graphs with convex qp stability number
play

Graphs with convex-QP stability number Domingos M. Cardoso - PowerPoint PPT Presentation

Universytet im. Adama Mickiewicza Pozna n, January 2004 Graphs with convex-QP stability number Domingos M. Cardoso (Universidade de Aveiro) 1 Summary Introduction. The class of Q -graphs. Adverse


  1. ✬ ✩ Universytet im. Adama Mickiewicza Pozna´ n, January 2004 Graphs with convex-QP stability number Domingos M. Cardoso (Universidade de Aveiro) ✫ ✪ 1

  2. ✬ ✩ Summary � Introduction. � The class of Q -graphs. � Adverse graphs and ( k, τ )-regular sets. � Analysis of particular families of graphs. � Relations with the Lov´ az’s ϑ -function. � Final remarks and open problems. ✫ ✪ 2

  3. ✬ ✩ Introduction Let us consider the simple graph G = ( V, E ) of order n, where V = V ( G ) is the set of nodes and E = E ( G ) is the set of edges. A G will denote the adjacency matrix of the graph G and λ min ( A G ) the minimum eigenvalue of A G . It is well known that if G has at least one edge, then λ min ( A G ) ≤ − 1 . Actually � λ min ( A G ) = 0 iff G has no edges, � λ min ( A G ) = − 1 iff G has at least one edge and every component complete, √ ✫ ✪ � λ min ( A G ) ≤ − 2 otherwise. 3

  4. ✬ ✩ Introduction (cont.) A graph G is ( H 1 , . . . , H k )-free if G contains no copy of the graphs H 1 , . . . , H k , as induced subgraphs. � In particular, G is H -free if G has no copy of H as an induced subgraph. � A claw-free graph is a K 1 , 3 -free graph. Let us define the quadratic programming problem ( P G ( τ )): e T x − x T (1 υ G ( τ ) = max { 2ˆ τ A G + I n ) x : x ≥ 0 } , with τ > 0 . If x ∗ ( τ ) is an optimal solution for ( P G ( τ )) then 0 ≤ x ∗ ( τ ) ≤ 1 . ✫ ✪ 4

  5. ✬ ✩ Introduction (cont.) ∀ τ > 0 1 ≤ υ G ( τ ) ≤ n. The fucntion υ G :]0 , + ∞ [ �→ [1 , n ] has the following properties: � ∀ τ > 0 α ( G ) ≤ υ G ( τ ) . � 0 < τ 1 < τ 2 ⇒ υ G ( τ 1 ) ≤ υ G ( τ 2 ) . � υ G (1) = α ( G ). � If τ ∗ > 0 , then the following are equivalent. τ ∈ ]0 , τ ∗ [ such that υ G (¯ τ ) = υ G ( τ ∗ ); – ∃ ¯ – υ G ( τ ∗ ) = α ( G ); – ∀ τ ∈ ]0 , τ ∗ [ x ∗ ( τ ) is not spurious; – ∀ τ ∈ ]0 , τ ∗ ] υ G ( τ ) = α ( G ) . � ∀ U ⊂ V ( G ) ∀ τ > 0 υ G − U ( τ ) ≤ υ G ( τ ). ✫ ✪ 5

  6. ✬ ✩ Introduction (cont.) ✉ ✉ ✉ ✉ g j f i ❏ ✡ ❏ ✡ ❏ ✡ ❏ ✡ ❏ ✡ ❏ ✡ ✡ ❏ ✡ ❏ ② ✡ ❏ ② ✉ ② ✡ ❏ ② c ✡ ❏ ✡ ❏ ◗ ✑✑✑✑✑✑✑✑ ◗ a e b d ◗ ◗ ◗ ◗ ✉ ◗ ◗ h Figure 1: A graph G with λ min ( A G ) = − 2 and υ G (2) = α ( G ) = 4. ✫ ✪ 6

  7. ✬ ✩ Introduction (cont.) ✁ � ✂ ✄ ☎ ✆ ✝ ✂ ✄ ✂ ✝ ✆ ✄ ✆ ✝ ✞ ✄ ✟ Figure 2: Function υ G ( τ ), where G is the above graph. ✫ ✪ 7

