Accumulate function that accepts a predicate for when to stop

accumulate_until.R

# inspiration: https://twitter.com/MilesMcBain/status/915448555027828737

accumulate_until <- function(.x, .f, .p, ..., .init = NULL) {
  n   <- length(.x)
  m   <- mode(.x)
  out <- vector(m, n)
  
  if(!is.null(.init)) {
    .x <- c(.init, .x)
  }
  
  out[1] <- .x[1]

  for(i in seq(2, n)) {
    y <- .f(out[i - 1], .x[i], ...)
    if(.p(y)) return(out[seq_len(i - 1)])
    out[i] <- y
  }
  
  out
}

accumulate_until(1:10, `+`, function(x) x > 10)
# [1]  1  3  6 10

transition_mc <- function(steps, start, mat, .p) {
  i <- seq_len(nrow(mat))
  transition <- function(x, y) sample(i, 1, prob = (i == x) %*% mat)
  accumulate_until(seq_len(steps), transition, .p, .init = start)
}

transition_matrix <- rbind(c(.1, .9, 0), c(.45, .45, .1), c(0, 0, 1))
transition_mc(30, 2, transition_matrix, .p = function(x) x == 3)
# e.g. [1] 2 2 2 2 2 2 1 2 2