Question

Below is a simplified version of a segment of code that I’m working on (a lot of additional calculations are left out to avoid confusion). It’s just a modified form of the cumsum function. I don’t want to re-invent the wheel, so does this function already exist? If not, what scheme would provide the best speed?

#Set up the data   
set.seed(1)   
junk <- rnorm(1000000)   
junk1 <- rnorm(1000000)   
cumval <- numeric(1000000)   

#Initialize the accumulator   
cumval[1] <- 1   

#Perform the modified cumsum
system.time({   
for (i in 2:1000000) cumval[i] <- junk[i] + (junk1[i] * cumval[i-1])       
})   

#Plot the result
plot(cumval, type="l")    

Answer 1

This algorithm is something that fits the compiler package perfectly!

#Set up the data   
set.seed(1)   
junk <- rnorm(1000000)   
junk1 <- rnorm(1000000)

# The original code
f <- function(junk, junk1) {
  cumval <- numeric(1000000)
  cumval[1] <- 1
  for (i in 2:1000000) cumval[i] <- junk[i] + (junk1[i] * cumval[i-1])
  cumval
}
system.time( f(junk, junk1) ) # 4.11 secs

# Now try compiling it...
library(compiler)
g <- cmpfun(f)
system.time( g(junk, junk1) ) # 0.98 secs

…so it would be interesting to know if this algorithm is in any way “typical” – in that case the compiler could perhaps be even more optimized for situations like this…