This constructs a discrete constraint satisfaction problem (CSP) and then solves it. A discrete CSP consists of a number of variables each having a discrete domain along with a number of constraints between those variables. Solving a CSP searches for assignments to the variables which satisfy those constraints. At the moment the only constraint propagation technique available is arc consistency.

Here is a simple example which solves Sudoku puzzles, project Euler problem 96.

import Data.List
import Control.Monad.CSP

solveSudoku :: (Enum a, Eq a, Num a) => [[a]] -> [[a]]
solveSudoku puzzle = oneCSPSolution $ do
  dvs <- mapM (mapM (\a -> mkDV $ if a == 0 then [1 .. 9] else [a])) puzzle
  mapM_ assertRowConstraints dvs
  mapM_ assertRowConstraints $ transpose dvs
  sequence_ [assertSquareConstraints dvs x y | x <- [0,3,6], y <- [0,3,6]]
  return dvs
      where assertRowConstraints =  mapAllPairsM_ (constraint2 (/=))
            assertSquareConstraints dvs i j = 
                mapAllPairsM_ (constraint2 (/=)) [(dvs !! x) !! y | x <- [i..i+2], y <- [j..j+2]]

 mapAllPairsM_ :: Monad m => (a -> a -> m b) -> [a] -> m ()
 mapAllPairsM_ f []     = return ()
 mapAllPairsM_ f (_:[]) = return ()
 mapAllPairsM_ f (a:l) = mapM_ (f a) l >> mapAllPairsM_ f l

sudoku3 = [[0,0,0,0,0,0,9,0,7],
           [0,0,0,4,2,0,1,8,0],
           [0,0,0,7,0,5,0,2,6],
           [1,0,0,9,0,4,0,0,0],
           [0,5,0,0,0,0,0,4,0],
           [0,0,0,5,0,7,0,0,9],
           [9,2,0,1,0,8,0,0,0],
           [0,3,4,0,5,9,0,0,0],
           [5,0,7,0,0,0,0,0,0]]
 

>>> solveSudoku sudoku3
[[4,6,2,8,3,1,9,5,7],[7,9,5,4,2,6,1,8,3],[3,8,1,7,9,5,4,2,6],[1,7,3,9,8,4,2,6,5],[6,5,9,3,1,2,7,4,8],[2,4,8,5,6,7,3,1,9],[9,2,6,1,7,8,5,3,4],[8,3,4,2,5,9,6,7,1],[5,1,7,6,4,3,8,9,2]]