[LeetCode] 230. Kth Smallest Element in a BST（二叉搜索树里的第 k 小元素）

• Difficulty: Medium

• Related Topics: Binary Search, Tree

• Link: https://leetcode.com/problems/kth-smallest-element-in-a-bst/

Description

Given a binary search tree, write a function kthSmallest to find the kth smallest element in it.
给一棵二叉搜索树，写一个函数 kthSmallest 找出其中第 k 小的元素。

Examples

Example 1

Input: root = [3,1,4,null,2], k = 1
   3
  / 
 1   4
  
   2
Output: 1

Example 2

Input: root = [5,3,6,2,4,null,null,1], k = 3
       5
      / 
     3   6
    / 
   2   4
  /
 1
Output: 3

Follow up

What if the BST is modified (insert/delete operations) often and you need to find the kth smallest frequently? How would you optimize the kthSmallest routine?
如果二叉搜索树经常被修改（插入/删除节点）并且你又需要经常查询第 k 小元素呢？你该怎样优化你的代码？

Constraints

• The number of elements of the BST is between 1 to 10^4.
  二叉搜索树的节点范围在 1 到 10^4 之间。

• You may assume k is always valid, 1 ≤ k ≤ BST's total elements.
  你可以假设 k 总是有效的，1 ≤ k ≤ 二叉搜索树的节点数量。

Hints

1. Try to utilize the property of a BST.
   尝试利用二叉搜索树的性质。

2. Try in-order traversal. (Credits to @chan13)
   尝试中序遍历。

3. What if you could modify the BST node's structure?
   如果你能修改二叉搜索树节点的结构呢？

4. The optimal runtime complexity is O(height of BST).
   更优的时间复杂度是 O(二叉搜索树的高度)。

Solution

由于对二叉树进行中序遍历即可得到有序数组，所以可以利用中序遍历来寻找第 k 小的元素，时间复杂度 (O(N))，代码如下：

/**
 * Example:
 * var ti = TreeNode(5)
 * var v = ti.`val`
 * Definition for a binary tree node.
 * class TreeNode(var `val`: Int) {
 *     var left: TreeNode? = null
 *     var right: TreeNode? = null
 * }
 */
class Solution {
    private var index = 0
    private var result = -1

    fun kthSmallest(root: TreeNode?, k: Int): Int {
        inOrder(root, k)
        return result
    }

    private fun inOrder(root: TreeNode?, k: Int) {
        if (index >= k && result != -1) {
            return
        }
        if (root != null) {
            inOrder(root.left, k)
            index++
            if (index == k) {
                result = root.`val`
                return
            }
            inOrder(root.right, k)
        }
    }
}

既然二叉搜索树有这种性质，为什么不能用二分搜索呢？翻了一下 Discussion，确实有这样的解法，代码如下：

/**
 * Example:
 * var ti = TreeNode(5)
 * var v = ti.`val`
 * Definition for a binary tree node.
 * class TreeNode(var `val`: Int) {
 *     var left: TreeNode? = null
 *     var right: TreeNode? = null
 * }
 */
class Solution {
    fun kthSmallest(root: TreeNode?, k: Int): Int {
        val leftNodes = nodeCount(root?.left)

        return when {
            k <= leftNodes -> kthSmallest(root?.left, k)
            k > leftNodes + 1 -> kthSmallest(root?.right, k - 1 - leftNodes)
            else -> root!!.`val`
        }
    }

    private fun nodeCount(root: TreeNode?): Int = if (root == null) {
        0
    } else {
        1 + nodeCount(root.left) + nodeCount(root.right)
    }
}

不过这样算节点的数量也是要消耗额外时间的，所以需要将其存储起来，这也就是提示 3 里的内容。

官方给出的 Solution 则是利用栈这一结构，将递归查找变成了迭代，代码如下（虽然我看不出这有啥优势）：

/**
 * Example:
 * var ti = TreeNode(5)
 * var v = ti.`val`
 * Definition for a binary tree node.
 * class TreeNode(var `val`: Int) {
 *     var left: TreeNode? = null
 *     var right: TreeNode? = null
 * }
 */
class Solution {
    fun kthSmallest(root: TreeNode?, k: Int): Int {
        val stack = ArrayDeque<TreeNode>()

        var p = root
        var count = k
        while (true) {
            while (p != null) {
                stack.push(p)
                p = p.left
            }
            p = stack.pop()
            if (--count == 0) {
                return p.`val`
            }
            p = p.right
        }
    }
}