摘要:
本章介绍了二叉查找树的概念及操作。主要内容包括二叉查找树的性质,如何在二叉查找树中查找最大值、最小值和给定的值,如何找出某一个元素的前驱和后继,如何在二叉查找树中进行插入和删除操作。在二叉查找树上执行这些基本操作的时间与树的高度成正比,一棵随机构造的二叉查找树的期望高度为O(lgn),从而基本动态集合的操作平均时间为θ(lgn)。
1、二叉查找树
二叉查找树是按照二叉树结构来组织的,因此可以用二叉链表结构表示。二叉查找树中的关键字的存储方式满足的特征是:设x为二叉查找树中的一个结点。如果y是x的左子树中的一个结点,则key[y]≤key[x]。如果y是x的右子树中的一个结点,则key[x]≤key[y]。根据二叉查找树的特征可知,采用中根遍历一棵二叉查找树,可以得到树中关键字有小到大的序列。http://www.cnblogs.com/Anker/archive/2013/01/27/2878594.html介绍了二叉树概念及其遍历。一棵二叉树查找及其中根遍历结果如下图所示:
书中给出了一个定理:如果x是一棵包含n个结点的子树的根,则其中根遍历运行时间为θ(n)。
问题:二叉查找树性质与最小堆之间有什么区别?能否利用最小堆的性质在O(n)时间内,按序输出含有n个结点的树中的所有关键字?
2、查询二叉查找树
二叉查找树中最常见的操作是查找树中的某个关键字,除了基本的查询,还支持最大值、最小值、前驱和后继查询操作,书中就每种查询进行了详细的讲解。
(1)查找SEARCH
在二叉查找树中查找一个给定的关键字k的过程与二分查找很类似,根据二叉查找树在的关键字存放的特征,很容易得出查找过程:首先是关键字k与树根的关键字进行比较,如果k大比根的关键字大,则在根的右子树中查找,否则在根的左子树中查找,重复此过程,直到找到与遇到空结点为止。例如下图所示的查找关键字13的过程:(查找过程每次在左右子树中做出选择,减少一半的工作量)
书中给出了查找过程的递归和非递归形式的伪代码:
TREE_SEARCH(x,k)
if x=NULL or k=key[x]
then return x
if(k<key[x])
then return TREE_SEARCH(left[x],k)
else
then return TREE_SEARCH(right[x],k)
ITERATIVE_TREE_SEARCH(x,k)
while x!=NULL and k!=key[x]
do if k<key[x]
then x=left[x]
else
then x=right[x]
return x
(2)查找最大关键字和最小关键字
根据二叉查找树的特征,很容易查找出最大和最小关键字。查找二叉树中的最小关键字:从根结点开始,沿着各个节点的left指针查找下去,直到遇到NULL时结束。如果一个结点x无左子树,则以x为根的子树中,最小关键字就是key[x]。查找二叉树中的最大关键字:从根结点开始,沿着各个结点的right指针查找下去,直到遇到NULL时结束。书中给出了查找最大最小关键字的伪代码:
1 TREE_MINMUM(x) 2 while left[x] != NULL 3 do x=left[x] 4 return x
1 TREE_MAXMUM(x) 2 while right[x] != NULL 3 do x= right[x] 4 return x
(3)前驱和后继
给定一个二叉查找树中的结点,找出在中序遍历顺序下某个节点的前驱和后继。如果树中所有关键字都不相同,则某一结点x的前驱就是小于key[x]的所有关键字中最大的那个结点,后继即是大于key[x]中的所有关键字中最小的那个结点。根据二叉查找树的结构和性质,不用对关键字做任何比较,就可以找到某个结点的前驱和后继。
查找前驱步骤:先判断x是否有左子树,如果有则在left[x]中查找关键字最大的结点,即是x的前驱。如果没有左子树,则从x继续向上执行此操作,直到遇到某个结点是其父节点的右孩子结点。例如下图查找结点7的前驱结点6过程:
查找后继步骤:先判断x是否有右子树,如果有则在right[x]中查找关键字最小的结点,即使x的后继。如果没有右子树,则从x的父节点开始向上查找,直到遇到某个结点是其父结点的左儿子的结点时为止。
书中给出了求x结点后继结点的伪代码:
TREE_PROCESSOR(x)
if right[x] != NULL
then return TREE_MINMUM(right(x))
y=parent[x]
while y!= NULL and x ==right[y]
do x = y
y=parent[y]
return y
定理:对一棵高度为h的二叉查找,动态集合操作SEARCH、MINMUM、MAXMUM、SUCCESSOR、PROCESSOR等的运行时间均为O(h)。
3、插入和删除
插入和删除会引起二叉查找表示的动态集合的变化,难点在在插入和删除的过程中要保持二叉查找树的性质。插入过程相当来说要简单一些,删除结点比较复杂。
(1)插入
插入结点的位置对应着查找过程中查找不成功时候的结点位置,因此需要从根结点开始查找带插入结点位置,找到位置后插入即可。下图所示插入结点过程:
书中给出了插入过程的伪代码:
TREE_INSERT(T,z)
y = NULL;
x =root[T]
while x != NULL
do y =x
if key[z] < key[x]
then x=left[x]
else x=right[x]
parent[z] =y
if y=NULL
then root[T] =z
else if key[z]>key[y]
then keft[y] = z
else right[y] =z
插入过程运行时间为O(h),h为树的高度。
(2)删除
从二叉查找树中删除给定的结点z,分三种情况讨论:
<1>结点z没有左右子树,则修改其父节点p[z],使其为NULL。删除过程如下图所示:
<2>如果结点z只有一个子树(左子树或者右子树),通过在其子结点与父节点建立一条链来删除z。删除过程如下图所示:
<3>如果z有两个子女,则先删除z的后继y(y没有左孩子),在用y的内容来替代z的内容
书中给出了删除过程的伪代码:
TREE_DELETE(T,z)
if left[z] ==NULL or right[z] == NULL
then y=z
else y=TREE_SUCCESSOR(z)
if left[y] != NULL
then x=left[y]
else x=right[y]
if x!= NULL
then parent[x] = parent[y]
if p[y] ==NULL
then root[T] =x
else if y = left[[prarnt[y]]
then left[parent[y]] = x
else right[parent[y]] =x
if y!=z
then key[z] = key[y]
copy y's data into z
return y
定理:对高度为h的二叉查找树,动态集合操作INSERT和DELETE的运行时间为O(h)。
4、实现测试
采用C++语言实现一个简单的二叉查找树,支持动态集合的基本操作:search、minmum、maxmum、predecessor、successor、insert和delete。设计的二叉查找树结构如下所示:
template <class T>
