Why?

February 16, 2016

RANDU: The case of the bad RNG

Filed under: Computing, R — Tags: , , — csgillespie @ 12:15 pm

The German Federal Office for Information Security (BSI) has established
criteria for quality random number generator (rng):

  • A sequence of random numbers has a high probability of containing no identical consecutive elements.
  • A sequence of numbers which is indistinguishable from true random’ numbers (tested using statistical tests.
  • It should be impossible to calculate, or guess, from any given sub-sequence, any previous or future values in the sequence.
  • It should be impossible, for all practical purposes, for an attacker to calculate, or guess the values used in the random number algorithm.

Points 3 and 4 are crucial for many applications. Everytime you make a
phone call, contact to a wireless point, pay using your credit card random
numbers are used.

Designing a good random number generator is hard and as a general rule you should never try to. R comes with many good quality random generators. The default generator is the Mersenne-Twister. This rng has a huge period of 2^{19937}-1 (how many random numbers are generated before we have a repeat).

Linear congruential generators

A linear congruential generator (lcg) is a relatively simple rng (popular in the 60’s and 70’s). It has a simple form of

r_{i}=(ar_{i-1}+b) \textrm{ mod }m, \quad i=1, 2, \ldots, m

where $latexr_0$ is the initial number, known as the seed, and \(a,b,m\) are the multiplier, additive constant and modulo respectively. The parameters are all integers.

The modulo operation means that at most m different numbers can be generated
before the sequence must repeat – namely the integers 0,1,2, \ldots, m-1. The
actual number of generated numbers is h \leq m, called the period of
the generator.

The key to random number generators is in setting the parameters.

RANDU

RANDU was a lcg with parameters m=2^{31}, a=65539 and b=0. Unfortunately this is a spectacularly bad choice of
parameters. On noting that a=65,539=2^{16}+3, then

r_{i+1} = a r_i = 65539 \times r_i = (2^{16}+3)r_i \;.

So

r_{i+2} = a\;r_{i+1} = (2^{16}+3) \times r_{i+1} = (2^{16}+3)^2 r_i \;.

On expanding the square, we get

r_{i+2} = (2^{32}+6\times 2^{16} + 9)r_i = [6 (2^{16}+3)-9]r_i = 6 r_{i+1} - 9 r_i \;.

Note: all these calculations should be to the mod 2^{31}. So there is a large
correlation between the three points!

If compare randu to a standard rng (code in a gist)

Rplot1

It’s obvious that randu doesn’t produce good random numbers. Plotting  x_i, x_{i-1} and x_{i-2} in 3d

Rplot2

Generating the graphics

The code is all in a gist and can be run via


devtools::source_gist("https://gist.github.com/csgillespie/0ba4bbd8da0d1264b124")

You can then get the 3d plot via


scatterplot3d::scatterplot3d(randu[,1], randu[,2], randu[,3],
angle=154)
## Interactive version
threejs::scatterplot3js(randu[,1], randu[,2], randu[,3])

Advertisements

4 Comments »

  1. Yeah, I ran across randu in code that was *written* in the early 1990s and still in use in 2014 at a certain large US defense contractor. All this even though the FORTRAN community had made it clear,in the early 1960s, that RANDU was not to be used. People just don’t learn.

    Comment by Carl Witthoft — February 17, 2016 @ 5:51 pm

    • Wow, why on earth would anyone want to use it??

      Comment by csgillespie — February 17, 2016 @ 7:24 pm

      • @csgillespie – because nobody is stodgier or less able to adjust to changes than a stack of engineers who grew up in the slide rule era and never saw a reason to change. Sad but true.

        Comment by Carl Witthoft — February 18, 2016 @ 1:54 pm

  2. Look up George Marsaglia and his DIEHARD battery of tests for RNG. They’ve been updated a bit since then but they form a solid basis for RNG testing.

    Comment by Mark Ferguson — February 18, 2016 @ 1:06 am


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at WordPress.com.

%d bloggers like this: