Lecture 016

Re-think about the complexity

• loopup has complexity of O(h) where h is the height of the tree

• h can be in O(n) or in O(log n)

Balanced Tree: $h \in O(\log n)$

• 

• lookup, insert, find_min = O(log n)

Self-balancing Tree: a tree remains balanced as we insert new nodes

• AVL trees

• Red-black trees

• Splay trees

• B-trees

Goal: additional representation invariants

1. algorithmically-checkable O(log n)
2. cheap to maintain O(log n)

AVL Trees

Height invariant: at every node, the height of the left and right sub-trees differ by at most 1

How to identify AVL Trees

• count from root to leaf

• count from left to right

• every path should have length <=1 in differ

Insertion in AVL Trees

Approach: two invariant

• insert one but break one of invariant

  • preserve order
  • break height

• patch the broken invariant

If a operation requires insertion and rotation, then the height of a tree never change

These rotation requires the heaviest tree is from outside nodes

• each type of rotation only cost O(1)

• at most 1 (double) rotation for each insertion

• you use double rotate when the heaviest tree in the middle

• you should fix the bottom violation first

• the heaviest tree is determined by the following diagram (notice the end note don't get counted, because we only trace through 2 segments)

  

Implementation

Traversal