dpmem.md

layout	title	model-status	model-category	model-tags	model-language
model	Dirichlet Process	code	Nonparametric Models	mem, nonparametrics	church

Using memoization (mem), we can implement the Dirichlet process:

(define (pick-a-stick sticks J)
  (if (flip (sticks J))
      J
      (pick-a-stick sticks (+ J 1))))

(define (make-sticks alpha)
  (let ((sticks (mem (lambda (x) (beta 1.0 alpha)))))
    (lambda () (pick-a-stick sticks 1))))

(define my-sticks (make-sticks 1))

(hist (repeat 1000 my-sticks) "Dirichlet Process")

Based on the Dirichlet Process, we can write a stochastic memoizer for any function (DPmem):

(define (pick-a-stick sticks J)
  (if (flip (sticks J))
      J
      (pick-a-stick sticks (+ J 1))))

(define (make-sticks alpha)
  (let ((sticks (mem (lambda (x) (beta 1.0 alpha)))))
    (lambda () (pick-a-stick sticks 1))))

(define (DPmem alpha base-dist)
  (let ((augmented-proc
          (mem (lambda (args stick-index) (apply base-dist args))))
        (DP (mem (lambda (args) (make-sticks alpha)))))
    (lambda argsin
      (let ((stick-index ((DP argsin))))
        (augmented-proc argsin stick-index)))))

(define memoized-gaussian (DPmem 1.0 gaussian))

(hist (repeat 1000 (lambda () (memoized-gaussian 0.0 1.0))) "Dirichlet Process")

References:

• Cite:Ferguson1973
• Cite:Goodman2008uq
• Cite:ProbMods