dirichlet.clj

;; Probability Distribution
;; counter: counts of objects
;; lambda: smoothing constants
;; num-keys: number of possible keys, needed to normalize
(defrecord DirichletMultinomial [counter lambda num-keys])

(defn new-dirichlet [lambda num-keys]
  (DirichletMultinomial. (Counter. {} 0) lambda num-keys))

(defn log-prob 
  "log prob. from a DirichletMultinomial"  
  [distr key]                                                
  {:post [(> % Double/NEGATIVE_INFINITY)]}
  (let [{:keys [counter lambda num-keys]} distr]        
    (Math/log (/ (+ (get-count counter key) lambda) 
                 (+ (:total counter) (* lambda num-keys))))))

(defn observe 
 "make an observation to a DirichletMultinomial"
 [distr key weight]
 (let [{:keys [counter,total,lambda,num-keys]} distr]  
  (DirichletMultinomial. (inc-count counter key weight) lambda num-keys)))