1 / 27

Randomized Min-Cut

Randomized Min-Cut. Meisam Navaki Arefi Computer Science Department Amirkabir University of Technology October 2010. Outline. Contraction Algorithm Faster Randomized Min Cut Algorithm ( Recursive Contraction Algorithm ) Open Problems References. Contraction Algorithm.

noel
Download Presentation

Randomized Min-Cut

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Randomized Min-Cut Meisam Navaki Arefi Computer Science Department Amirkabir University of Technology October 2010

  2. Outline • Contraction Algorithm • Faster Randomized Min Cut Algorithm ( Recursive Contraction Algorithm ) • Open Problems • References Randomized Min Cut

  3. Contraction Algorithm • Start with H=G • Repeat (n-2) times 1.Choose a random edge e in H 2.Contract edge e • Finally, H has only two vertices Return the resulting cut as the min-cut Randomized Min Cut

  4. Example F A contract B E G C D FG FGD A A B E B E C D C contract Randomized Min Cut

  5. Example ABC EFGD Randomized Min Cut

  6. Analysis • The probability of success : • Probability of Error : Randomized Min Cut

  7. Analysis • Repeat k times, the failure probability is • If K= n(n-1)/2 , this is According to : Failure probability ≤ 1/e (Constant probability ) Randomized Min Cut

  8. Running Time Analysis • Theorem 10.10: The Contraction Algorithm can be implemented to run in O(n² ) time. Randomized Min Cut

  9. Running Time Analysis • W : an n*n weighted adjacency matrix Procedure to contract edge (u , v) Let W(u , v) W( v, u) 0 For each vertex w except u and v Let W(u , w) W(u , w) + W(v , w) Let W(w , u) W(w , u) + W(w , v) Let W(v , w) W(w, v) 0 Time : O(n) Randomized Min Cut

  10. Running Time Analysis • Start with H=G • Repeat (n-2) times 1.Choose a random edge e in H 2.Contract edge e • Finally, H has only two vertices Return the resulting cut as the min-cut Time : O(n² ) Randomized Min Cut

  11. Running Time Analysis • Run the algorithm O(n²) times and expect to get C with constant probability. Total time: • No better than Ford-Fulkerson! Can we do better? Randomized Min Cut

  12. Recursive Contraction Algorithm Randomized Min Cut

  13. IDEA! The situation: n² results, one of which is expected to be good. O(n²) time Randomized Min Cut

  14. What if ? Could compute: n² results, one of which is expected to be good. O(log n) time Randomized Min Cut

  15. Contraction Algorithm But note: The probability of not hitting C at stage 1 is (almost 1). It grows as we run more and more stages. What happens if we stop in the middle? The probability that C not touched in the first i steps ≥ Randomized Min Cut

  16. Recursive Contraction Algorithm • When is the probability of not hitting C exactly ½ ? -- When nodes are left. • If we run two times until nodes are left ,we expect that one of them still did not touch C . • So, for each of these runs, when nodes are left, stop running and recurse ! Randomized Min Cut

  17. Recursive Contraction Algorithm Randomized Min Cut

  18. Running Time Analysis The closed form is: Much better! But what is the overall probability? Randomized Min Cut

  19. Probability • Theorem 10.17 The Recursive Contraction Algorithm finds a particular minimum cut with probability . Randomized Min Cut

  20. Probability • Assume we are in level d of the recursion tree, with the leaves being level 1. • Let Pd(x)be the probability that there is a path from node x at level d that does not touch C. • We have: Pd(x) = ½ * ( Probability that at least one of the two recursions does not touch C) Randomized Min Cut

  21. Probability Recall: Pr(A or B)= Pr(A)+Pr(B)-Pr(A and B) We have: Pd(x) = ½(2 Pd-1 –(Pd-1)²) Where Pd-1 is the probability that a child of x does not hit C. = Pd-1 – ½ (Pd-1)² Claim: For d >1, Pd>1/d Proof: Randomized Min Cut

  22. Probability We have: Pd = Pd-1 – ½ (Pd-1)². Inductively: Pd-1 >1/(d-1). So, because of monotonicity , Pd-1 – ½ (Pd-1)² > Randomized Min Cut

  23. Probability Conclude: Plog n(x) > 1/log n Which means:If we run the algorithm log n times, we get constant probability.So… Total time: O(n² log² n) If we run the algorithm log² n times, we get high probability.So… Total time: O(n² log3n) Randomized Min Cut

  24. Best Min Cut • Best known randomized min cut algorithm: [Karger 2000] O(m log3n). • Faster than best known max flow algorithm ordeterministic global min cut algorithm. Randomized Min Cut

  25. Open Problems • Is it easier to find a minimum cut than a maximum flow in directed graphs ? • How close to linear-time can we get in solving the minimum cut problem in theory and in practice? • Design a fast Las Vegas algorithm for min cut . Randomized Min Cut

  26. References • Randomized Algorithms, R.Motwani ,P. Raghavan ,1995. • D.R.Karger , A New Approach To The Minimum Cut Problme,1996. • D.R.Karger,Minimum Cuts in Near-Linear Time,ACM, Vol. 47, No. 1, January 2000, pp. 46 –76. Randomized Min Cut

  27. HAVE A NICE RANDOMIZED LIFE Randomized Min Cut

More Related