Trees

CMSC 420: Lecture 5

Hierarchies
Many ways to represent tree-like information:
A B D E F C G

linked hierarchy

I. A 1. B a. D i. E b. F 2. C a. G

outlines, indentations

A E D G C F B

(((E):D), F):B, (G):C):A nested, labeled parenthesis nested sets

Denition Rooted Tree

! is a tree If T1, T2, ..., Tk are trees with roots r1, r2, ..., rk and r is a node ! any Ti, then the structure that consists of the Ti, node r, and edges (r, ri) is also a tree. r3 r2 r4 T3 r1 r2 r r3 r4

r1 T1

T4

T2

Terminology

r is the parent of its children r1, r2, ..., rk. r1, r2, ..., rk are siblings. root = distinguished node, usually drawn at top. Has no parent. If all children of a node are !, the node is a leaf. Otherwise, the node is a internal node. A path in the tree is a sequence of nodes u1, u2, ..., um such that each of the edges (u, ui+1) exists. A node u is an ancestor of v if there is a path from u to v. A node u is a descendant of v if there is a path from v to u.

root internal node

path from root to w parent of w

children of u (they are siblings)

w
leaves

Height & Depth

The height of node u is the length of the longest path from u to a leaf. The depth of node u is the length of the path from the root to u. Height of the tree = maximum depth of its nodes. A level is the set of all nodes at the same depth.

Depth = 0 Depth = 1 Depth = 2 Depth = 3 0 1 0

3 2 0 0 1 0 0

Numbers in nodes give heights

Subtrees, forests, and graphs

A subtree rooted at u is the tree formed from u and all its descendants. A forest is a (possibly empty) set of trees. The set of subtrees rooted at the children of r form a forest. As weve dened them, trees are not a special case of graphs:

Our trees are oriented (there is a root which implicitly denes directions on the edges). A free tree is a connected graph with no cycles.

Alternative Denition Rooted Tree

A tree is a nite set T such that:

one element r " T is designated the root. the remaining nodes are partitioned into k ! 0 disjoint sets T1, T2, ..., Tk, each of which is a tree.

This denition emphasizes the partitioning aspect of trees: As we move down the were dividing the set of elements into more and more parts. Each part has a distinguished element (that can represent it).

Basic Properties

Every node except the root has exactly one parent.

A tree with n nodes has n-1 edges (every node except the root has an edge to its parent).

There is exactly one path from the root to each node. (Suppose there were 2 paths, then some node along the 2 paths would have 2 parents.)

Binary Trees !Denition

An ordered tree is a tree for which the order of the children of each node is considered important.
r r

r1

r2

r3

r4

"

r4

r3

r1

r2

A binary tree is an ordered tree such that each node has ! 2 children. Call these two children the left and right children.

Example Binary Trees

The edge cases: ! Empty Binary Tree

Only left child

Only right child

Single node

Small binary tree:

Extended Binary Trees

Replace each missing child with external node Do you need a special ag to tell which nodes are external? Binary tree Every internal node has exactly 2 children. Every leaf (external node) has exactly 0 children. Each external node corresponds to one ! in the original tree lets us distinguish different instances of !. Extended binary tree

# of External Nodes in Extended Binary Trees

Thm. An extended binary tree with n internal nodes has n+1 external nodes. Proof. By induction on n. X(n) := number of external nodes in binary tree with n internal nodes. Base case: X(0) = 1 = n + 1. Induction step: Suppose theorem is true for all i < n. Because n # 1, we have: Extended binary tree X(n) $= X(k) + X(n-k-1) $ = k+1 + n-k-1 + 1 $ =n+1
n-k-1 nodes k nodes (for some 0 ! k < n)

Related to Thm 5.2 in your book.

Alternative Proof
Thm. An extended binary tree with n internal nodes has n+1 external nodes.

