Infinite recursion error - how to solve?

Question

Still new to R so trying to practice working with functions, I have created some (probably quite inefficient) code which I will look at modifying later, currently it returns the result I need but I want to repeat the simulation 100 times within the function itself. This would effectively be turn[i] which returns outcomes -1 or 1 with a specified function probability:

game = function(n,pr) {
turn = cumsum(2*rbinom(n,1,prob=pr)-1)
bankrupcy.test = which(turn == -25)
winner.test = which(turn == 50)
if(length(bankrupcy.test)==0){bankrupcy.test=c(0)}
if(length(winner.test)==0){winner.test=c(0)}
if(bankrupcy.test==0 && winner.test==0){turn[n]}else
if(bankrupcy.test[1]>winner.test[1]){-25}else{
50}
return(replicate(100,game(n)))
}

I have tried creating a for loop but I can't seem to structure that correctly, hence I am looking to use the replicate command within the function I created, however I recieve the following error:

"evaluation nested too deeply: infinite recursion / options(expressions=)? Error during wrapup: evaluation nested too deeply: infinite recursion /options(expressions=)"

Where am I going wrong? I want to return a vector with 100 outcomes of the above simulation, in which the game is played until 50 profit is made or a 25 loss is made, whichever occurs first. If neither of the previous outcomes occur then the final value of the vector is taken.

You are calling game from inside the function, and so game ends up calling itself. This is causing the recursion. — Oliver Frost
I figured that might be the case but how else can I replicate the simulation within the function? Is the only way a for loop? I can never seem to get those to work if the function gets even slightly complicated... — Aesler
is there a reason you want the replication within the function? if you remove the return(replicate(100,game(n))) and then outside the function run replicate(100, game(n = n, pr = pr)) it should work (but maybe I don't understand what you are trying to do) — Lucy
I want to add to the function further, so there will be additional factors which are reliant on the final vector being of 100 simulations (or more), so I have a sample of data to return information about. I will expand the function to several more variables (function(x,y,z....) and with only one result in the vector it isn't possible to do this as of yet. — Aesler
Can you give an example input? I.e. some values for n and pr when you call the function for the first time? — Oliver Frost

Oliver Frost Oliver Frost · Accepted Answer · 2017-03-10T23:53:39

Try this:

n <- 1000
pr <- 0.50

game <- function(n, pr) {
  result = 
    replicate(100, {{turn = cumsum(2*rbinom(n, 1,prob=pr)-1)
      bankrupcy.test = which(turn == -25)
      winner.test = which(turn == 50)
      if(length(bankrupcy.test)==0){bankrupcy.test = 0}
      if(length(winner.test)==0){winner.test = 0}
      if(bankrupcy.test==0 && winner.test==0){turn[n]} else
        if(bankrupcy.test[1]>winner.test[1]) {-25} else {50}
      }})
  return(result)
}

game(n, pr)

Infinite recursion error - how to solve?

1 Answers