Function not finding an argument when vectorized in R? - r

Question

I'm wondering why my vectorized function below works fine when I use cii(peta = c(.3, .4), N = 120, df1 = 3, df2 = 116) BUT when I use cii(F.value = c(30, 40), N = 120, df1 = 3, df2 = 116) the function gives "peta" is missing?
P.S. I have made it clear in my function that when F.value is NA, function should use peta and ELSE use F.value. But why I'm getting the "peta" is missing error?
cii <- function(peta, F.value = NA, N, df1, df2, conf.level = .9){ # Upper-Level FUNCTION
ci <- Vectorize(function(peta, F.value, N, df1, df2, conf.level){ # Lower-Level FUNCTION
options(warn = -1)
q = ifelse(is.na(F.value), (-peta * df2) / ((peta * df1) - df1), F.value)
alpha = (1 - conf.level)/2
f <- function (ncp, alpha, q, df1, df2) {
abs(suppressWarnings(pf(q = q, df1 = df1, df2 = df2, ncp, lower.tail = FALSE)) - alpha)}
I <- sapply(c(alpha, 1-alpha), function(x) optimize(f, interval = c(-30, 30), alpha = x, q = q, df1 = df1, df2 = df2)[[1]])
round(data.frame(lower = I[1], upper = I[2], conf.level = conf.level, F.value = q), 6)
})
data.frame(t(ci(peta = peta, F.value = F.value, N = N, df1 = df1, df2 = df2, conf.level = conf.level)))
}
### TWO EXAMPLES OF USE: ###
cii(F.value = c(30, 40), N = 120, df1 = 3, df2 = 116) # Gives Error!!!
cii(peta = c(.3, .4), N = 120, df1 = 3, df2 = 116) # Works Fine !!!

Answer 1

Your problem is that cii() expects there to be an input for the peta argument, and you are not providing one. You have a few options:
1. Call cii() with peta = NA
cii(peta = NA, F.value = c(30, 40), N = 120, df1 = 3, df2 = 116)
# Output:
lower upper conf.level F.value
1 29.99996 29.99996 0.9 30
2 29.99996 29.99996 0.9 40
2. Re-write cii() to assign a default value to peta
e.g.
cii <- function(peta = c(.3, .4), F.value = NA, N, df1, df2, conf.level = .9){ ... }
cii(F.value = c(30, 40), N = 120, df1 = 3, df2 = 116)
# Output
lower upper conf.level F.value
1 29.99996 29.99996 0.9 30
2 29.99996 29.99996 0.9 40
As for this case:
cii(peta = c(.3, .4), N = 120, df1 = 3, df2 = 116)
You don't get an error because in your function you've already assigned a default value of NA to F.value.

Answer 2

BTW although Marcus's solution works - your error occurs because you are using Vectorize
You can call a function with a single argument if the second argument is not required for evaluation
myfun <- function(a, b) {
ifelse(a==1, print(a), print(b))
if (a==1) { print(a) } else { print(b) }
}
myfun(1)
# [1] 1
# [1] 1
myfun(2)
# Error in print(b) : argument "b" is missing, with no default
This is also true for a nested(?) (internal) function
myfun <- function(a, b) {
internalfun <- function(a, b) { ifelse(a==3, print(a), print(b)) }
if (a == 1) { print(a) } else { print(b) }
internalfun(a = 3)
}
myfun(1)
# [1] 1
# [1] 3
# [1] 3
# BTW, I do not understand why `3` is printed twice
This still works when you explicitly name the arguments when calling the nested function
myfun <- function(a, b) {
internalfun <- function(a, b) { ifelse(a==3, print(a), print(b)) }
if (a == 1) { print(a) } else { print(b) }
internalfun(a = 3, b = b)
}
myfun(1)
# [1] 1
# [1] 3
# [1] 3
But it fails when using Vectorize
myfun <- function(a, b) {
internalfun <- Vectorize(function(a, b) { ifelse(a==3, print(a), print(b)) })
if (a == 1) { print(a) } else { print(b) }
internalfun(a = 3, b = b)
}
myfun(1)
# [1] 1
# Error in FUN(X[[i]], ...) : argument "b" is missing, with no default
My guess is that when vectorizing your function, Vectorize has to 'evaluate' your arguments, which is leading to the error