CODEWITHSUNDEEP
rfunction
moth
moth

Reputation: 2389

equivalent of *args in R and summing up a function to itself

Say I have a binomial.coeff function defined as:

combn <- function(n,r) {
  factorial(n)/(factorial(r)*factorial(n-r))
}

binomial.coeff <- function(n,x,p) {
  combn(n,x)*p^x*(1-p)^(n-x)
}

I would like to call the function binomial.coeff for different values of x and sum the calls to produce the result. I'm not sure if R takes many positional arguments like this:

binomial.coeff <- function(n,...,p) {
    combn(n,...)*p^...*(1-p)^(n-...)
}

My final goal is say x takes values [3,2] then I would like to have the function being summed on 3 and 2

binomial.coeff <- function(n,x,p) {
  combn(n,3)*p^3*(1-p)^(n-3)+combn(n,2)*p^2*(1-p)^(n-2)
}

Upvotes: 0

Views: 211

Answers (1)

MrFlick
MrFlick

Reputation: 206466

Since you are using functions that are themselves vectorized, you can pass in a vector for x with

binomial.coeff(3,2:3,.25)
# [1] 0.140625 0.015625

and sum that with

sum(binomial.coeff(3,2:3,.25))
# [1] 0.15625

More generally, if you want to call it multiple times with different parameter values, in R you'd use something like sapply. For example

sapply(2:3, function(x) binomial.coeff(3, x, .25))
# [1] 0.140625 0.015625
sum(sapply(2:3, function(x) binomial.coeff(3, x, .25)))
# [1] 0.15625

But the code you've written seems to be the pmf for the binomial distribution. There is a built in function for that dbinom You could do the same thing with

dbinom(2:3, 3, .25)
# [1] 0.140625 0.015625
sum(dbinom(2:3, 3, .25))
# [1] 0.15625

in one call because dbinom is vectorized over the x argument. Also this version would be more efficient.

Upvotes: 2

Related Questions