R, coding a discontinuous/interval function within a function - r

Question

I'm new to R, and I'm trying to code a function which requires it only chooses values in a certain interval, so I have decided to go with k=1 if it lies in [lower, upper] and 0 if it lies elsewhere (where lower and upper have been defined earlier in the function. However, when I try to assign values to the function, it always comes back with this
myfun(10,0.5,0.05)
#Error in k[i] <- function(p) ifelse(p >= lower & p <= upper, 1, 0) :
# incompatible types (from closure to double) in subassignment type fix
I don't really know what this means, I've tried finding an answer, but most pages just say how to fix their particular problem rather than saying what it actually means. Maybe I haven't been looking hard enough, and I apologise if I haven't, but any help would be greatly appreciated. Here is the full function, if it would help:
myfun <- function(a, q, m) {
k <- rep(0,a+1)
bin.prob <- rep(0,a+1)
for (i in 1:(a+1)) {
x <- i-1
qhat <- x/a
z <- qnorm(1-m/2)
upper <- qhat+(z*sqrt(qhat*(1-qhat)*(a^-1)))
lower <- qhat-(z*sqrt(qhat*(1-qhat)*(a^-1)))
k[i] <- function(q) ifelse(q>=lower & q<=upper, 1, 0)
bin.prob[i] <- dbinom(x,a,q)
}
C <- sum(k*bin.prob)
return(C)
}
myfun(10,0.5,0.05)
#Error in k[i] <- function(q) ifelse(q >= lower & q <= upper, 1, 0) :
# incompatible types (from closure to double) in subassignment type fix
NEW PROBLEM
Hey, I'm encountering a new problem when trying to adjust this function when trying to adjust the data set, i.e a becomes a+4 and x becomes x+2
> myfun2 <- function(a,q,m) {
+ fn <- function(a) a+4
+ abar <- fn(a)
+ kadj <- rep(0,abar+1)
+ bin.prob.adj <- rep(0,abar+1)
+ for (j in 1:(abar+1)) {
+ x <- j-1
+ fx <- function(x) x+2
+ xbar <- fx(x)
+ qhatadj <- xbar/abar
+ z <- (1-(m/2))
+ upperadj <- qhatadj+(z*sqrt(qhatadj*(1-qhatadj)*(abar^-1)))
+ loweradj <- qhatadj-(z*sqrt(qhatadj*(1-qhatadj)*(abar^-1)))
+ kadj[j] <- q>=loweradj & q<=upperadj
+ bin.prob.adj[j] <- dbinom(xbar,abar,q)
+ }
+ D <- sum(kadj*bin.prob.adj)
+ return(D)
+ }
> myfun2(10,0.5,0.05)
[1] NA
Warning messages:
1: In sqrt(qhatadj * (1 - qhatadj) * (abar^-1)) : NaNs produced
2: In sqrt(qhatadj * (1 - qhatadj) * (abar^-1)) : NaNs produced
3: In sqrt(qhatadj * (1 - qhatadj) * (abar^-1)) : NaNs produced
4: In sqrt(qhatadj * (1 - qhatadj) * (abar^-1)) : NaNs produced
I've been trying to find an answer as to why this has arised, and have found that the NaNs warning could mean there is a negative square root? However I can't see why that would have arisen. It may be bad coding on my part, or could be something else entirely (I'm new to R). Thanks for any help.

Answer 1

As the error message suggests, the problem starts at the line:
k[i] <- function(q) ifelse(q >= lower & q <= upper, 1, 0)
In the line above you are assigning a function function(q) ifelse(q >= lower & q <= upper, 1, 0) to each element of the vector k, when you really want to be assigning the result of evaluating this function on the scalar q given as an argument to the original function. Note also that the closure function(q) has an environment separate from that of the function in which it is defined. It must be explicitly called with an argument in order for it to evaluate. Hence, when you hit the line:
C <- sum(k * bin.prob)
R tries to multiply the function function(q) itself by bin.prob, throwing an error, when what you want to be doing is multiplying the result of evaluating function(q) for the scalar q defined in the arguments to the original function. In this case, there appears to be no need for you to define function(q) at all. The assignment can be replaced with:
k[i] <- ifelse(q >= lower & q <= upper, 1, 0)
Since R coerces logical vectors to numeric vectors where necessary, treating TRUE as 1 and FALSE as 0, the above assigment can be expressed more succinctly as:
k[i] <- q >= lower & q <= upper