Binary Search Trees

What is a binary tree?

• Property 1: each node can have up to two
successor nodes.
 What is a binary tree? (cont.)
• Property 2: a unique path exists from the
root to every other node

Not a valid binary tree!

 Some terminology
• The successor nodes of a node are called its children
• The predecessor node of a node is called its parent
• The "beginning" node is called the root (has no parent)
• A node without children is called a leaf
 Some terminology (cont’d)
• Nodes are organize in levels (indexed from 0).

• Level (or depth) of a node: number of edges in the path

from the root to that node.

not full!
• Height of a tree h: #levels = L
(Warning: some books define h
as #levels-1).

• Full tree: every node has exactly

two children and all the
leaves are on the same level.
 What is the max #nodes
at some level l?
The max #nodes at level l is 2l where l=0,1,2, ...,L-1

2 0
21
2 2
2 3
What is the total #nodes N
of a full tree with height h?

0 1 h 1 h
N 2  2  ...  2 2  1
l=0 l=1 l=h-1

using the geometric series:

n 1
0 1
x  x  ...  x n 1
 x  i xn  1
x 1
i 0
 What is the height h
of a full tree with N nodes?

h
2  1 N
h
 2 N  1
 h log( N  1)  O(log N )
 Why is h important?
• Tree operations (e.g., insert, delete, retrieve
etc.) are typically expressed in terms of h.

• So, h determines running time!

•What is the max height of a tree with
N nodes? N (same as a linked list)

•What is the min height of a tree with

N nodes? log(N+1)
 How to search a binary tree?
(1) Start at the root
(2) Search the tree level
by level, until you find
the element you are
searching for or you reach
a leaf.

Is this better than searching a linked list?

No  O(N)
 Binary Search Trees (BSTs)
• Binary Search Tree Property:
The value stored at
a node is greater than
the value stored at its
left child and less than
the value stored at its
right child
 Binary Search Trees (BSTs)

In a BST, the value

stored at the root of
a subtree is greater
than any value in its
left subtree and less
than any value in its
right subtree!
 Binary Search Trees (BSTs)

Where is the
smallest element?
Ans: leftmost element

Where is the largest

element?
Ans: rightmost element
How to search a binary search tree?
(1) Start at the root
(2) Compare the value of
the item you are
searching for with
the value stored at
the root
(3) If the values are
equal, then item
found; otherwise, if it
is a leaf node, then
not found
How to search a binary search tree?
(4) If it is less than the value
stored at the root, then
search the left subtree
(5) If it is greater than the
value stored at the root,
then search the right
subtree
(6) Repeat steps 2-6 for the
root of the subtree chosen
in the previous step 4 or 5
How to search a binary search tree?

Is this better than searching

a linked list?

Yes !! ---> O(logN)

Tree node structure

template<class ItemType>
struct TreeNode<ItemType> {
ItemType info;
TreeNode<ItemType>* left;
TreeNode<ItemType>* right;
};
Binary Search Tree Specification
#include <fstream.h>

struct TreeNode<ItemType>;

enum OrderType {PRE_ORDER, IN_ORDER, POST_ORDER};

template<class ItemType>
class TreeType {
public:
TreeType();
~TreeType();
TreeType(const TreeType<ItemType>&);
void operator=(const TreeType<ItemType>&);
void MakeEmpty();
bool IsEmpty() const;
bool IsFull() const;
int NumberOfNodes() const;
Binary Search Tree Specification
(cont.)
void RetrieveItem(ItemType&, bool& found);
void InsertItem(ItemType);
void DeleteItem(ItemType);
void ResetTree(OrderType);
void GetNextItem(ItemType&, OrderType, bool&);
void PrintTree(ofstream&) const;
private:
TreeNode<ItemType>* root;
};
 Function NumberOfNodes
• Recursive implementation
#nodes in a tree =
#nodes in left subtree + #nodes in right subtree + 1

• What is the size factor?

Number of nodes in the tree we are examining
• What is the base case?
The tree is empty
• What is the general case?
CountNodes(Left(tree)) + CountNodes(Right(tree)) + 1
 Function NumberOfNodes
template<class ItemType>
(cont.)
int TreeType<ItemType>::NumberOfNodes() const
{
return CountNodes(root);
}

template<class ItemType>
int CountNodes(TreeNode<ItemType>* tree)
{
if (tree == NULL) Running Time?
return 0; O(N)
else
return CountNodes(tree->left) + CountNodes(tree->right) + 1;
}
Function Retrieve Item
 Function Retrieve Item
• What is the size of the problem?
Number of nodes in the tree we are examining
• What is the base case(s)?
1) When the key is found
2) The tree is empty (key was not found)
• What is the general case?
Search in the left or right subtrees
 Function Retrieve Item (cont.)
template <class ItemType>
void TreeType<ItemType>:: RetrieveItem(ItemType& item, bool& found)
{
Retrieve(root, item, found);
}

template<class ItemType>
void Retrieve(TreeNode<ItemType>* tree, ItemType& item, bool& found)
{
if (tree == NULL) // base case 2
found = false;
else if(item < tree->info) Running Time?
Retrieve(tree->left, item, found);
else if(item > tree->info) O(h)
Retrieve(tree->right, item, found);
else { // base case 1
item = tree->info;
found = true;
}
}
 Function
Insert Item
• Use the
binary search
tree property
to insert the
new item at
the correct
place
Function Insert
Item
(cont.)

