I am new to Haskell.
I have this code (my solution to one of the exercise from Ninety-Nine Haskell Problems)
data Structure a = Single a | Multiple (a, Int) deriving (Show)
encodeM ::(Eq a)=> [a]->[Structure a]
encodeM l = map(\x -> (let size = length x
--h = head x
in if size>1 then Multiple ( head x, size) else Single (head x)
)
) $ group l
When I uncomment "-h = head x" I get: "parse error on input `='"
But
xxx l= let size = length l
h = head l
in size
works fine, why it doesn't compile when I use "let" with multiple statement inside the lambda?
I have tried to replace let by where
encodeM2 ::(Eq a)=> [a]->[Structure a]
encodeM2 l = map(\x->if si>1 then Multiple ( head x, si) else Single (head x)
where si = length x)
but it doesn't compile as well, whats wrong with it?
This is your code properly indented: (note how the let bindings align vertically)
encodeM :: Eq a => [a] -> [Structure a]
encodeM l = map (\x -> let size = length x
h = head x in
if size > 1
then Multiple (h, size)
else Single h) $
group l
This is your code readable:
encodeM :: Eq a => [a] -> [Structure a]
encodeM = map runLength . group
where
runLength x =
let size = length x
h = head x in
if size > 1
then Multiple (h, size)
else Single h
This is your code idiomatic:
encodeM :: Eq a => [a] -> [Structure a]
encodeM = map runLength . group
where
runLength [x] = Single x
runLength xs = Multiple (head xs, length xs)
I prefer to use pattern matching to if/then/else, so your code becomes:
encodeM :: (Eq a) => [a] -> [Structure a]
encodeM lst = map fun $ group lst
where
fun [x] = Single x
fun l = Multiple (head l, length l)
In Haskell whitespace matters.
Align assignemnts in your let. And you can't use where in lambda.
Related
I have an algorithm for parallel sorting a list of a given length:
import Control.Parallel (par, pseq)
import Data.Time.Clock (diffUTCTime, getCurrentTime)
import System.Environment (getArgs)
import System.Random (StdGen, getStdGen, randoms)
parSort :: (Ord a) => [a] -> [a]
parSort (x:xs) = force greater `par` (force lesser `pseq`
(lesser ++ x:greater))
where lesser = parSort [y | y <- xs, y < x]
greater = parSort [y | y <- xs, y >= x]
parSort _ = []
sort :: (Ord a) => [a] -> [a]
sort (x:xs) = lesser ++ x:greater
where lesser = sort [y | y <- xs, y < x]
greater = sort [y | y <- xs, y >= x]
sort _ = []
parSort2 :: (Ord a) => Int -> [a] -> [a]
parSort2 d list#(x:xs)
| d <= 0 = sort list
| otherwise = force greater `par` (force lesser `pseq`
(lesser ++ x:greater))
where lesser = parSort2 d' [y | y <- xs, y < x]
greater = parSort2 d' [y | y <- xs, y >= x]
d' = d - 1
parSort2 _ _ = []
force :: [a] -> ()
force xs = go xs `pseq` ()
where go (_:xs) = go xs
go [] = 1
randomInts :: Int -> StdGen -> [Int]
randomInts k g = let result = take k (randoms g)
in force result `seq` result
testFunction = parSort
main = do
args <- getArgs
let count | null args = 500000
| otherwise = read (head args)
input <- randomInts count `fmap` getStdGen
start <- getCurrentTime
let sorted = testFunction input
putStrLn $ "Sort list N = " ++ show (length sorted)
end <- getCurrentTime
putStrLn $ show (end `diffUTCTime` start)
I want to get the time to perform parallel sorting on 2, 3 and 4 processor cores less than 1 core.
At the moment, this result I can not achieve.
Here are my program launches:
1. SortList +RTS -N1 -RTS 10000000
time = 41.2 s
2.SortList +RTS -N3 -RTS 10000000
time = 39.55 s
3.SortList +RTS -N4 -RTS 10000000
time = 54.2 s
What can I do?
