Function inputs from a list - r

Question

How can I run a function (in R) where some of the inputs are pulled from a list (or data frame)? Am I right in thinking that this would be more efficient than running a for-loop?
I am running simulations and want to change the variable values, but as they take a long time to run I want them to run overnight and to just tick through the different values automatically.
Here's the code for the function:
n = 10000
mu = 0
sd = 1
n.sub = 100
iboot = 100
isim = 1000 ### REDUCED FOR THIS EXAMPLE ###
var.values <- NULL
var.values.pop <- NULL
hist.fn <- function(n,mu,sd,n.sub,iboot)
{
Pop <- rnorm(n,mu,sd)
var.pop <- var(Pop)
Samp <- sample(Pop, n.sub, replace = FALSE)
var.samp <- var(Samp)
for(i in 1:isim) {
for(j in 1:iboot) {
Boot <- sample(Samp, n.sub, replace = TRUE)
var.values[j] <- var(Boot)
}
Samp <- sample(Pop, n.sub, replace = FALSE)
var.values.pop[i] <- var(Samp)
}
hist.pop <- hist(var.values.pop,plot=F)
hist.boot <- hist(var.values,plot=F)
#mypath = file.path("C:", "Output", paste("hist.boot_n.", n.sub, "_var.", sd^2, "_isim.", isim, "_iboot.", iboot, ".wmf", sep=""))
#win.metafile(file=mypath)
plot.new() #### ADDED FOR THIS EXAMPLE INSTEAD OF OUTPUTTING TO FILE ####
plot(hist.pop, freq=FALSE, xlim=range(var.values.pop, var.values), ylim=range(hist.pop$density, hist.boot$density), main = paste("Histogram of variances \n n=",n.sub," mu=",mu,"var=",sd^2,"\n n.sim=",isim,"n.boot=",iboot,"\n"), cex.main=0.8, xlab="Variance", col="red")
plot(hist.boot, freq=FALSE, col="blue", border="blue", add=T, density=20, angle=45)
abline(v=var.pop, lty=2, col="black", lwd=2)
legend("topright", legend=c("sample","bootstrap"),col=c("red","blue"),lty=1,lwd=2,bty="n",cex=0.7)
#dev.off()
}
hist.fn(n,mu,sd,n.sub,iboot)
Then I want sd, n.sub, and iboot to change by running through the following values:
sd <- c(1,10,100,1000)
n.sub <- c(4,10,100,1000)
iboot <- c(100,1000,10000)

Answer 1

Perhaps something like this?
n = 10000
mu = 0
sd = 1
n.sub = 100
iboot = 100
isim = 1000
sd <- c(1,10,100,1000)
n.sub <- c(4,10,100,1000)
iboot <- c(100,1000,10000)
# hist.fn parameters: n,mu,sd,n.sub,iboot
params <- expand.grid(n = n, mu = mu, sd = sd,
n.sub = n.sub, iboot = iboot)
apply(params, 1, FUN = function(x) do.call(hist.fn, as.list(x) ) )
You probably want to put these:
var.values <- NULL
var.values.pop <- NULL
Inside hist.fn, because assigning values to variables outside a function doesn't work like you seem to think.

Answer 2

You should use do.call, which will apply the function using arguments in a list. I have simplified your example to run less loops for the example. You can modify the printline of the script in order to monitor your progress for a larger job:
# The function
hist.fn <- function(n,mu,isim,sd,n.sub,iboot)
{
Pop <- rnorm(n,mu,sd)
var.pop <- var(Pop)
Samp <- sample(Pop, n.sub, replace = FALSE)
var.samp <- var(Samp)
var.values <- NaN*seq(isim) # sets up an empty vector for results
var.values.pop <- NaN*seq(isim) # sets up an empty vector for results
for(i in seq(isim)) {
for(j in seq(iboot)) {
Boot <- sample(Samp, n.sub, replace = TRUE)
var.values[j] <- var(Boot)
print(paste("i =", i, "; j =", j))
}
Samp <- sample(Pop, n.sub, replace = FALSE)
var.values.pop[i] <- var(Samp)
}
list(var.values=var.values, var.values.pop=var.values.pop) #returns results in the form of a list
}
# Global variables
n = 100
mu = 0
isim = 10
# Changing variables
sd <- c(1,10,20,30)
n.sub <- c(4,10,20,30)
iboot <- c(100,200,300,400)
df <- data.frame(sd=sd, n.sub=n.sub, iboot=iboot)
res <- vector(mode="list", nrow(df)) # sets up an empty list for results
for(i in seq(nrow(df))){
res[[i]] <- do.call(hist.fn, c(n=n, mu=mu, isim=isim, df[i,]) )
}
res # show results

Answer 3

sd <- 1:3
n.sub <- 4:6
iboot <- 7:9
funct1<-function(x,y,z) print(x+y+z)
for (i in 1:length(sd)){
funct1(sd[i],n.sub[i],iboot[i])
}
just an example. Doing it with loop.