Replacing specific values in vector with different samples from another vector

Question

I have a vector "a" with integer values, some of which might have become 0 due to other parts of code that are running. I would like to replace the occurrences of 0 in this vector with a random sample from another vector "b" that I have. However, if there are multiple 0 values in "a", I would like them to all different samples from "b". So for instance:

a <- c(1, 2, 3, 0, 0, 0)
b <- 1:100

I would like the last three 0 values of "a" to be replaced with random values within "b", but I would like to avoid using 1, 2, or 3. Those are already in a.

Currently, I am using a while loop, so:

while(0 %in% a) {
  s = sample(1, b)
    while(s %in% a) {
      s = sample(1, b)
    }
  a[a==0][1] = s
}

Is there a better way to do this? it seems like this double while loop might take a long time to run.

Answer 1

You could do something like the following

indx <- which(!a) # identify the zeroes locations
# Or less golfed `indx <- which(a == 0)`
a[indx] <- sample(setdiff(b, a), length(indx)) # replace by a sample from `setdiff(b, a)`

We haven't specified replace = TRUE so the new values will be always different of each other.