Path separability of Graphs Emilie Diot and Cyril Gavoille LaBRI, - - PowerPoint PPT Presentation

path separability of graphs
SMART_READER_LITE
LIVE PREVIEW

Path separability of Graphs Emilie Diot and Cyril Gavoille LaBRI, - - PowerPoint PPT Presentation

Path separability of Graphs Emilie Diot and Cyril Gavoille LaBRI, University of Bordeaux, France 8 novembre 2010 JGA, Luminy Outline Motivation Definition Results What about the 1 -path separable graphs Outline Motivation Definition


slide-1
SLIDE 1

Path separability of Graphs

Emilie Diot and Cyril Gavoille LaBRI, University of Bordeaux, France

8 novembre 2010

JGA, Luminy

slide-2
SLIDE 2

Outline

Motivation Definition Results What about the 1-path separable graphs

slide-3
SLIDE 3

Outline

Motivation Definition Results What about the 1-path separable graphs

slide-4
SLIDE 4

Motivations

To separate graphs in order to apply “Divide and Conquer”

slide-5
SLIDE 5

Motivations

To separate graphs in order to apply “Divide and Conquer” The notion of k-path separability defined by Abraham et

  • al. (PODC’06), to solve “Objects Location Problem”

Compact routing with O(k log2 n)-bit tables Distance labelling with O(k log n log D)-bit labels Navigation in “Small-World” with O(k2 log2 n) hops

slide-6
SLIDE 6

Outline

Motivation Definition Results What about the 1-path separable graphs

slide-7
SLIDE 7

k-path separable graphs.

Intuitively : Separate recursely the input graph with separators composed of at most k shortest paths.

slide-8
SLIDE 8

k-path separable graphs.

Intuitively : Separate recursely the input graph with separators composed of at most k shortest paths.

Definition (k-path separability)

S = P0 ∪ P1 ∪ . . ., where each subgraph Pi is the union

  • f ki minimum cost paths in G \

j<i Pj where

  • i ki k ; and

every connected component of G \ S (if any) is k-path separable and weigth at most ω(G)/2.

slide-9
SLIDE 9

Example : Petersen Graph

slide-10
SLIDE 10

Example : Petersen Graph

slide-11
SLIDE 11

Example : Petersen Graph

slide-12
SLIDE 12

Example : Petersen Graph

slide-13
SLIDE 13

Example : Petersen Graph

slide-14
SLIDE 14

Example : Petersen Graph

⇒ 2-path separable

slide-15
SLIDE 15

Outline

Motivation Definition Results What about the 1-path separable graphs

slide-16
SLIDE 16

Related Works.

Trees are 1-path separable.

slide-17
SLIDE 17

Related Works.

Trees are 1-path separable. K4r is r-path separable.

slide-18
SLIDE 18

Related Works.

Trees are 1-path separable. K4r is r-path separable. Treewidth-k graphs are ⌈(k + 1)/2⌉-path separable.

slide-19
SLIDE 19

Related Works.

Trees are 1-path separable. K4r is r-path separable. Treewidth-k graphs are ⌈(k + 1)/2⌉-path separable.

Theorem (Thorup - FOCS’01/JACM’04)

Planar graphs are 3-path separable.

slide-20
SLIDE 20

Related Works.

Trees are 1-path separable. K4r is r-path separable. Treewidth-k graphs are ⌈(k + 1)/2⌉-path separable.

Theorem (Thorup - FOCS’01/JACM’04)

Planar graphs are 3-path separable.

Theorem (Abraham and Gavoille - PODC ’06)

H-minor free graphs are f(H)-path separable.

slide-21
SLIDE 21

Outline

Motivation Definition Results What about the 1-path separable graphs

slide-22
SLIDE 22

The family of k-path separable graphs

Definition

PSk is the family of graphs that are k-paths separable for every weight function.

slide-23
SLIDE 23

The family of k-path separable graphs

Definition

PSk is the family of graphs that are k-paths separable for every weight function. Trees ⊂ PS1 Treewidth-3 ⊂ PS2 Planar graphs ⊂ PS3

slide-24
SLIDE 24

Minor graphs

A minor of G is a subgraph of a graph obtained from G by edge contraction. A H-minor free graph is a graph without minor H.

slide-25
SLIDE 25

Minor graphs

A minor of G is a subgraph of a graph obtained from G by edge contraction. A H-minor free graph is a graph without minor H.

slide-26
SLIDE 26

Minor graphs

A minor of G is a subgraph of a graph obtained from G by edge contraction. A H-minor free graph is a graph without minor H.

slide-27
SLIDE 27

Minor graphs

Proposition

PSk is closed under minor taking.

slide-28
SLIDE 28

Proof

(G) (H)

slide-29
SLIDE 29

Proof

(G) (H)

slide-30
SLIDE 30

Proof

(G) (H)

slide-31
SLIDE 31

Proof

(G) (H)

slide-32
SLIDE 32

Forbidden minors

Corollaire (Roberston & Seymour)

G ∈ PSk iff G excludes a finite list of “forbidden” minors. Therefore, for constant k, membership for PSk” can be tested in cubic time ... if the list is given.

slide-33
SLIDE 33

Forbidden minors (at least 16)

The unique non-planar graph in PS1 is K3,3.

slide-34
SLIDE 34

Forbidden minors (at least 16)

The unique non-planar graph in PS1 is K3,3. And for planar graphs :

slide-35
SLIDE 35

Particular example

slide-36
SLIDE 36

Forbidden minors

slide-37
SLIDE 37

Perspectives

List all forbidden minors for PS1. What about planar graphs ?

slide-38
SLIDE 38

Thank you

slide-39
SLIDE 39