Update 1:
testFunction = parSort2 60
Here's one idea you can play around with, using Data.Map. For simplicity and performance, I assume substitutivity for the element type, so we can count occurrences rather than storing lists of elements. I'm confident that you can get better results using some fancy array algorithm, but this is simple and (essentially) functional.
When writing a parallel algorithm, we want to minimize the amount of work that must be done sequentially. When sorting a list, there's one thing that we really can't avoid doing sequentially: splitting up the list into pieces for multiple threads to work on. We'd like to get that done with as little effort as possible, and then try to work mostly in parallel from then on.
Let's start with a simple sequential algorithm.
{-# language BangPatterns, TupleSections #-}
import qualified Data.Map.Strict as M
import Data.Map (Map)
import Data.List
import Control.Parallel.Strategies
type Bag a = Map a Int
ssort :: Ord a => [a] -> [a]
ssort xs =
let m = M.fromListWith (+) $ (,1) <$> xs
in concat [replicate c x | (x,c) <- M.toList m]
How can we parallelize this? First, let's break up the list into pieces. There are various ways to do this, none of them great. Assuming a small number of capabilities, I think it's reasonable to let each of them walk the list itself. Feel free to experiment with other approaches.
-- | Every Nth element, including the first
everyNth :: Int -> [a] -> [a]
everyNth n | n <= 0 = error "What you doing?"
everyNth n = go 0 where
go !_ [] = []
go 0 (x : xs) = x : go (n - 1) xs
go k (_ : xs) = go (k - 1) xs
-- | Divide up a list into N pieces fairly. Walking each list in the
-- result will walk the original list.
splatter :: Int -> [a] -> [[a]]
splatter n = map (everyNth n) . take n . tails
Now that we have pieces of list, we spark threads to convert them to bags.
parMakeBags :: Ord a => [[a]] -> Eval [Bag a]
parMakeBags xs =
traverse (rpar . M.fromListWith (+)) $ map (,1) <$> xs
Now we can repeatedly merge pairs of bags until we have just one.
parMergeBags_ :: Ord a => [Bag a] -> Eval (Bag a)
parMergeBags_ [] = pure M.empty
parMergeBags_ [t] = pure t
parMergeBags_ q = parMergeBags_ =<< go q where
go [] = pure []
go [t] = pure [t]
go (t1:t2:ts) = (:) <$> rpar (M.unionWith (+) t1 t2) <*> go ts
But ... there's a problem. In each round of merges, we use only half as many capabilities as we did in the previous one, and perform the final merge with just one capability. Ouch! To fix this, we'll need to parallelize unionWith. Fortunately, this is easy!
import Data.Map.Internal (Map (..), splitLookup, link)
parUnionWith
:: Ord k
=> (v -> v -> v)
-> Int -- Number of threads to spark
-> Map k v
-> Map k v
-> Eval (Map k v)
parUnionWith f n t1 t2 | n <= 1 = rseq $ M.unionWith f t1 t2
parUnionWith _ !_ Tip t2 = rseq t2
parUnionWith _ !_ t1 Tip = rseq t1
parUnionWith f n (Bin _ k1 x1 l1 r1) t2 = case splitLookup k1 t2 of
(l2, mb, r2) -> do
l1l2 <- parEval $ parUnionWith f (n `quot` 2) l1 l2
r1r2 <- parUnionWith f (n `quot` 2) r1 r2
case mb of
Nothing -> rseq $ link k1 x1 l1l2 r1r2
Just x2 -> rseq $ link k1 fx1x2 l1l2 r1r2
where !fx1x2 = f x1 x2
Now we can fully parallelize bag merging:
-- Uses the given number of capabilities per merge, initially,
-- doubling for each round.
parMergeBags :: Ord a => Int -> [Bag a] -> Eval (Bag a)
parMergeBags !_ [] = pure M.empty
parMergeBags !_ [t] = pure t
parMergeBags n q = parMergeBags (n * 2) =<< go q where
go [] = pure []
go [t] = pure [t]
go (t1:t2:ts) = (:) <$> parEval (parUnionWith (+) n t1 t2) <*> go ts
We can then implement a parallel merge like this:
parMerge :: Ord a => [[a]] -> Eval [a]
parMerge xs = do
bags <- parMakeBags xs
-- Why 2 and not one? We only have half as many
-- pairs as we have lists (capabilities we want to use)
-- so we double up.
