1 / 26

Graph Orientations and Submodular Flows

Graph Orientations and Submodular Flows. Lecture 6: Jan 26. Outline. Graph connectivity Graph orientations Submodular flows Survey of results Open problems. Edge Disjoint Paths. s . t .

davis
Download Presentation

Graph Orientations and Submodular Flows

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. Graph Orientations and Submodular Flows Lecture 6: Jan 26

  2. Outline • Graph connectivity • Graph orientations • Submodular flows • Survey of results • Open problems

  3. Edge Disjoint Paths s t [Menger 1927] maximum number of edge disjoint s-t paths = minimum size of an s-t cut.

  4. Graph Connectivity (Robustness) A graph is k-edge-connected if removal of any k-1 edges the remaining graph is still connected. (Connectedness) A graph is k-edge-connected if any two vertices are linked by k edge-disjoint paths. By Menger, these two are equivalent.

  5. Graph Connectivity (Robustness) A graph is k-vertex-connected if removal of any k-1 vertices the remaining graph is still connected. (Connectedness) A graph is k-vertex-connected if any two vertices are linked by k internally vertex-disjoint paths. Are these two are equivalent? Yes, again by Menger!

  6. Vertex Connectivity G’ G v v- v+ k internally vertex disjoint s-t paths in G  k edge disjoint s-t paths in G’

  7. An Inductive Proof of Menger’s Theorem [Menger] maximum number of edge disjoint s-t paths = minimum size of an s-t cut. (Proof by contradiction) Consider a counterexample G with minimum number of edges. So, every edge of G is in some minimum s-t cut

  8. An Inductive Proof of Menger’s Theorem Claim: there is no edge between two vertices in V(G)-{s,t}

  9. An Inductive Proof of Menger’s Theorem s t s t G’ G x x edge-splitting at x So, in G, the only edges are between s and t. But then Menger’s theorem must be true, a contradiction. Conclusion, G doesn’t exist!

  10. Graph Orientations Scenario: Suppose you have a road network. For each road, you need to make it into an one-way street. Question: Can you find a direction for each road so that every vertex can still reach every other vertex by a directed path? What is a necessary condition?

  11. Robbin’s Theorem [Robbins 1939]G has a strongly connected orientation  G is 2-edge-connected

  12. A Useful Inequality d(X) + d(Y) ≥ d(X ∩ Y) + d(X U Y) We call such function a submodular function.

  13. Minimally k-edge-connected graph Claim: A minimally k-ec graph has a degree k vertex. Another cut of size k A smallest cut of size k k + k = d(X) + d(Y) ≥ d(X ∩ Y) + d(X U Y) ≥ k + k

  14. A Proof of Robbin’s Theorem By the claim, a minimally 2-ec graph has a degree 2 vertex. G’ G x x Done! G’ G x x

  15. Nash-Williams’ Theorem [Nash-Williams 1960]G has a strongly k-edge-connected orientation  G is 2k -edge-connected

  16. Mader’s Edge Splitting-off Theorem edge-splitting at x G’ G x x A suitable splitting at x, if for every pair a,b  V(G)-x, # edge-disjoint a,b-paths in G = # edge-disjoint a,b-paths in G’ [Mader] x not a cut vertex, x is incident with 3 edges  there exists a suitable splitting at x

  17. A Proof of Nash-Williams’ Theorem Find a vertex v of degree 2k. Keep finding suitable splitting-off at v for k times. Apply induction. Reconstruct the orientation.

  18. Submodular Flows [Edmonds Giles 1970] Can Find a minimum cost such flow in polytime if g is a submodular function.

  19. Minimum Cost Flows • For sets that contain s but not t, g(X) = -k. • For sets that contain s but not t, g(X) = k. • Otherwise, g(X) = 0. g is submodular.

  20. Problems Recap Stable matchings Bipartite matchings Minimum spanning trees General matchings Maximum flows Shortest paths Minimum Cost Flows Submodular Flows Linear programming

  21. Frank’s approach [Frank] First find an arbitrary orientation, and then use a submodular flow to correct it. submodular [Frank] Minimum weight orientation, mixed graph orientation.

  22. V(G)-S – Steiner vertices S-Steiner tree (S-tree) Steiner Tree Packing Given an undirected multigraph G, S  V(G). S – terminal vertices Steiner Tree Packing Find a largest collection of edge-disjoint S-trees

  23. Special Cases [Menger] Edge-disjoint paths [Tutte, Nash-Williams, 1960] Edge-disjoint spanning trees in polynomial time. (Corollary)2k -edge-connected => k edge-disjoint spanning trees

  24. Kriesell’s Conjecture Steiner tree packing is NP complete Kriesell’s conjecture: [1999] 2k-S-edge-connected  k edge-disjoint S-trees

  25. Nash-Williams’ Theorem [Nash-Williams 1960]Strong Orientation Theorem Suppose each pair of vertices hasr(u,v)paths in G. Then there is an orientation D of G such that there arer(u,v)/2paths between u,v in D.

  26. Orientations with High Vertex Connectivity • Can we characterize those graphs which have a high vertex-connectivity orientation? [Jordán] Every 18-vertex-connected graph has a 2-vertex-connected orientation. Frank’s conjecture 1994: A graph G has a k-vc orientation  For every set X of j vertices, G-X is 2(k-j)-edge-connected.

More Related