Below are two examples.
(load "load")
;;;; set up a simple discrete test
(define (test1)
(let ((x (discrete '(0 1)))
(y (discrete '(0 1)))
(z (discrete '(0 1))))
(let ((sum (+ x y z)))
(emit (or (= sum 0)
(= sum 2))
#t)
(list x y z))))
;;;; inspect a few samples
(test1)
;Value 2: (1 1 0)
(test1)
;Value 3: (0 1 1)
(test1)
;Value 4: (0 0 0)
;;;; construct a stream of samples
(define ss (sample-stream test1 0 5))
(stream-head ss 10)
;Value: ((0 0 0) (0 0 0) (0 0 0) (1 1 0) (1 1 0) (1 1 0) (0 1 1) (0 1 1) (1 0 1) (1 1 0))
;;;; estimate the probability all are zero
(define (all-zero? s)
(if (= (reduce + 0 s) 0)
1
0))
(estimate-mean (stream-map all-zero? ss) 2000)
;Value: 103/400
(estimate-mean (stream-map all-zero? ss) 4000)
;Value: 1/4
;;;; the true probability is 1/4(load "load")
;;;; set up a simple Gaussian test
(define (test3)
(let ((x (gaussian 0 1))
(y (gaussian 0 4)))
(emit (+ x y) 3 (likelihood:additive-gaussian 0 1))
y))
;;;; inspect a few samples
(test3)
;Value: 2.627738902766393
(test3)
;Value: 1.4713636283192144
(test3)
;Value: 3.650811672098934
(test3)
;Value: 2.508292233153293
;;;; construct a stream of samples from the Markov chain
(define ss (sample-stream test3 5000 25))
(stream-head ss 5)
;Value: (2.362360472725903 2.457989361328277 .9664606900582463 2.245335036134793 2.241072726252283)
;;;; estimate the mean
(estimate-mean ss 2000)
;Value: 2.0789872746395206
(estimate-mean ss 3000)
;Value: 2.0538287789004213
;;;; the true posterior mean is 2