m <- parMergeBags 2 bags
pure $ concat [replicate c x | (x,c) <- M.toList m]
Putting the pieces together,
parSort :: Ord a => Int -> [a] -> Eval [a]
parSort n = parMerge . splatter n
pSort :: Ord a => Int -> [a] -> [a]
pSort n = runEval . parMerge . splatter n
There's just one sequential piece remaining that we can parallelize: converting the final bag to a list. Is it worth parallelizing? I'm pretty sure that in practice it is not. But let's do it anyway, just for fun! To avoid considerable extra complexity, I'll assume that there aren't large numbers of equal elements; repeated elements in the result will lead to some work (thunks) remaining in the result list.
We'll need a basic partial list spine forcer:
-- | Force the first n conses of a list
walkList :: Int -> [a] -> ()
walkList n _ | n <= 0 = ()
walkList _ [] = ()
walkList n (_:xs) = walkList (n - 1) xs
And now we can convert the bag to a list in parallel chunks without paying for concatenation:
-- | Use up to the given number of threads to convert a bag
-- to a list, appending the final list argument.
parToListPlus :: Int -> Bag k -> [k] -> Eval [k]
parToListPlus n m lst | n <= 1 = do
rseq (walkList (M.size m) res)
pure res
-- Note: the concat and ++ should fuse away when compiling with
-- optimization.
where res = concat [replicate c x | (x,c) <- M.toList m] ++ lst
parToListPlus _ Tip lst = pure lst
parToListPlus n (Bin _ x c l r) lst = do
r' <- parEval $ parToListPlus (n `quot` 2) r lst
res <- parToListPlus (n `quot` 2) l $ replicate c x ++ r'
rseq r' -- make sure the right side is finished
pure res
And then we modify the merger accordingly:
parMerge :: Ord a => Int -> [[a]] -> Eval [a]
parMerge n xs = do
bags <- parMakeBags xs
m <- parMergeBags 2 bags
parToListPlus n m []
I want to make a sudoku solver in Haskell (as an exercise). My idea is:
I have t :: [[Int]] representing a 9x9 grid so that it contains 0 in an empty field and 1-9 in a solved field.
A function solve :: [[Int]] -> [[Int]] returns the solved sudoku.
Here is a rough sketch of it (i'd like to point out i'm a beginner, i know it is not the most optimal code):
solve :: [[Int]] -> [[Int]]
solve t
| null (filter (elem 0) t) = t
| t /= beSmart t = solve (beSmart t)
| otherwise = guess t
The function beSmart :: [[Int]] -> [[Int]] tries to solve it by applying some solving algorithms, but if methodical approach fails (beSmart returns the unchanged sudoku table in that case) it should try to guess some numbers (and i'll think of that function later). In order to fill in an empty field, i have to find it first. And here's the problem:
beSmart :: [[Int]] -> [[Int]]
beSmart t = map f t
where f row
| elem 0 row = map unsolvedRow row
| otherwise = row
where unsolvedRow a
| a == 0 = tryToDo t r c --?!?!?!?! skip
| otherwise = a
The function tryToDo :: [[Int]]] -> Int -> Int - > Int needs the row and column of the field i'm trying to change, but i have no idea how to get that information. How do i get from map what element of the list i am in at the moment? Or is there a better way to move around in the table? I come from iterative and procedural programing and i understand that perhaps my approach to the problem is wrong when it comes to functional programing.
I know this is not really an answer to your question, but I would argue, that usually you would want a different representation (one that keeps a more detailed view of what you know about the sudoku puzzle, in your attempted solution you can only distinguish a solved cell from a cell that is free to assume any value). Sudoku is a classical instance of CSP. Where modern approaches offer many fairly general smart propagation rules, such as unit propagation (blocking a digit in neighboring cells once used somewhere), but also many other, see AC-3 for further details. Other related topics include SAT/SMT and you might find the algorithm DPLL also interesting. In the heart of most solvers there usually is some kind of a search engine to deal with non-determinism (not every instance must have a single solution that is directly derivable from the initial configuration of the instance by application of inference rules). There are also techniques such as CDCL to direct the search.
