04/27/09
18:30:17
1
38
CS61B:
Lecture 38
Monday, April 27, 2009
AMORTIZED ANALYSIS
==================
We’ve seen several data structures for which I claimed that the average time
for certain operations is always better than the worstcase time.
These data
structures include hash tables, treebased disjoint sets, and splay trees.
The analysis technique that proves these claims is called _amortized_analysis_.
Amortized analysis is a way of proving that even if an operation is
occasionally expensive, its cost is made up for by other, cheaper occurrences
of the same operation.
The Averaging Method

Let’s look at the averagecase time for inserting an item into a hash table.
Assume that the chains don’t grow longer than O(1), and any operation that
doesn’t resize the table takes O(1) timemore precisely, suppose it takes at
most one second.
The hash table has N buckets and n items.
Suppose the insert operation doubles
the size of the table (before adding the new item) if n = N.
That way, the
new item won’t cause the load factor to exceed one.
Suppose this resizing
operation takes at most 2n seconds.
When resizing is done, N doubles, the new
item is inserted, and n increases by one.
We construct an empty hash table with just one bucket.
After we perform i
insert operations, n = i and N is the smallest power of two >= n.
The total
seconds for _all_ the resizing operations is
2 + 4 + 8 + .
.. + N/4 + N/2 + N = 2N  2.
So the cost of i insert operations is at most i + 2N  2 seconds.
Because
N < 2n = 2i, the i insert operations take O(5i  2) = O(i) time, and so the
_average_ running time of an insertion operation is in O(i)/i = O(1).
We say that the _amortized_running_time_ of insertion is in O(1), even though
the worstcase running time is in Theta(n).
For almost any application, the amortized running time is more important than
the worstcase running time, because the amortized running time determines the
total running time of the application.
The main exceptions are some
applications that require fast interaction (like video games), for which one
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
This is the end of the preview. Sign up
to
access the rest of the document.
This note was uploaded on 02/21/2010 for the course CS 61B taught by Professor Canny during the Spring '01 term at Berkeley.
 Spring '01
 Canny
 Data Structures

Click to edit the document details