class BinarySearchTreeNode
{
public:
T elem;
struct BinarySearchTreeNode<T> *parent;
struct BinarySearchTreeNode<T>* left;
struct BinarySearchTreeNode<T>* right;
};
template <class T>
class BinarySearchTree
{
public:
BinarySearchTree();
void tree_insert(const T& elem);
int tree_remove(const T& elem );
BinarySearchTreeNode<T>* tree_search(const T& elem)const;
T tree_minmum(BinarySearchTreeNode<T>* root)const;
T tree_maxmum(BinarySearchTreeNode<T>* root)const;
T tree_successor(const T& elem) const;
T tree_predecessor(const T& elem)const;
int empty() const;
void inorder_tree_walk()const;
BinarySearchTreeNode<T>* get_root()const {return root;}
private:
BinarySearchTreeNode<T>* root;
};
完整程序:
#include <iostream>
#include <stack>
#include <cstdlib>
using namespace std;
template <class T>
class BinarySearchTreeNode
{
public:
T elem;
struct BinarySearchTreeNode<T> *parent;
struct BinarySearchTreeNode<T>* left;
struct BinarySearchTreeNode<T>* right;
};
template <class T>
class BinarySearchTree
{
public:
BinarySearchTree();
void tree_insert(const T& elem);
int tree_remove(const T& elem );
BinarySearchTreeNode<T>* tree_search(const T& elem)const;
T tree_minmum(BinarySearchTreeNode<T>* root)const;
T tree_maxmum(BinarySearchTreeNode<T>* root)const;
T tree_successor(const T& elem) const;
T tree_predecessor(const T& elem)const;
int empty() const;
void inorder_tree_walk()const;
BinarySearchTreeNode<T>* get_root()const {return root;}
private:
BinarySearchTreeNode<T>* root;
};
template <class T>
BinarySearchTree<T>::BinarySearchTree()
{
root = NULL;
}
template <class T>
void BinarySearchTree<T>::tree_insert(const T& elem)
{
if(!empty())
{
BinarySearchTreeNode<T> *pnode = root;
BinarySearchTreeNode<T> *qnode = NULL;
BinarySearchTreeNode<T> *newnode = new BinarySearchTreeNode<T>;
newnode->elem = elem;
newnode->parent = NULL;
newnode->left = NULL;
newnode->right = NULL;
while(pnode)
{
qnode = pnode;
if(pnode->elem > elem)
pnode = pnode->left;
else
pnode = pnode->right;
}
if(qnode->elem > elem)
qnode->left = newnode;
else
qnode->right = newnode;
newnode->parent = qnode;
}
else
{
root = new BinarySearchTreeNode<T>;
root->elem = elem;
root->parent =NULL;
root->left = NULL;
root->right = NULL;
}
}
template <class T>
int BinarySearchTree<T>::tree_remove(const T&elem)
{
BinarySearchTreeNode<T> *pnode;
BinarySearchTreeNode<T> *parentnode,*snode;
pnode = tree_search(elem);
if(pnode != NULL)
{
parentnode = pnode->parent;
if(pnode->right == NULL || pnode->left == NULL)
{
if(pnode->right != NULL)
{
if(parentnode->left == pnode)
parentnode->left = pnode->right;
if(parentnode->right == pnode)
parentnode->right = pnode->right;
pnode->right->parent = parentnode;
}
else if(pnode->left != NULL)
{
if(parentnode->left == pnode)
parentnode->left = pnode->left;
if(parentnode->right == pnode)
parentnode->right = pnode->left;
pnode->left->parent = parentnode;
}
else
{
if(parentnode->left == pnode)
parentnode->left = NULL;
if(parentnode->right == pnode)
parentnode->right = NULL;
}
delete pnode;
}
else
{
snode = tree_search(tree_successor(pnode->elem));
pnode->elem = snode->elem;
if(snode->parent->left == snode)
{
snode->parent->left = snode->right;
snode->right->parent = snode->parent->left;