To address the question in the title, to know where you are, its probably best if you abstract the traversal of your table so that each step has access to the coordinates, you can for example zip a list of rows with [0..] (zip [0..] rows) to number the rows, when you then map a function over the zipped lists, you will have access to pairs (index, row), the same applies to columns. Just a sketch of the idea:
mapTable :: (Int -> Int -> a -> b) -> [[a]] -> [[b]]
mapTable f rows = map (\(r, rs) -> mapRow (f r) rs) $ zip [0..] rows
mapRow :: (Int -> a -> b) -> [a] -> [b]
mapRow f cols = map (uncurry f) $ zip [0..] cols
or use fold to turn your table into something else (for example to search for a unit cell):
foldrTable :: (Int -> Int -> a -> b -> b) -> b -> [[a]] -> b
foldrTable f z rows = foldr (\(r, rs) b -> foldrRow (f r) b rs) z $ zip [0..] rows
foldrRow :: (Int -> a -> b -> b) -> b -> [a] -> b
foldrRow f z cols = foldr (uncurry f) z $ zip [0..] cols
to find which cell is unital:
foldrTable
(\x y v acc -> if length v == 1 then Just (x, y) else acc)
Nothing
[[[1..9],[1..9],[1..9]],[[1..9],[1..9],[1..9]],[[1..9],[1],[1..9]]]
by using Monoid you can refactor it:
import Data.Monoid
foldrTable' :: Monoid b => (Int -> Int -> a -> b) -> [[a]] -> b
foldrTable' f rows = foldrTable (\r c a b -> b <> f r c a) mempty rows
unit :: Int -> Int -> [a] -> Maybe (Int, Int)
unit x y c | length c == 1 = Just (x, y)
| otherwise = Nothing
firstUnit :: [[[a]]] -> Maybe (Int, Int)
firstUnit = getFirst . foldrTable' (\r c v -> First $ unit r c v)
so now you would do
firstUnit [[[1..9],[1..9],[1..9]],[[1,2],[3,4],[5]]]
to obtain
Just (1, 2)
correctly determining that the first unit cell is at position 1,2 in the table.
[[Int]] is a good type for a sodoku. But map does not give any info regarding the place it is in. This is one of the ideas behind map.
You could zip together the index with the value. But a better idea would be to pass the whole [[Int]] and the indexes to to the function. So its type would become:
f :: [[Int]] -> Int -> Int -> [[Int]]
inside the function you can now access the current element by
t !! x !! y
Already did this a while ago as a learning example. It is definitely not the nicest solution, but it worked for me.
