uniqueness-periods-vector-filters 0.2.0.0 → 0.3.0.0

raw patch · 3 files changed

+69/−34 lines, 3 filesPVP ok

version bump matches the API change (PVP)

API changes (from Hackage documentation)

Files

• ChangeLog.md +6/−2
• Languages/UniquenessPeriods/Vector/Filters.hs +61/−30
• uniqueness-periods-vector-filters.cabal +2/−2

@@ -6,8 +6,12 @@  ## 0.1.1.0 -- 2020-09-17 -* First version revised A. Defined intervalNRealFrac for the equal two first arguments (though this can be rarely useful). +* First version revised A. Defined intervalNRealFrac for the equal two first arguments (though this can be rarely useful).  ## 0.2.0.0 -- 2020-09-19 -* Second version. Some code re-defining for special cases. +* Second version. Some code re-defining for special cases.++## 0.3.0.0 -- 2020-10-07++* Third version. Added two new functions-examples. Some documentation improvements.

@@ -5,8 +5,8 @@ -- Stability   :  Experimental -- Maintainer  :  olexandr543@yahoo.com ----- A module allows to change the structure of the function output for the functions of --- elements from 'RealFrac' class. At the moment only the equal intervals are supported. +-- A module allows to change the structure of the function output for the functions of+-- elements from 'RealFrac' class. At the moment only the equal intervals are supported.  module Languages.UniquenessPeriods.Vector.Filters (   -- * One interval used@@ -16,32 +16,35 @@   -- * Several intervals   , unsafeRearrangeIG   , unsafeRearrangeIGV+  -- * Some basic usage examples+  , unsafeSwapIWithMaxI+  , unsafeSwapVecIWithMaxI ) where  import CaseBi import qualified Data.Vector as V --- | Given the minimum and maximum elements, a quantity of equal intervals, and an element in between the first two arguments (or equal to one of them), finds out the +-- | Given the minimum and maximum elements, a quantity of equal intervals, and an element in between the first two arguments (or equal to one of them), finds out the -- index of the interval, to which the element belongs (starting from 1). The minimum element belongs to the interval with the index 1.-intervalNRealFrac -  :: RealFrac b => b -  -> b +intervalNRealFrac+  :: RealFrac b => b+  -> b   -> Int-  -> b +  -> b   -> Int-intervalNRealFrac minE maxE n x +intervalNRealFrac minE maxE n x  | maxE == minE = ceiling (0.5 * fromIntegral n)  | otherwise = zero2One . ceiling $ fromIntegral n * (x - minE) / (maxE - minE)-{-# INLINE intervalNRealFrac #-} +{-# INLINE intervalNRealFrac #-}  zero2One :: Int -> Int zero2One x = if x == 0 then 1 else x-{-# INLINE zero2One #-} +{-# INLINE zero2One #-} --- | Moves (if needed) the given value so that its result divides the new [min..max] interval in the same proportion as the starting one. Is intended to be used --- for the arguments satisfying some additional constraints, but they are not checked (hence, its name prefix \"unsafe\"). For example, the second argument must be --- greater than the first one, the fourth -- than the third one, and the fifth must be located in between the first two. Then the result is also located in between --- the third and fourth arguments similarly. +-- | Moves (if needed) the given value so that its result divides the new [min..max] interval in the same proportion as the starting one. Is intended to be used+-- for the arguments satisfying some additional constraints, but they are not checked (hence, its name prefix \"unsafe\"). For example, the second argument must be+-- greater than the first one, the fourth -- than the third one, and the fifth must be located in between the first two. Then the result is also located in between+-- the third and fourth arguments similarly. unsafeTransfer1I5   :: RealFrac b => b   -> b@@ -49,10 +52,10 @@   -> b   -> b   -> b-unsafeTransfer1I5 minE0 maxE0 minE1 maxE1 x +unsafeTransfer1I5 minE0 maxE0 minE1 maxE1 x  | minE0 == maxE0 = x- | otherwise = minE1 + (x - minE0) * (maxE1 - minE1) / (maxE0 - minE0) -{-# INLINE unsafeTransfer1I5 #-} + | otherwise = minE1 + (x - minE0) * (maxE1 - minE1) / (maxE0 - minE0)+{-# INLINE unsafeTransfer1I5 #-}  -- | A variant of the 'unsafeTransfer1I5' where the lengths of the both intervals (the old and the new ones) are equal. transfer1IEq3@@ -61,34 +64,62 @@   -> b   -> b transfer1IEq3 minE0 minE1 = (+ (minE1 - minE0))-{-# INLINE transfer1IEq3 #-} +{-# INLINE transfer1IEq3 #-} --- | Makes a complex interval-based transformation moving the value from its own interval to the corresponding by the 'V.Vector' of tuples second element of the --- respective pair with the first element being the starting number of the interval (numeration of them begins at 1). The 'V.Vector' argument must be sorted --- by the first argument in the ascending order. Usually, its first elements in the tuples are from the range @[1..n]@. Number of the intervals are given as --- the third argument and for many cases should not be greater than 10. There do exist several semantical constraints for the possible