Function for simulation game in R - r

Question

I have a classic dice simulation problem, which I'm struggling to implement since I'm new with R syntax. The function (which I have called simu) works as follows:
Start with 0 points
Simulate n random draws of three six-sided dice
For each draw:
If sum of three dice >12 --> +1 point
If sum of three dice <6 --> -1 point
Otherwise (ie sum between 6 and 12):
If three dice have same number --> +5 points
Otherwise --> 0 points
Return total # of points obtained at the end of n simulations
Having tried a number of different methods I seem to be pretty close with:
simu <- function(n){
k <- 0
for(i in 1:n) {
a <- sample(y,1,replace=TRUE)
b <- sample(y,1,replace=TRUE)
c <- sample(y,1,replace=TRUE)
if ((a + b + c) > 12) {
k <- k+1
} else if ((a + b + c) < 6) {
k <- k-1
} else if ((a == b) & (b == c)) {
k <- k+5
} else k <- 0
}
return(k)
}
The problem seems to be that I am failing to iterate over new simulations (for a, b, c) for each "i" in the function.

Answer 1

I have commented the only issue I have found... The last else that always re-initialize k to 0. Instead it should have been k <- k + 0 but it does not change anything to remove it.
y <- seq(1,6) # 6-sided dice
simu <- function(n){
k <- 0
for(i in 1:n) {
a <- sample(y,1,replace=TRUE)
b <- sample(y,1,replace=TRUE)
c <- sample(y,1,replace=TRUE)
if ((a + b + c) > 12) {
k <- k+1
} else if ((a + b + c) < 6) {
k <- k-1
} else if ((a == b) & (b == c)) {
k <- k+5
} #else k <- 0
}
return(k)
}
The results look quite fine :
> simu(1000)
[1] 297
> simu(100)
[1] 38

Answer 2

If you are going to use R, then you should learn to create vectorized operations instead of 'for' loops. Here is a simulation of 1 million rolls of the dice that took less than 1 second to calculate. I am not sure how long the 'for' loop approach would have taken.
n <- 1000000 # trials
start <- proc.time() # time how long it takes
result <- matrix(0L, ncol = 6, nrow = n)
colnames(result) <- c('d1', 'd2', 'd3', 'sum', 'same', 'total')
# initial the roll of three dice
result[, 1:3] <- sample(6L, n * 3L, replace = TRUE)
# compute row sum
result[, 'sum'] <- as.integer(rowSums(result[, 1:3]))
# check for being the same
result[, 'same'] <- result[,1L] == result[, 2L] & result[, 2L] == result[, 3L]
result[, 'total'] <- ifelse(result[, 'sum'] > 12L,
1L,
ifelse(result[, 'sum'] < 6L,
-1L,
ifelse(result[, 'same'] == 1L,
5L,
0L
)
)
)
table(result[, 'total'])
-1 0 1 5
46384 680762 259083 13771
cat("simulation took:", proc.time() - start, '\n')
simulation took: 0.7 0.1 0.8 NA NA

Answer 3

I am not sure that's what you need, but you can try something like that:
# Draw the dice(s) - returns vector of length == n_dices
draw <- function(sides = 6, dices = 3){
sample(1:sides, dices, replace = T)
}
# test simulation x and return -1, 0, 1, 1 or 5
test <- function(x){
(sum(x) > 12)*1 + (sum(x) < 6)*(-1) + (sum(x) >= 6 &
sum(x) <= 12 &
var(x) == 0)*5
}
# simulate n draws of x dices with y sides
simu <- function(sides = 6, dices = 3, n = 100){
sum(replicate(n, test(draw(sides, dices))))
}
# run simulations of 100 draws for 1, 2, ..., 11, 12-side dices (3 dices each simulation)
dt <- lapply(1:12, function(side) replicate(100, simu(side, 3, 100)))
# plot dicstribution of scores
par(mfrow = c(3,4))
lapply(1:length(dt), function(i) hist(dt[[i]],
main = sprintf("%i sides dice", i),
xlab = "Score"
)
)