  8. ✬ ✩ The class of Q -graphs � The graphs G such that υ G ( − λ min ( A G )) = α ( G ) are called graphs with convex- QP stability number where QP means quadratic program. � The class of these graphs will be denoted by Q and its elements called Q -graphs. � Since the components of the optimal solutions of ( P G ( τ )) are between 0 and 1 , then υ G ( τ ) = α ( G ) if and only if ( P G ( τ )) has an integer optimal solution. Theorem [Luz, 1995] If G has at least one edge then G ∈ Q if and only if, for a maximum stable set S (and then for all), − λ min ( A G ) ≤ min {| N G ( i ) ∩ S | : i �∈ S } . (1) ✫ ✪ 8

  9. ✬ ✩ The class of Q -graphs (cont.) There exists an infinite number of graphs with convex- QP stability number. Theorem [Cardoso, 2001] A connected graph with at least one edge, which is nor a star neither a triangle, has a perfect matching if and only if its line graph has convex- QP stability number. As immediate consequence, we have the following corollary. Corollary [Cardoso, 2001] If G is a connected graph with an even number of edges then L ( L ( G )) has convex- QP stability number. ✫ ✪ 9

  10. ✬ ✩ The class of Q -graphs (cont.) There are several famous Q -graphs. � The Petersen graph P , where λ min ( A P ) = − 2 and α ( P ) = υ P (2) = 4. � The Hoffman-Singleton graph HS , where λ min ( A HS ) = − 3 and α ( HS ) = υ HS (3) = 15. � If the fourth graph of Moore M 4 there exists with α ( M 4 ) = 400, as it is expected, then it is a Q -graph. � Additionally, taking into account ( ?? ), graphs defined by the disjoint union of complete subgraphs and complete bipartite graphs are trivial examples of Q -graphs. ✫ ✪ 10

  11. ✬ ✩ The class of Q -graphs (cont.) Additional examples of Q -graphs ③ ✉ ❅ � ❅ � ❅ � � � ❅ � � ✉ ✉ ③ ❅ � � ❅ � � ❅ � ❅ � ❅ � ❅ ③ � ✉ ❅ � ❅ Figure 3: Graph G such that λ min ( A G ) = − 2 and υ G (2) = 3 = α ( G ). ✫ ✪ 11

  12. ✬ ✩ The class of Q -graphs (cont.) ③ ③ ③ ③ ❅ � ❅ � ③ ③ ❅ � ❍ ✟ ❍ ❅ ✟✟✟✟✟ � ❍ ✉ ✉ ❍ ❅ � ❍ ❅ ❍ � ✟ ❍❍❍❍❍ ✟ � ❅ ✟ ③ ③ ✟ � ❅ ✟ ✟ ❍ � ❅ � ❅ ③ ③ ③ ③ � ❅ � ❅ Figure 4: Graph G such that λ min ( A G ) = − 3 and υ G (3) = 12 = α ( G ). ✫ ✪ 12

  13. ✬ ✩ The class of Q -graphs (cont.) � A graph belongs to Q if and only if each of its components belongs to Q . � Every graph G has a subgraph H ∈ Q such that α ( G ) = α ( H ). � If G ∈ Q and ∃ U ⊆ V ( G ) such that α ( G ) = α ( G − U ) then G − U ∈ Q . � If ∃ v ∈ V ( G ) such that υ G ( τ ) � = max { υ G −{ v } ( τ ) , υ G − N G ( v ) ( τ ) } , with τ = − λ min ( A G ), then G �∈ Q . ✫ ✪ 13

  14. ✬ ✩ The class of Q -graphs (cont.) � Consider that ∃ v ∈ V ( G ) such that υ G −{ v } ( τ ) � = υ G − N G ( v ) ( τ ) and τ = − λ min ( A G ) . 1. If υ G ( τ ) = υ G −{ v } ( τ ) then G ∈ Q iff G − { v } ∈ Q . 2. If υ G ( τ ) = υ G − N G ( v ) ( τ ) then G ∈ Q iff G − N G ( v ) ∈ Q . ✫ ✪ 14

  15. ✬ ✩ The class of Q -graphs (cont.) � Assuming that τ 1 = − λ min ( A G ) > − λ min ( A G − U ) = τ 2 , with U ⊂ V ( G ). Then υ G ( τ 1 ) = υ G − U ( τ 2 ) ⇒ G ∈ Q , ⇒ G �∈ Q or U ∩ S � = ∅ , υ G ( τ 1 ) > υ G − U ( τ 2 ) for each maximum stable set S of G . ✫ ✪ 15