import Data.List
import Data.Maybe
import Data.Char
sodoku="\
\-9-----1-\
\8-4-2-3-7\
\-6-9-7-2-\
\--5-3-1--\
\-7-5-1-3-\
\--3-9-8--\
\-2-8-5-6-\
\1-7-6-4-9\
\-3-----8-"
sodoku2="\
\----13---\
\7-5------\
\1----547-\
\--418----\
\951-67843\
\-2---4--1\
\-6235-9-7\
\--7-98--4\
\89----1-5"
data Position = Position (Int, Int) deriving (Show)
data Sodoku = Sodoku [Int]
insertAtN :: Int -> a -> [a] -> [a]
insertAtN n y xs = intercalate [y] . groups n $ xs
where
groups n xs = takeWhile (not.null) . unfoldr (Just . splitAt n) $ xs
instance Show Sodoku where
show (Sodoku s) = (insertAtN 9 '\n' $ map intToDigit s) ++ "\n"
convertDigit :: Char -> Int
convertDigit x = case x of
'-' -> 0
x -> if digit>=1 && digit<=9 then
digit
else
0
where digit=digitToInt x
convertSodoku :: String -> Sodoku
convertSodoku x = Sodoku $ map convertDigit x
adjacentFields :: Position -> [Position]
adjacentFields (Position (x,y)) =
[Position (i,y) | i<-[0..8]] ++
[Position (x,j) | j<-[0..8]] ++
[Position (u+i,v+j) | i<-[0..2], j<-[0..2]]
where
u=3*(x `div` 3)
v=3*(y `div` 3)
positionToField :: Position -> Int
positionToField (Position (x,y)) = x+y*9
fieldToPosition :: Int -> Position
fieldToPosition x = Position (x `mod` 9, x `div` 9)
getDigit :: Sodoku -> Position -> Int
getDigit (Sodoku x) pos = x !! (positionToField pos )
getAdjacentDigits :: Sodoku -> Position -> [Int]
getAdjacentDigits s p = nub digitList
where
digitList=filter (\x->x/=0) $ map (getDigit s) (adjacentFields p)
getFreePositions :: Sodoku -> [Position]
getFreePositions (Sodoku x) = map fieldToPosition $ elemIndices 0 x
isSolved :: Sodoku -> Bool
isSolved s = (length $ getFreePositions s)==0
isDeadEnd :: Sodoku -> Bool
isDeadEnd s = any (\x->x==0) $ map length $ map (getValidDigits s)$ getFreePositions s
setDigit :: Sodoku -> Position -> Int -> Sodoku
setDigit (Sodoku x) pos digit = Sodoku $ h ++ [digit] ++ t
where
field=positionToField pos
h=fst $ splitAt field x
t=tail$ snd $ splitAt field x
getValidDigits :: Sodoku -> Position -> [Int]
getValidDigits s p = [1..9] \\ (getAdjacentDigits s p)
-- Select numbers with few possible choices first to increase execution time
sortImpl :: (Position, [Int]) -> (Position, [Int]) -> Ordering
sortImpl (_, i1) (_, i2)
| length(i1)<length(i2) = LT
| length(i1)>length(i2) = GT
| length(i1)==length(i2) = EQ
selectMoves :: Sodoku -> Maybe (Position, [Int])
selectMoves s
| length(posDigitList)>0 = Just (head posDigitList)
| otherwise = Nothing
where
posDigitList=sortBy sortImpl $ zip freePos validDigits
validDigits=map (getValidDigits s) freePos
freePos=getFreePositions s
createMoves :: Sodoku -> [Sodoku]
createMoves s=
case selectMoves s of
Nothing -> []
(Just (pos, digits)) -> [setDigit s pos d|d<-digits]
solveStep :: Sodoku -> [Sodoku]
solveStep s
| (isSolved s) = [s]
| (isDeadEnd s )==True = []
| otherwise = createMoves s
solve :: Sodoku -> [Sodoku]
solve s
| (isSolved s) = [s]
| (isDeadEnd s)==True = []
| otherwise=concat $ map solve (solveStep s)
s=convertSodoku sodoku2
readSodoku :: String -> Sodoku
readSodoku x = Sodoku []
What is the best way to map across a list, using the result of each map as you go along, when your result is of a different type to the list.
for example
f :: Int -> Int -> String -> String
l = [1,2,3,4]
I would like to have something that walks along the list l and does:
f 1 2 [] = result1 => f 2 3 result1 = result2 => f 3 4 result3 ==> return result3.
I can sort of get this to work with a an accumulator, but it seems rather cumbersome. Is there a standard way to do this... or is this something for Monads??
Thanks!
NB the function above is just for illustration.
This is just a fold left over the pairs in the input list:
f :: Int -> Int -> String -> String
f = undefined
accum :: [Int] -> String
accum xs = foldl (flip . uncurry $ f) "" $ zip xs (drop 1 xs)
You probably want to use Data.List.foldl' instead of foldl, but this is an answer that works with just Prelude.