Proof. Every node has 2 children pointers, for a total of 2n pointers. Every node except the root has a parent, for a total of n - 1 nodes with parents. These n - 1 parented nodes are all children, and each takes up 1 child pointer. (pointers) - (used child pointers) = (unused child pointers) 2n - (n-1) = n + 1 Thus, there are n + 1 null pointers. Every null pointer corresponds to one external node by construction.

Full and Complete Binary Trees

If every node has either 0 or 2 children, a binary tree is called full. If the lowest d-1 levels of a binary tree of height d are lled and level d is partially lled from left to right, the tree is called complete. If all d levels of a height-d binary tree are lled, the tree is called perfect.

full

complete

perfect

# Nodes in a Perfect Tree of Height h

Thm. A perfect tree of height h has 2h+1 - 1 nodes.
Proof. By induction on h. Let N(h) be number of nodes in a perfect tree of height h. Base case: when h = 0, tree is a single node. N(0) = 1 = 20+1 - 1. Induction step: Assume N(i) = 2i+1 - 1 for 0 ! i < h. A perfect binary tree of height h consists of 2 perfect binary trees of height h-1 plus the root: N(h) $= 2 % N(h - 1) + 1 $ = 2 % (2h-1+1 - 1) + 1 $ = 2 % 2h - 2 + 1 $ = 2h+1 - 1

2h are leaves 2h - 1 are internal nodes

Full Binary Tree Theorem

Thm. In a non-empty, full binary tree, the number of internal nodes is always 1 less than the number of leaves. Proof. By induction on n. L(n) := number of leaves in a non-empty, full tree of n internal nodes. Base case: L(0) = 1 = n + 1. Induction step: Assume L(i) = i + 1 for i < n. Given T with n internal nodes, remove two sibling leaves. T has n-1 internal nodes, and by induction hypothesis, L(n-1) = n leaves. Replace removed leaves to return to tree T. Turns a leaf into an internal node, adds two new leaves. Thus: L(n) = n + 2 - 1 = n + 1. Thm 5.1 in your book.

Array Implementation for Complete Binary Trees

A

1
B

2
E F

3
G Mapping of nodes to integers

4
I J

A B C D E
1 2 3 4 5

F G H
6 7 8

I
9

J
10

K L M
11 12 13 14 15

left(i): 2i if 2i ! n otherwise 0 right(i): (2i + 1) if 2i + 1 ! n otherwise 0 parent(i): #i/2$ if i # 2 otherwise 0

Binary Tree ADT

A tree can be represented as a linked collection of its nodes: template <class ValType> class BinaryTree { public: virtual ValType & value() = 0; virtual void set_value(const ValType &) = 0; virtual BinaryTree * left() const = 0; virtual void set_left(BinNode *) = 0; virtual BinaryTree * right() const = 0; virtual void set_right(BinNode *) = 0; virtual bool is_leaf() = 0; }; virtual % this function can be overridden by subclassing. = 0 % a pure function with no implementation. Must subclass to get implementation.

Linked Binary Tree Implementation

template <class ValType> class BinNode : public BinaryTree<ValType> { public: BinNode(ValType * v); ~BinNode(); ValType & value(); void set_value(const ValType&); BinNode * left() const; void set_left(BinNode *); BinNode * right() const; void set_right(BinNode *); bool is_leaf(); private: ValType * _data; BinNode<ValType> * _left_child; BinNode<ValType> * _right_child; }; _data _left_child _right_child

Binary Tree Representation

_data _left_child _right_child

_data _left_child _right_child

_data _left_child _right_child

0
_data _left_child _right_child _data _left_child _right_child

List Representation of General Trees

A

A children
Child next Child next Child 0

D children C children B children

Child next Child next Child 0 Child 0 Child 0

Representing General Trees with Binary Trees

A A

General K-ary Tree Each node represented by:

Representation as Binary Tree

_data _rst_child _right_sibling

How would you implement an ordered general tree using a binary tree?