• Implementing
insertion
resursively
e.g., insert 11
Construct Binary Search tree using the following numbers

45, 56, 32, 14, 27, 18, 12, 50, 66, 49, 61, 7, 10, 76
 Function InsertItem (cont.)
• What is the size of the problem?
Number of nodes in the tree we are examining
• What is the base case(s)?
The tree is empty
• What is the general case?
Choose the left or right subtree
Function InsertItem (cont.)
template<class ItemType>
void TreeType<ItemType>::InsertItem(ItemType item)
{
Insert(root, item);
}

template<class ItemType>
void Insert(TreeNode<ItemType>*& tree, ItemType item)
{
if(tree == NULL) { // base case
tree = new TreeNode<ItemType>;
tree->right = NULL;
tree->left = NULL;
tree->info = item;
} Running Time?
else if(item < tree->info)
Insert(tree->left, item);
else
O(h)
Insert(tree->right, item);
}
Function InsertItem (cont.)
Insert 11
 Does the order of inserting
elements into a tree matter?
• Yes, certain orders might produce very
unbalanced trees!
Does the
order of
inserting
elements into
a tree matter?
(cont.)
 Does the order of inserting
elements into a tree matter?
(cont’d)
• Unbalanced trees are not desirable because
search time increases!
• Advanced tree structures, such as red-black
trees, guarantee balanced trees.
 Function DeleteItem

• First, find the item; then, delete it

• Binary search tree property must be
preserved!!
• We need to consider three different cases:
(1) Deleting a leaf
(2) Deleting a node with only one child
(3) Deleting a node with two children
(1) Deleting a leaf
(2) Deleting a node with
only one child
(3) Deleting a node with two
children
(3) Deleting a node with two
children (cont.)
• Find predecessor (i.e., rightmost node in the
left subtree)
• Replace the data of the node to be deleted
with predecessor's data
• Delete predecessor node
 Function DeleteItem (cont.)
• What is the size of the problem?
Number of nodes in the tree we are examining
• What is the base case(s)?
Key to be deleted was found
• What is the general case?
Choose the left or right subtree
Make use of the Binary tree perform the following deletion operation

45, 56, 32, 14, 27, 18, 12, 50, 66, 49, 61, 7, 10, 76

Delete 10
Delete 50
Delete 14
 Function DeleteItem (cont.)
template<class ItemType>
void TreeType<ItmeType>::DeleteItem(ItemType item)
{
Delete(root, item);
}

template<class ItemType>
void Delete(TreeNode<ItemType>*& tree, ItemType item)
{
if(item < tree->info)
Delete(tree->left, item);
else if(item > tree->info)
Delete(tree->right, item);
else
DeleteNode(tree);
}
 Function DeleteItem (cont.)
template <class ItemType>
void DeleteNode(TreeNode<ItemType>*& tree)
{
ItemType item;
TreeNode<ItemType>* tempPtr;

tempPtr = tree;
if(tree->left == NULL) { // right child
tree = tree->right; 0 children or
delete tempPtr;
} 1 child
else if(tree->right == NULL) { // left child
tree = tree->left;
delete tempPtr; 0 children or
} 1 child
else {
GetPredecessor(tree->left, item);
tree->info = item;
Delete(tree->left, item); 2 children
}
}
 Function DeleteItem (cont.)
template<class ItemType>
void GetPredecessor(TreeNode<ItemType>* tree, ItemType& item)
{
while(tree->right != NULL)
tree = tree->right;
item = tree->info;
}
Comparing Binary Search Trees to Linear Lists
Big-O Comparison
Binary Array- Linked
Operation
Search Tree based List List
Constructor O(1) O(1) O(1)
Destructor O(N) O(1) O(N)
IsFull O(1) O(1) O(1)
IsEmpty O(1) O(1) O(1)
RetrieveItem O(logN)* O(logN) O(N)
InsertItem O(logN)* O(N) O(N)
DeleteItem O(logN)* O(N) O(N)
*assuming h=O(logN)
 Finding the Minimum and
Maximum
In the Binary Search Tree the minimum is the left most element
and the maximum is the right most element
Exercises 37-41 (p. 539)
Exercise 17 (p. 537)
Exercise 18 (p. 537)