hasty-hamiltonian

Speedy traversal through parameter space.

http://github.com/jtobin/hasty-hamiltonian

Version on this page:	1.3.0
LTS Haskell 21.24:	1.3.4
Stackage Nightly 2023-12-10:	1.3.4
Latest on Hackage:	1.3.4

See all snapshots hasty-hamiltonian appears in

MIT licensed by Jared Tobin

Maintained by jared@jtobin.ca

This version can be pinned in stack with:hasty-hamiltonian-1.3.0@sha256:7f6453f7e5e1c76e5ebeed67aef103136594e1633b55b1b11033f6ba2b423679,2330

Module documentation for 1.3.0

Numeric
- Numeric.MCMC
  - Numeric.MCMC.Hamiltonian

Depends on 8 packages(full list with versions):

base, kan-extensions, lens, mcmc-types, mwc-probability, pipes, primitive, transformers

Used by 1 package in lts-9.21(full list with versions):

declarative

Gradient-based traversal through parameter space.

This implementation of HMC 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.

If you don't want to calculate your gradients by hand you can use the handy ad library for automatic differentiation.

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

import Numeric.AD (grad)
import Numeric.MCMC.Hamiltonian

target :: RealFloat a => [a] -> a
target [x0, x1] = negate ((x0 + 2 * x1 - 7) ^ 2 + (2 * x0 + x1 - 5) ^ 2)

gTarget :: [Double] -> [Double]
gTarget = grad target

booth :: Target [Double]
booth = Target target (Just gTarget)

main :: IO ()
main = withSystemRandom . asGenIO $ mcmc 10000 0.05 20 [0, 0] booth