Painless general-purpose sampling.


Version on this page:1.5.0@rev:1
LTS Haskell 20.26:1.5.2
Stackage Nightly 2022-11-17:1.5.2
Latest on Hackage:1.5.2

See all snapshots flat-mcmc appears in

MIT licensed by Jared Tobin
Maintained by jared@jtobin.ca
This version can be pinned in stack with:flat-mcmc-1.5.0@sha256:d689eaad9d5ab6468b3300791d0e4e1b4287c312fc23a0b634dd1b46f4f653de,3004

Module documentation for 1.5.0

flat-mcmc is a Haskell library for painless, efficient, general-purpose sampling from continuous distributions.

flat-mcmc uses an ensemble sampler that is invariant to affine transformations of space. It wanders a target probability distribution's parameter space as if it had been "flattened" or "unstretched" in some sense, allowing many particles to explore it locally and in parallel.

In general this sampler is useful when you want decent performance without dealing with any tuning parameters or local proposal distributions.

flat-mcmc exports an mcmc function that prints a trace to stdout, as well as a flat transition operator that can be used more generally.

import Numeric.MCMC.Flat
import qualified Data.Vector.Unboxed as U (unsafeIndex)

rosenbrock :: Particle -> Double
rosenbrock xs = negate (5  * (x1 - x0 ^ 2) ^ 2 + 0.05 * (1 - x0) ^ 2) where
  x0 = U.unsafeIndex xs 0
  x1 = U.unsafeIndex xs 1

origin :: Ensemble
origin = ensemble [
    particle [negate 1.0, negate 1.0]
  , particle [negate 1.0, 1.0]
  , particle [1.0, negate 1.0]
  , particle [1.0, 1.0]

main :: IO ()
main = withSystemRandom . asGenIO $ mcmc 12500 origin rosenbrock