The Psyleron system, unfortunately, costs hundreds of dollars and only runs on Windows, so I wrote some quick Clojure to do experimentation with randomness, using /dev/random as the source. (It appears that /dev/random on Mac OS X does not provide the same quality guarantees as it does on Linux, but it’s fine for a prototype and the implementation allows easy substitution of another randomness source later.)

Clojure random walk generator

(ns randomwalker.core
  (:use [incanter core charts]))

(defn random-walk
  "Returns a random walk of length steps. random-fn should return a random integer."
  [steps list random-fn]
  (if (= 1 steps)
    list
    (recur (dec steps)
           (conj list
                   (+
                    (if (odd? (random-fn)) 1 -1)
                    (or (last list) 0)))
           random-fn)))

(defn dev-random-walk
  "Returns a random walk of length steps based on random numbers read from /dev/random."
  [steps]
  (with-open [random (java.io.FileInputStream. "/dev/random")]
    (random-walk steps [0] (fn [] (.read random)))))

(defn view-walk
  "Visualize a random walk."
  [steps]
  (view (incanter.charts/xy-plot (range 0 (dec steps)) (dev-random-walk steps))))

(defn walk-ratio
  "Generates the given number of random walks and returns the ratio of (walks ending > 0) / (walks ending < 0)."
  [runs steps-per-run greater less]
  (if (zero? runs)
    (double (/ greater less))
    (let [final (last (dev-random-walk steps-per-run))
          ;foo (println final)
          new-greater (if (> final 0) (inc greater) greater)
          new-less (if (< final 0) (inc less) less)]
      (recur (dec runs) steps-per-run new-greater new-less))))

This code generates a simple random walk. The user can easily visualize the random walk with Incanter, and it looks suspiciously like a stock chart.

Screenshot of the Clojure Random Walker in action

Psyleron type functionality first vaguely appears in walk-ratio. This generates the requested number of random walks and returns the ratio of those that ended above zero to those that ended below. I imagine this will become the basis for some Psyleron Reflector style games and experiments.

This is probably not the most efficient way to implement this; any suggestions are welcome. Any ideas for a better source of randomness? Any insight into how the Psyleron games work? Their descriptions are a bit short on the details.