Site icon R-bloggers

Simulatin speed: GNU R vs Julia

[This article was first published on R snippets, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
Recently there is a lot of noise about Julia. I have decided to test its speed in simulation tasks on my toy Cont model. I thought I had vectorized my GNU R code pretty well, but Julia is much faster.

The model was described in my earlier posts so let us go down to a comparison:
Here is my GNU R code:

library(e1071)< o:p>

cont.run <- function(burn.in, reps, n, d, l, s) {< o:p>
  tr <- rep(0, n)< o:p>
  r <- rep(0, reps)< o:p>
  for (i in 1:reps) {< o:p>
    sig <- rnorm(1, 0, d)< o:p>
    mul <- 1< o:p>
    if (sig < 0) {< o:p>
        sig <- sig< o:p>
        mul <- 1< o:p>
    }< o:p>
    r[i] <- mul * sum(sig > tr) / (l * n)< o:p>
    tr[runif(n) < s] <- abs(r[i])< o:p>
  }< o:p>
  return(kurtosis(r[burn.in:reps]))< o:p>
}< o:p>

system.time(replicate(10,< o:p>
    cont.run(1000, 10000, 1000, 0.005, 10.0, 0.01)))

It’s execution time is a bit below 10 seconds on my laptop.

An equivalent Julia code is the following:

using Distributions< o:p>

function cont_run(burn_in, reps, n, d, l, s)< o:p>
    tr = Array(Float64, n)< o:p>
    r = Array(Float64, reps)< o:p>
    for i in 1:reps< o:p>
        aris = 0< o:p>
        sig = randn() * d< o:p>
        mul = 1< o:p>
        if sig < 0< o:p>
            sig = sig< o:p>
            mul = 1< o:p>
        end< o:p>
        for k in 1:n< o:p>
            if sig > tr[k]< o:p>
                aris += 1< o:p>
            end< o:p>
        end< o:p>
        ari = aris / (l * n)< o:p>
        r[i] = mul * ari< o:p>
        for j in 1:n< o:p>
            if rand() < s< o:p>
                tr[j] = ari< o:p>
            end< o:p>
        end< o:p>
    end< o:p>
    kurtosis(r[burn_in:reps])< o:p>
end< o:p>

n = 10< o:p>
t_start = time()< o:p>
k = Array(Float64, n)< o:p>
for i in 1:n< o:p>
    k[i] = cont_run(1000, 10000, 1000, 0.005, 10.0, 0.01)< o:p>
end< o:p>
println(time() t_start)

And on my machine it takes a bit less than 0.7 seconds to run.

So we get over tenfold speedup. This is a significant difference for simulation experiments.
I will have to dig more into Julia in the future.

To leave a comment for the author, please follow the link and comment on their blog: R snippets.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.