1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
|
module Probability.Distribution.Bernoulli where
import Probability.Random
import MCMC
foreign import bpcall "Distribution:" sample_bernoulli :: Double -> IO Int
-- Could we allow this to return Int, Integer, Bool, etc?
-- Could we allow this to store the probability in an arbitrary Fractional or Pow type?
-- Should we put the result into the distribution class, a la (Dist Bernoulli Int)?
-- we could use fromRational... but converting to rationals seems like a terrible idea.
data Bernoulli = Bernoulli Prob
instance Dist Bernoulli where
type Result Bernoulli = Int
dist_name _ = "bernoulli"
instance IOSampleable Bernoulli where
sampleIO (Bernoulli p) = sample_bernoulli $ toFloating $ p
instance HasPdf Bernoulli where
pdf (Bernoulli p) n | n == 0 = toFloating $ 1-p
| n == 1 = toFloating $ p
| otherwise = 0
instance Dist1D Bernoulli where
cdf (Bernoulli p) n | n < 0 = 0
| n == 1 = toFloating $ 1-p
| otherwise = 1
lower_bound _ = Just 0
upper_bound _ = Just 1
instance MaybeMean Bernoulli where
maybeMean (Bernoulli p) = Just $ toFloating p
instance Mean Bernoulli
instance MaybeVariance Bernoulli where
maybeVariance (Bernoulli p) = Just $ toFloating $ p * (1 - p)
instance Variance Bernoulli
instance HasAnnotatedPdf Bernoulli where
annotated_densities dist@(Bernoulli p) n = do
in_edge "p" p
return ([pdf dist n],())
instance Sampleable Bernoulli where
sample dist = RanDistribution2 dist bernoulli_effect
bernoulli_effect x = addMove 1 $ discreteUniformAvoidMH x 0 1
bernoulli :: Double -> Bernoulli
bernoulli p = Bernoulli (toFloating p)
rbernoulli q = Bernoulli (1-toFloating(q))
|