accurate arguments, --- but they are not checked. For example, the first argument must be less than the second one; the fifth argument must be located between the first two ones; --- the 'Int' argument must be greater than zero. -unsafeRearrangeIG +-- | Makes a complex interval-based transformation moving the value from its own interval to the corresponding by the 'V.Vector' of tuples second element of the+-- respective pair with the first element being the starting number of the interval (numeration of them begins at 1). The 'V.Vector' argument must be sorted+-- by the first argument in the ascending order. Usually, its first elements in the tuples are from the range @[1..n]@. Number of the intervals are given as+-- the third argument and for many cases should not be greater than 10. There do exist several semantical constraints for the possible accurate arguments,+-- but they are not checked. For example, the first argument must be less than the second one; the fifth argument must be located between the first two ones;+-- the 'Int' argument must be greater than zero.+unsafeRearrangeIG   :: RealFrac b => b   -> b   -> Int   -> V.Vector (Int,Int)   -> b   -> b-unsafeRearrangeIG minE maxE n v x +unsafeRearrangeIG minE maxE n v x  | minE == maxE = x  | otherwise = x + fromIntegral (getBFst' (n0, v) n0 - n0) * (maxE - minE) / fromIntegral n       where n0 = intervalNRealFrac minE maxE n x --- | An unzipped variant of the 'unsafeRearrangeIG' function where the 'V.Vector' argument is internally 'V.zip'ped as the second argument with the 'V.Vector' @[1..n]@. +-- | An unzipped variant of the 'unsafeRearrangeIG' function where the 'V.Vector' argument is internally 'V.zip'ped as the second argument with the 'V.Vector' @[1..n]@. -- This allows to shorten the time of the arguments writing given only the resulting backpermutted indexes in the 'V.Vector'.-unsafeRearrangeIGV +unsafeRearrangeIGV   :: RealFrac b => b   -> b   -> Int   -> V.Vector Int   -> b   -> b-unsafeRearrangeIGV minE maxE n v = unsafeRearrangeIG minE maxE n . V.zip (V.enumFromN 1 n) . V.filter (\y -> compare y 0 /= LT && compare y n == LT) $ v -{-# INLINE unsafeRearrangeIGV #-}            +unsafeRearrangeIGV minE maxE n v = unsafeRearrangeIG minE maxE n . V.zip (V.enumFromN 1 n) $ v+{-# INLINE unsafeRearrangeIGV #-}++-- | Swaps the k-th inner interval values with the maximum one's (that is the n-th one) values.+unsafeSwapIWithMaxI+  :: RealFrac b => b+  -> b+  -> Int -- ^ It is expected to be greater than 0, though this is not checked.+  -> Int -- ^ It is expected to be less than the previous argument, but greater than 0, though this is not checked.+  -> b -- ^ It is expected to lie between the first two arguments, though this is not checked.+  -> b+unsafeSwapIWithMaxI minE maxE n k = unsafeRearrangeIGV minE maxE n (V.generate n (\i -> if i == k - 1 then n - 1 else if i == n - 1 then k - 1 else i))+{-# INLINE unsafeSwapIWithMaxI #-}++-- | Swaps the inner intervals values (given by the 'V.Vector' of 'Int' elements that represent numbers-indices starting from 1 to n) with the maximum one's+-- (that is the n-th one) values. The 'V.Vector' must be not empty and sorted in the ascending order, though it is not checked. Be aware that this can+-- significantly change the density of the values and break some other properties for distributions.+unsafeSwapVecIWithMaxI+  :: RealFrac b => b+  -> b+  -> Int -- ^ It is expected to be greater than 0, though this is not checked.+  -> V.Vector Int -- ^ It is expected the 'V.Vector' to be sorted in the ascending order (indices are counted in it starting with 1 opposed to the usual behaviour for 'V.Vector's and are the numbers of the intervals in the range from 1 to n), and besides all the elements to be less than the previous argument, greater than 0 and to be not pairwise equal, though it is not checked.+  -> b -- ^ It is expected to lie between the first two arguments, though this is not checked.+  -> b+unsafeSwapVecIWithMaxI minE maxE n vI = unsafeRearrangeIGV minE maxE n (V.generate n h)+  where h i+         | getBFst' (False, V.zip (V.map (+ (-1)) vI) . V.replicate n $ True) i = n - 1+         | i == n - 1 = V.unsafeIndex vI 0+         | otherwise = i+{-# INLINE unsafeSwapVecIWithMaxI #-}

@@ -2,9 +2,9 @@ --   For further documentation, see http://haskell.org/cabal/users-guide/  name:                uniqueness-periods-vector-filters-version:             0.2.0.0+version:             0.3.0.0 synopsis:            A library allows to change the structure of the 'RealFrac' function output.-description:         A library allows to change the structure of the 'RealFrac' function output. At the moment only the equal intervals are supported. +description:         A library allows to change the structure of the 'RealFrac' function output. At the moment only the equal intervals are supported.  homepage:            https://hackage.haskell.org/package/uniqueness-periods-vector-filters license:             MIT