堆：左倾树

堆是一种树型数据结构，一般组织成二叉树的形式。堆具有这样的属性，对于每一个节点，父节点上的元素大于子孙节点。具体实现可以用链表，也可以用数组。优先队列(Priority Queue)可以用堆来实现。下面介绍一种左倾树(Leftist Tree)的实现方式，参考Algorithms:a functional programming approach.

 

左倾树首先是堆，具有堆的属性。左倾树的每一个节点除空节点外，具有这些域：数据，距离，左子树，右子树。此处的距离不是指树的深度，而是指节点到达空节点经历的最少节点个数。规定空节点的距离是0。左倾树具有这样的特点，左子树上的距离大于等于右子树上的距离，左倾树的每一个子树都是左倾树。

 

左倾树一般提供这些操作：取最大值，插入一个元素，删除最大元素，归并(Merge)左倾树。插入一个元素操作，可以通过归并原左倾树和待插入元素构成的独苗来完成。删除最大元素，可以通过移除左倾树的根节点，然后归并左右子树来完成。

 

左倾树不是平衡树，之所以采用这种不平衡的结构是因为它归并操作比较省时，只需lgn的时间复杂度，比较容易保持堆的属性（对于每一个节点，父节点上的元素大于子孙节点）。左倾树体现了一种不平衡的美。AVL树体现平衡的美。各有各的美。

 

下面是实现代码：

-- LeftistTree.hs module LeftistTree where data (Ord a) => LeftistTree a = EmptyTree | LeftistTree a Int (LeftistTree a) (LeftistTree a) treeEmpty EmptyTree = True treeEmpty _ = False makeEmptyTree = EmptyTree distance :: (Ord a)=> LeftistTree a -> Int distance EmptyTree = 0 distance (LeftistTree elem dist left right) = dist makeLeftistTree:: (Ord a)=> a -> LeftistTree a -> LeftistTree a -> LeftistTree a makeLeftistTree elem left right | distance left >= distance right = LeftistTree elem (distance right +1) left right | otherwise = LeftistTree elem (distance left +1) right left mergeLeftistTree:: (Ord a)=>LeftistTree a -> LeftistTree a -> LeftistTree a mergeLeftistTree EmptyTree right = right mergeLeftistTree left EmptyTree = left mergeLeftistTree left@(LeftistTree e1 d1 l1 r1) right@(LeftistTree e2 d2 l2 r2) | e1 >= e2 = makeLeftistTree e1 l1 (mergeLeftistTree r1 right) | otherwise = makeLeftistTree e2 l2 (mergeLeftistTree left r2) getMax :: (Ord a ) => LeftistTree a -> a getMax EmptyTree = error "getMax on empty tree" getMax (LeftistTree e _ _ _) = e insertLeftistTree::(Ord a)=>LeftistTree a -> a -> LeftistTree a insertLeftistTree tree1 a = mergeLeftistTree tree1 (makeLeftistTree a EmptyTree EmptyTree ) removeLeftistTree::(Ord a)=>LeftistTree a -> LeftistTree a removeLeftistTree EmptyTree = error "remove on empty tree" removeLeftistTree (LeftistTree _ _ left right) = mergeLeftistTree left right instance (Show a, Ord a)=> Show (LeftistTree a) where show tree = showLevel tree 0 where showLevel EmptyTree level= "" showLevel (LeftistTree e d l r) level = levelSpaces level ++ show(e)++ ":" ++show(d) ++ "/r/n" ++ showLevel l (level+1) ++ showLevel r (level+1) levelSpaces level | level==0 = "" | otherwise = ' ':' ':' ': (levelSpaces ( level -1 ) ) main=test test = do let elems = [1,3,5,7,9,2,4,6,8,0] let tree = makeEmptyTree let tree1 = foldl insertLeftistTree tree elems putStrLn "tree1" print tree1 putStrLn "max1" print $ getMax tree1 let tree2 = removeLeftistTree tree1 putStrLn "tree2" print tree2 putStrLn "max2" print $ getMax tree2

下面是试验结果：

#>runghc LeftistTree.hs tree1 9:2 8:2 6:1 4:1 2:1 0:1 7:1 5:1 3:1 1:1 max1 9 tree2 8:2 7:2 5:1 3:1 1:1 0:1 6:1 4:1 2:1 max2 8