Amortized Analysis (chap. 17).ppt

Amortized costs: more cases insert, delete α ≥ 1/2, α 1/2 (use αi, since α can vary a lot) size does/doesn’t change Amortized cost for each operation Splay tree A binary search tree (not balanced) Height may be larger than log n, even n-1. However a sequence of n operations takes O(nlog n). Assumptions: data values are distinct and form a totally order set Operations: Member(i,S) Insert(i,S) Delete(i,S) Merge(S,S’) Split(i,S) All based on splay(i,S), reorganize tree so that i to be root if i?S, otherwise, the new root is either max{k ?S |ki} or min{k ?S |ki} Splay tree (cont.) For examples, merge(S,S’) Call Splay(?, S) and then make S’ the right child Delete(i,S), call Splay(i,S), remove I, then merge(left(i), right(i)). Similar for others. Constant number of splays called. Splay tree (cont.) Splay operation is based on basic rotate(x) operation (either left or right). Three cases: y is the parent of x and x has not grandparent rotate(x) x is the left (or right) child of y and y is the left (or right) child of z, rotate(y) and then rotate(x)\ x is the left (or right) child of y and y is the right (or left) child of z, rotate(x) and then rotate(x) Splay tree (cont.) Credit invariant: Node x always has at least log ?(x) credits on deposit. Where ?(S)=log (|S|) and ?(x)=?(S(x)) Lemma: Each operation splay(x,S) requires no more than 3(?(S)-?(x))+1 credits to perform the operation and maintain the credit invariant. Theorem: A sequence of m operations involving n inserts takes time O(mlog(n)). Summary Amortized analysis Different from probabilistic analysis Three methods and their differences how to analyze Amortized Analysis (chap. 17) Not just consider one operation, but a sequence of operations on a given data structure. Average cost over a sequence of operations. Probabilistic analysis: Average case running time: average over all possible inputs for one algorithm (operation). If using probability, called expected running time. Amortized analysis: No in