}
if(snode->parent->right == snode)
{
snode->parent->right = snode->right;
snode->right->parent = snode->parent->right;
}
delete snode;
}
return 0;
}
return -1;
}
template <class T>
BinarySearchTreeNode<T>* BinarySearchTree<T>::tree_search(const T& elem)const
{
BinarySearchTreeNode<T> *pnode = root;
while(pnode)
{
if(pnode->elem == elem)
break;
else if(pnode->elem > elem)
pnode = pnode->left;
else
pnode = pnode->right;
}
return pnode;
}
template <class T>
T BinarySearchTree<T>::tree_minmum(BinarySearchTreeNode<T>* root)const
{
BinarySearchTreeNode<T> *pnode = root;
if(pnode->left)
{
while(pnode->left)
pnode = pnode->left;
}
return pnode->elem;
}
template <class T>
T BinarySearchTree<T>::tree_maxmum(BinarySearchTreeNode<T>* root)const
{
BinarySearchTreeNode<T> *pnode = root;
if(pnode->right)
{
while(pnode->right)
pnode = pnode->right;
}
return pnode->elem;
}
template <class T>
T BinarySearchTree<T>::tree_successor(const T& elem) const
{
BinarySearchTreeNode<T>* pnode = tree_search(elem);
BinarySearchTreeNode<T>* parentnode;
if(pnode != NULL)
{
if(pnode->right)
return tree_minmum(pnode->right);
parentnode = pnode->parent;
while(parentnode && pnode == parentnode->right)
{
pnode = parentnode;
parentnode = parentnode->parent;
}
if(parentnode)
return parentnode->elem;
else
return T();
}
return T();
}
template <class T>
T BinarySearchTree<T>::tree_predecessor(const T& elem)const
{
BinarySearchTreeNode<T>* pnode = tree_search(elem);
BinarySearchTreeNode<T>* parentnode;
if(pnode != NULL)
{
if(pnode->right)
return tree_maxmum(pnode->right);
parentnode = pnode->parent;
while(parentnode && pnode == parentnode->left)
{
pnode = parentnode;
parentnode = pnode->parent;
}
if(parentnode)
return parentnode->elem;
else
return T();
}
return T();
}
template <class T>
int BinarySearchTree<T>::empty() const
{
return (NULL == root);
}
template <class T>
void BinarySearchTree<T>::inorder_tree_walk()const
{
if(NULL != root)
{
stack<BinarySearchTreeNode<T>*> s;
BinarySearchTreeNode<T> *ptmpnode;
ptmpnode = root;
while(NULL != ptmpnode || !s.empty())
{
if(NULL != ptmpnode)
{
s.push(ptmpnode);
ptmpnode = ptmpnode->left;
}
else
{
ptmpnode = s.top();
s.pop();
cout<<ptmpnode->elem<<" ";
ptmpnode = ptmpnode->right;
}
}
}
}
int main()
{
BinarySearchTree<int> bstree;
BinarySearchTreeNode<int>* ptnode,*proot;
bstree.tree_insert(32);
bstree.tree_insert(21);
bstree.tree_insert(46);
bstree.tree_insert(54);
bstree.tree_insert(16);
bstree.tree_insert(38);
bstree.tree_insert(70);
cout<<"inorder tree walk is: ";
bstree.inorder_tree_walk();
proot = bstree.get_root();
cout<<"\nmax value is: "<<bstree.tree_maxmum(proot)<<endl;
cout<<"min value is: "<<bstree.tree_minmum(proot)<<endl;
ptnode = bstree.tree_search(38);
if(ptnode)
cout<<"the element 38 is exist in the binary tree.\n";
else
cout<<"the element 38 is not exist in the binary tree.\n";
cout<<"the successor of 38 is: "<<bstree.tree_successor(38)<<endl;
cout<<"the predecessor of 38 is:"<<bstree.tree_predecessor(38)<<endl;
if(bstree.tree_remove(46) == 0)
cout<<"delete 46 successfully"<<endl;
else
cout<<"delete 46 failed"<<endl;
cout<<"inorder tree walk is: ";
bstree.inorder_tree_walk();
exit(0);
}