{-# LANGUAGE PatternGuards #-} module Numeric.Histogram ( Range , binBounds , histValues , histWeightedValues , histWithBins ) where import qualified Data.Vector as V import qualified Data.Vector.Mutable as MV import Control.Monad.ST type Range a = (a,a) -- | 'binBounds a b n' generates bounds for 'n' bins spaced linearly between -- 'a' and 'b' -- -- Examples: -- -- >>> binBounds 0 3 4 -- [(0.0,0.75),(0.75,1.5),(1.5,2.25),(2.25,3.0)] binBounds :: RealFrac a => a -> a -> Int -> [Range a] binBounds a b n = map (\i->(lbound i, lbound (i+1))) [0..n-1] where lbound i = a + (b-a) * realToFrac i / realToFrac n -- | 'histValues a b n vs' returns the bins for the histogram of -- 'vs' on the range from 'a' to 'b' with 'n' bins histValues :: RealFrac a => a -> a -> Int -> [a] -> V.Vector (Range a, Int) histValues a b n = histWithBins (V.fromList $ binBounds a b n) . zip (repeat 1) -- | 'histValues a b n vs' returns the bins for the weighted histogram of -- 'vs' on the range from 'a' to 'b' with 'n' bins histWeightedValues :: RealFrac a => a -> a -> Int -> [(Double,a)] -> V.Vector (Range a, Double) histWeightedValues a b n = histWithBins (V.fromList $ binBounds a b n) -- | 'histWithBins bins xs' is the histogram of weighted values 'xs' with 'bins' -- -- Examples: -- -- >>> :{ -- histWithBins -- (V.fromList [(0.0, 0.75), (0.75, 1.5), (1.5, 2.25), (2.25, 3.0)]) -- [(1, 0), (1, 0), (1, 1), (1, 2), (1, 2), (1, 2), (1, 3)] -- :} -- [((0.0,0.75),2),((0.75,1.5),1),((1.5,2.25),3),((2.25,3.0),1)] histWithBins :: (Num w, RealFrac a) => V.Vector (Range a) -> [(w, a)] -> V.Vector (Range a, w) histWithBins bins xs = let n = V.length bins testBin :: RealFrac a => a -> (Int, Range a) -> Bool testBin x (i, (a,b)) = if i == n - 1 then x >= a && x <= b else x >= a && x < b f :: (RealFrac a, Num w) => V.Vector (Range a) -> MV.STVector s w -> (w, a) -> ST s () f bins1 bs (w,x) = case V.dropWhile (not . testBin x) $ V.indexed bins1 of v | V.null v -> return () v | (idx,_) <- V.head v -> do m <- MV.read bs idx MV.write bs idx $! m+w counts = runST $ do b <- MV.replicate n 0 mapM_ (f bins b) xs V.freeze b in V.zip bins counts