 40
#such as IDs of populations or individuals from a list

#pairwise comparisons
set1 <- sample.int(60,10)
set2 <- sample.int(60,10)

set1 %in% set3
#comparison gives logical vector
#[1]  TRUE  TRUE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE

####function to check matching/add to list####
#first make list containing starter set
plist <- list(set1=set1)

#defaults generate set of 10 IDs from a pool of 60 possible IDs with an overlap between any two sets of 4 or less
generate_set <- function(setList, pool=60, size=10, overlap=4) {
  testset <- sample.int(pool,size)
  for(i in 1:length(setList)) {
    n <- testset %in% setList[[i]]
    if (sum(n, na.rm=TRUE)>overlap) {
      fail <- TRUE
    } 
  }
  if (fail==FALSE) {return(testplot)}
}

####while loop to run enough times to generate the required number of set assignments####
#here, produces list of 20 sets
while (length(plist) < 20) {
  fail <- FALSE
  plist <- c(plist, list(generate_plot(plist)))
}