Seems like a job for fold:
func f l = foldl (\s (x, y) -> f x y s) "" (zip l (tail l))
-- just some placeholder function
f :: Int -> Int -> String -> String
f x y s = s ++ " " ++ show(x) ++ " " ++ show(y)
l = [1,2,3,4]
main = print $ func f l
prints:
" 1 2 2 3 3 4"
(if you can change the signature of f, you can get rid of the ugly lambda that rearranges arguments)
Of course, rather than zipping, you could pass along the previous element inside the fold's accumulator. For example:
l = [1,2,3,4]
f x y = (x,y)
g b#(accum,prev) a = (accum ++ [f prev a],a)
main = print (foldl g ([],head l) (tail l))
Output:
([(1,2),(2,3),(3,4)],4)
Recently, I try to sovle the Haskell 99 Problems, the 66th (layout a tree compactly). I successed, but got confused by the solutions here(http://www.haskell.org/haskellwiki/99_questions/Solutions/66).
layout :: Tree a -> Tree (a, Pos)
layout t = t'
where (l, t', r) = layoutAux x1 1 t
x1 = maximum l + 1
layoutAux :: Int -> Int -> Tree a -> ([Int], Tree (a, Pos), [Int])
layoutAux x y Empty = ([], Empty, [])
layoutAux x y (Branch a l r) = (ll', Branch (a, (x,y)) l' r', rr')
where (ll, l', lr) = layoutAux (x-sep) (y+1) l
(rl, r', rr) = layoutAux (x+sep) (y+1) r
sep = maximum (0:zipWith (+) lr rl) `div` 2 + 1
ll' = 0 : overlay (map (+sep) ll) (map (subtract sep) rl)
rr' = 0 : overlay (map (+sep) rr) (map (subtract sep) lr)
-- overlay xs ys = xs padded out to at least the length of ys
-- using any extra elements of ys
overlay :: [a] -> [a] -> [a]
overlay [] ys = ys
overlay xs [] = xs
overlay (x:xs) (y:ys) = x : overlay xs ys
why the caculation of 'x1' and 'sep' don't cause infinit loop?
How they been calculated?
The reason for this to work is non-strict evaluation mode of Haskell rather than strict evaluation that you see in most languages.
In the example you have given, maximum l is possible to calculate because the l returned from layoutAux function doesn't contain any dependency on x1. x1 is used in the t' part of the returned value.
Another simple example to show similar behavior is below code:
hello :: [Int] -> [Int]
hello x = x' where
x' = hello' l x
l = length x'
hello' i lst = map (+i) lst
This will not loop forever because to get the length of a list you don't need to know it's content and that's why the list content dependency on l doesn't cause it to loop forever. Whereas if you had something like maximum instead of length, that would cause it to loop forever as maximum needs to know the content of list and the content depends on the result of maximum.
This seems to be a very common operation but I can't find it in hoogle for some reason. Either way, it's an interesting thought exercise. My naive implementation:
pluckL :: [a] -> Int -> Maybe ( a, [a] )
pluckL xs idx = if idx < length xs then Just $ pluck' xs idx else Nothing
where
pluck' l n = let subl = drop n l in ( head subl, rest l n ++ tail subl )
rest l n = reverse $ drop ( length l - n ) $ reverse l
My main gripe is that I'm flipping the list too many times, so I'm looking for a creative way where you can traverse the list once and generate the tuple.
There will never be an efficient way. But there can at least be a pretty way:
pluckL xs i = case splitAt i xs of
(b, v:e) -> Just (v, b ++ e)
_ -> Nothing
You can get by with one fewer reverse and fewer operations on the list if you use an accumulator:
pluckL :: [a] -> Int -> Maybe (a, [a])
pluckL xs idx = pluck xs idx [] where
pluck (x:xs) 0 acc = Just $ ( x, (reverse acc) ++ xs )
pluck (x:xs) i acc = pluck xs (i-1) (x:acc)
pluck [] i acc = Nothing
You can use elem to check if the elem is in the list or not, then depending of the result return Nothing or use delete x to remove x from the list, as follow for example,
pluckL :: Eq a => [a] -> a -> Maybe (a, [a])
pluckL xs0 x =
if (x `elem` xs0)
then Just (x, xs)
else Nothing
where xs = delete x xs0