Table of Contents
CSE 589 -- Lecture 5
Plan for Today
Easy vs. Hard Problems
Polynomial Time Reductions
Essence of NP-completeness
Definitions, etc.
Proving a Problem is NP-complete
Remarkable Theorem of Steve Cook (1971)Proved that there exist NP-complete problems
Idea of Proof
Traveling Salesman Problem
Clique
Vertex Coloring
Integer Linear Programming
Techniques for Dealing with NP-complete Problems
PPT Slide
BacktrackingExample: Finding a 3-coloring of a graph
Graph Recursion Tree
Branch-and-Bound
Example: Minimum number of colors in graph coloring
Branch-and-Bound
How might you obtain such a bound?
Bounds for clique problem
Two things you can do with linear program
PPT Slide
Approximation Algorithms
Example 1: Vertex Cover
The Vertex Cover Produced Is At Most Twice The Size Of The Optimal VC
Example 2: Approximation Algorithm for Euclidean Traveling Salesman Problem
Approximation Algorithm For Euclidean TSP
PPT Slide
Better version of algorithm
PPT Slide
PPT Slide
This algorithm has provably performance guarantee
PPT Slide
PPT Slide
Randomized Rounding
Example: Global Wiring In A Gate Array
Simplifications
Cast as Integer Program
Randomized Rounding
PPT Slide
Local Search Algorithms
Local Search Procedure for TSP
Does Greedy Local Search Lead Eventually To Optimal Tour?
Solution Spaces
Other Types of Local Moves For TSP Used
PPT Slide
Tabu Search
Augment Greedy Local Search on TSP with Ideas From Tabu Search Paradigm
Limit Acceptance of Moves that Worsen Cost of Current Tour
PPT Slide
Simulated Annealing
Notation
Metropolis Algorithm
What’s going on?
Some Details
Your turn to do an NP-completeness proof
Your turn to analyze an approximation algorithm for an NP-complete problem.
|