The bst(binary search tree)'s inorder traverse is ascend. In the following solution, I created an assistant point to record current subtree’s root’s pre nodes in inorder traverse. The key point here is that exactly two nodes were swapped. which could result in one or two inverse pairs. Suppose the sequence is a1,a2, …, an and ai, aj which is in the sequence is the inverse pair nodes, and if ai is next to aj, only one inverse pair <ai, aj> will be found, and when ai isn’t next to aj, two inverse pairs would be found, which is <ai, ai+1> and <aj-1, aj> (i < j). So in the swap function, when I find the first inverse pair, I just save it and when I found a second inverse pair, I exchange ai and aj which is a[0] and root to recover the tree. And if I don’t find a second inverse pair, I can now make a conclusion that ai is next to aj, so I just need to exchange a[0] and a[1] which is exactly ai and aj. And the time complexity is the same as space complexity which is O(n).
func swap(pre **TreeNode, root *TreeNode, a []*TreeNode, exchanged *bool){
if (*pre)!=nil{
if (*pre).Val < root.Val{
//normal
}else{
if a[0]==nil {
a[0]=*pre
a[1]=root
}else{
temp:=root.Val
root.Val=a[0].Val
a[0].Val=temp
(*exchanged)=true
}
}
}
}
func fakeRecoverTree(pre **TreeNode, root *TreeNode, a []*TreeNode, exchanged *bool) {
if root==nil{
return
}
fakeRecoverTree(pre,root.Left,a,exchanged)
swap(pre,root,a,exchanged)
*pre=root
fakeRecoverTree(pre,root.Right,a,exchanged)
}
func recoverTree(root *TreeNode) {
var a []*TreeNode = []*TreeNode{nil,nil}
var pre *TreeNode = nil
var exchanged bool=false
fakeRecoverTree(&pre,root,a,&exchanged)
if !exchanged&&a[1]!=nil{
temp:=a[1].Val
a[1].Val=a[0].Val
a[0].Val=temp
}
}
博客围绕二叉搜索树(BST)展开,指出其按中序遍历是升序的。当BST中恰好两个节点被交换时,会出现一到两个逆序对。通过特定方法找到逆序对节点并交换,恢复二叉搜索树,时间和空间复杂度均为O(n)。
1575

被折叠的 条评论
为什么被折叠?



