Properties of Binary Trees

often expressed recursively (following the denition of binary tree) depth (or level) of a node:
root has level 1 otherwise 1+ level of parent Denition

Some tree examples:

TA TC TB TD

height of a tree:
if the tree is empty, its height is 0 otherwise, its height is 1 + max{ height TL, height TR }, where TL and TR designate left and right subtrees

Other ways to dene the height of a tree length of the longest root-leaf path max( level(n) ) over all nodes n in the tree So the height of a singleton tree is 1. (Huh?) Think of the height as the max # of nodes on a root-leaf path, not the number of branches.

Code

public static int height( BinaryTreeInterface t ) // Post: Returns height of subtree. if( t.isEmpty() ) return 0; else { int leftHt = height( t.getLeftSubtree() ); int rightHt = height( t.getRightSubtree() ); return 1 + Math.max( leftHt, rightHt ); } }

A tree of height h having the maximum number of nodes (2h-1) is called a perfect tree.

CS 134 JCBeatty

Page 111

Proving Properties About Trees

Theorem: A binary tree of height h has at most 2h -1 nodes. Proof: (by induction on tree height h) base case: h = 0 By denition, h = 0 implies the tree is empty. Thus, there are 0 nodes. Since 20 -1 = 0, the theorem is true for h = 0. inductive hypothesis: Assume the theorem is true for all h < k, for some k > 0. inductive step: Assume T has height k for some value of k > 0. Then both subtrees are of height < k.
T TL TR

Lets try it out on a few trees ... see if its true see how its true
TA TB TC TD

By the inductive hypothesis, TL has at most 2k-1 -1 nodes and TR has at most 2k-1 -1 nodes. Thus T has at most 1+2(2k-1 -1) = 2k -1 nodes. QED or as required (ie were done).

height < k means height at most k-1 (and since k > 0, k-1 0)

CS 134 JCBeatty

Page 112

Inductive Proof Template

Notation: <introduce any notation you need> Theorem: <state it here> Proof: (by induction on <# nodes, height, >) Base case(s): < prove for 1 or more small cases> Inductive Hypothesis: Assume the theorem holds for all cases with n up to k-1 ( ie for all n k-1). Inductive Step: Assume we have a problem (tree) of size n=k. Show: How to apply the inductive hypothesis to the subtrees. That all variations of TL and TR are covered. How to combine the smaller problems to derive the statement in the theorem.

TL

TR

CS 134 JCBeatty

Page 113

Strong Induction
To prove P(n) true for all n 0:
Base case: Show that P(0) is true Inductive hypothesis: Assume property P(i) is true for i = 0,1,...,k-1. Inductive step: Show, using the inductive hypothesis, that P(k) is true.

Variations
base case something other than 0 several base cases assume P(0), P(1), ..., P(k) true and use them to prove P(k+1) is true weak (aka natural) induction: assume P(k-1) true and use it alone to prove P(k) is true. (More common for proving formulas regarding sequences of integers, eg 1 + 2 + 3 + ... + n = n(n+1)/2 ).

The form of the inductive proof usually matches the form of the recursion.

CS 134 JCBeatty

Page 114

Number of Leaves in a Binary Tree

Dene a full node to be a node with exactly two children (ie non-empty subtrees). Theorem: In any non-empty binary tree, the number of leaves is one more than the number of full nodes. Proof:

CS 134 JCBeatty

Page 115