  16. ✬ ✩ Adverse graphs and ( k, τ ) -regular sets � Using the above results, we may recognize if a graph G is (or not) a Q -graph, unless an induced subgraph H = G − U (where U ⊂ V ( G ) can be empty) is obtained, such that = λ min ( A G ) = λ min ( A H ) , (2) τ υ G ( τ ) = υ H ( τ ) , (3) ∀ v ∈ V ( H ) λ min ( A H ) = λ min ( A H − N G ( v ) ) , (4) ∀ v ∈ V ( H ) υ H ( τ ) = υ H − N G ( v ) ( τ ) . (5) � A subgraph H of G without isolated vertices, for which the conditions ( ?? )-( ?? ) are fulfilled is called adverse . ✫ ✪ 16

  17. ✬ ✩ Adverse graphs and ( k, τ ) -regular sets (cont.) s s s s ✟ ❍❍❍❍❍❍❍ ✟ ❍❍❍❍❍❍❍ ✟ ❍❍❍❍❍❍❍ ✟ ❍❍❍ ✟✟✟✟✟✟✟ ✟✟✟✟✟✟✟ ✟✟✟✟✟✟✟ ✟✟✟ ✉ ✉ ✉ ✉ ✉ ❍ ❍❍❍ ✟ s s s s ✟✟✟ ❍ ❍ ❍ ❍ Figure 5: Adverse graph G, with λ min ( A G ) = − 2 and υ G (2) = α ( G ) = 5 . ✫ ✪ 17

  18. ✬ ✩ Adverse graphs and ( k, τ ) -regular sets (cont.) � Based in the above results, a procedure which recognizes if a graph G is (or not) in Q or determines an adverse subgraph can be implemented. � A subset of vertices S ⊂ V ( G ) is ( k, τ )- regular if induces in G a k -regular subgraph and ∀ v �∈ S | N G ( v ) ∩ S | = τ. � The maximum stable sets of the graphs of figures 1, ?? and ?? are (0 , 2)-regular and the maximum stable set of the graph of figure ?? is (0 , 6)-regular. ✫ ✪ 18

  19. ✬ ✩ Adverse graphs and ( k, τ ) -regular sets (cont.) � The Petersen graph P includes the (0 , 2)-regular set S = { 1 , 2 , 3 , 4 } and the (2 , 1)-regular sets T 1 = { 1 , 2 , 5 , 7 , 8 } and T 2 = { 3 , 4 , 6 , 9 , 10 } . t 5 t ✑ ◗◗◗ ❞ ✑ ❞ 6 ✑ t t ✑ ✂ ❇ ◗ 1 2 ❍ ✟ ❍ ✟✟✟✟ ❇ ✂ ❇ ✂ ❇ ✂ ❍ ❞ ❞ ❍ ✂ ❇❇ ❇ ✂ 9 10 ❍ ✂ t t ❇ ✂ ✔ ❚ 3 4 ❇ ✔ ❚ ✂ 7 8 Figure 6: The Petersen graph. � L ( P ) includes the (0 , 2)-regular set {{ 1 , 9 } , { 5 , 6 } , { 2 , 10 } , { 4 , 8 } , { 3 , 7 }} (a perfect matching) and the (0 , 1)-regular set {{ 5 , 6 } , { 9 , 10 } , { 7 , 8 }} (a perfect induced matching). ✫ ✪ 19

  20. ✬ ✩ Adverse graphs and ( k, τ ) -regular sets (cont.) Theorem Let G be adverse and τ = − λ min ( A G ) . Then G ∈ Q if and only if ∃ S ⊂ V ( G ) which is (0 , τ ) -regular. Theorem Let G be p -regular, with p > 0 . Then G ∈ Q if and only if ∃ S ⊂ V ( G ) which is (0 , τ ) -regular, with τ = − λ min ( A G ) . Theorem [Thompson, 1981] Let G be a p -regular graph and x ( S ) the characteristic vector of S ⊂ V ( G ) . Then S is ( k, τ ) -regular if and only if e − p − ( k − τ ) x ( S )) ∈ Ker ( A G − ( k − τ ) I n ) , (ˆ τ where ˆ e is the all-ones vector. ✫ ✪ 20

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend