Ways to find the N minimum value in a vector

findmins.R

#Looking for a fast way to find the N minimum values in a vector of values.  For use in finding nearest neighbors when one already has a (large) distance matrix.

x <- runif(5000)
n <- 3

system.time({              # What I'm using now
  for(i in 1:10000) {
    a <- sort(x)[1:n]
    b <- match(a, x)
  }
})

# This saves a quarter of a second.  Can you vectorize your 10000 replicates?

system.time({            
  for(i in 1:10000) {
    order(x)[seq.int(n)]
  }
})

# Edit: this seems like it might even be slightly faster:

values = seq.int(n)

system.time({            
  for(i in 1:10000) {
    a = .Internal(order(na.last = TRUE, decreasing = FALSE, x))[values]
  }
})