module Coal where import Probability import Data.Frame import Data.Array -- When we get polymorphism, maybe we can create an temporary array as a view on xs' -- I presume that using n directly is more efficient that using [length xs'] orderedSample n dist = do xs' <- sample $ iid n dist let xs = sort xs' return $ listArray n xs get_intervals (r:[]) (t:[]) tend = [(r,t,tend)] get_intervals (r:rs) (t1:ts@(t2:_)) tend = (r,t1,t2):get_intervals rs ts tend model (t1,t2) times = do let lam = 3; nMax = 30; a = 1; b = 365/200 -- n change points, n+1 intervals n <- min nMax <$> prior (poisson lam) -- even numbered order statistics over the interval [t1,t2] s' <- orderedSample (2*n+1) (uniform t1 t2) let s = t1:[s'!(2*i-1) | i <- [1..n]] -- n+1 rates g <- prior $ iid (n+1) (gamma a b) let intervals = get_intervals g s t2 observe times $ poisson_processes intervals return [ "n" %=% n, "s" %=% s, "g" %=% g, "intervals" %=% intervals] main dirname = do frame <- readTable "coal-times.csv" let times = frame $$ "time" :: [Double] return $ model (1851, 1963) times