gistfile1.clj

(ns ScriptIJ.distance-matrix
  (:import [java.awt.geom Point2D$Double]))

(defn ->point [m]
  (Point2D$Double. (Double/parseDouble (get m "XM"))
		   (Double/parseDouble (get m "YM"))))

(defn compute-matrix [table]
  (partition (count table)
	     (for [a table, b table]
	       (.distance (->point a) (->point b)))))

(defn verify-matrix [matrix]
  (loop [i 0, matrix matrix]
    (when-not (empty? matrix)
      (assert (zero? (nth (first matrix) i)))
      (recur (inc i) (rest matrix)))))

(defn allocate-bins [distance tolerance]
  (partition 2
	     (interleave (range 0 (+ distance tolerance) tolerance)
			 (range tolerance (+ distance tolerance) tolerance))))

(defn within-range
  "Return the number of items in row that is between lower (inclusive) and upper (exclusive)"
  [row [lower upper]]
  (count (filter #(and (<= lower %) (< % upper)) row)))

(def range-comparator
     (comparator (fn [[a _] [b _]] (< a b))))

(defn count-occurrances [row bins]
  (apply assoc
	 (sorted-map-by range-comparator)
	 (interleave bins (map #(within-range row %) bins))))

(defn matrix->histogram
  "Given an n-matrix return a list of length 'distance/tolerance', counting the number of entries per bin"
  [matrix distance tolerance]
  (let [bins (allocate-bins distance tolerance)]

    ;; it's better when it's parallel!
    (reduce #(merge-with + %1 %2)
	    (doall (pmap #(count-occurrances % bins) matrix)))

    ;; (loop [histogram (sorted-map-by range-comparator), matrix matrix]
    ;;   (if (empty? matrix)
    ;; 	(do (println "Done!")
    ;; 	    histogram)
    ;; 	(let [row (first matrix)
    ;; 	      occurances (count-occurrances row bins)]
    ;; 	  (recur (merge-with + histogram occurances)
    ;; 		 (rest matrix)))))
    ))

;; (matrix->histogram (partition 10 (take 1000 (repeatedly #(rand-int 100))))  100 10)