speedy-slice

Speedy slice sampling, as per Neal, 2003.

This implementation of the slice sampling algorithm uses lens as a means to operate over generic indexed traversable functors, so you can expect it to work if your target function takes a list, vector, map, sequence, etc. as its argument.

Additionally you can sample over anything that's an instance of both Num and Variate, which is useful in the case of discrete parameters.

Exports a mcmc function that prints a trace to stdout, a chain function for working with results in memory, and a slice transition operator that can be used more generally.

import Numeric.MCMC.Slice
import Data.Sequence (Seq, index, fromList)

bnn :: Seq Double -> Double
bnn xs = -0.5 * (x0 ^ 2 * x1 ^ 2 + x0 ^ 2 + x1 ^ 2 - 8 * x0 - 8 * x1) where
  x0 = index xs 0
  x1 = index xs 1

main :: IO ()
main = withSystemRandom . asGenIO $ mcmc 10000 1 (fromList [0, 0]) bnn

speedy-slice is a member of the declarative suite of libraries, containing a bunch of MCMC algorithms that play nicely together.