on strongly regular graphs attaining the claw bound
play

On strongly regular graphs attaining the claw bound M. Ma caj - PowerPoint PPT Presentation

On strongly regular graphs attaining the claw bound M. Ma caj Comenius University, Bratislava, Slovakia Villanova, June 5th 2014 Strongly regular graphs A k -regular graph on n vertices is strongly regular with parame- ters ( n, k, , )


  1. On strongly regular graphs attaining the claw bound M. Maˇ caj Comenius University, Bratislava, Slovakia Villanova, June 5th 2014

  2. Strongly regular graphs A k -regular graph on n vertices is strongly regular with parame- ters ( n, k, λ, µ ) if a ∼ b ⇔ | N ( a ) ∩ N ( b ) | = λ, a �∼ b ⇔ | N ( a ) ∩ N ( b ) | = µ. (Usually) a SRG has three distinct eigenvalues k > r > s , r > 0 and s < − 1 and all the eigenvalues are integers.

  3. Partial geometries (according to Bose) A partial geometry PG ( R, K, T ) is a triple ( V, B, I ) of sets V and B , and the relation I of incidence on V × B such that A1 Any two points lie on at most one line. A2 Each point lies on exactly R lines. A3 Each line has exactly K points. A4 If the point P does not lie on line l , then there are exactly T lines containing P which intersects l .

  4. An example A Steiner triple system of order n is a pair ( V, B ) of sets such � V � that | V | = n , B ⊆ , and each pair of vertices lies exactly in 3 one triple of B . A STS ( n ) exists iff n mod 6 ∈ { 1 , 3 } . Each STS ( n ) is a partial geometry PG (( n − 1) / 2 , 3 , 3) ( I is the ∈ relation).

  5. The dual geometry If ( V, B, I ) is a PG ( R, K, T ), then ( B, V, I R ) is a PG ( K, R, T ). It is the dual geometry of ( V, B, I ). The dual geometry of a STS ( n ) has parameters (3 , ( n − 1) / 2 , 3).

  6. The point graph of a geometry The point graph of a partial geometry ( V, B, I ) is the graph on V in which two points are adjacent iff they are collinear. The block graph of a partial geometry is the point graph of the dual geometry.

  7. The point graph of a geometry Theorem (Bose, 1963). The point graph of a PG ( R, K, T ) is strongly regular with parameters n = K (1 + ( K − 1)( R − 1) ) , k = R ( K − 1) , T λ = K − 2 + ( R − 1)( T − 1) , µ = RT, and eigenvalues r = K − 1 − T, s = − R.

  8. An example The point graph of a STS ( v ) is complete. The block graph of a STS ( v ) has parameters n = v ( v − 1) k = 3 v − 9 , , 6 2 λ = v + 3 , µ = 9 . 2

  9. (Pseudo-)geometric SRGs A SRG ( n, k, λ, µ ) is pseudo-geometric with characteristic ( R, K, T ) if n = K (1 + ( K − 1)( R − 1) ) , k = R ( K − 1) , T λ = K − 2 + ( R − 1)( T − 1) , µ = RT. A pseudo-geometric SRG is geometric if it is the point graph of a PG ( R, K, T ).

  10. The result of Bose Theorem (Bose 1963). Any pseudo-geometric SRG with char- acteristic ( R, K, T ) is geometric if K > 1 2[ R ( R − 1) + T ( R + 1)( R 2 − 2 R + 2)] . (1) Corollary. Any SRG with parameters of the block graph of a STS ( v ) is geometric if v ≥ 69 .

  11. The improvement of Neumaier Theorem (Neumaier 1979). Let Γ be a SRG ( n, k, λ, µ ) with eigenvalues k > r > s . If r > − 1 + 1 2 s ( s + 1)( µ + 1) , (2) then Γ is geometric and µ ∈ { s ( s + 1) , s 2 } (equivalently, T ∈ { R − 1 , R } ). Remark If Γ is pseudo-geometric with characteristic ( R, K, T ), then (1) and (2) are equivalent.

  12. Overview of the proof: basic ideas • Maximal cliques in (SRGs) have small intersection. • Large maximal cliques are edge disjoint. • If the size(= number of leaves) of the largest claw with root v is small, then there are large maximal cliques containing v .

  13. Overview of the proof: Bose • If STC 1 ∗ are satisfied and the size of the largest claw with root v is − s = R , then the neighborhood of v can be decom- posed into − s maximal cliques of size k/ ( − s ). • Moreover, if the size of the largest claw with root v is − s for all vertices in Γ, then the above cliques form blocks of a PG ( R, K, T ). • If K > [ R ( R − 1) + T ( R + 1)( R 2 − 2 R + 2)] / 2, then STC 1 are satisfied and the size of the largest claw with root v is − s = R for all vertices in Γ. ( STC 1 is a set of technical conditions.)

  14. Overview of the proof: Neumaier • If we replace (1) by (2), then we can drop the assumption that the graph is pseudo-geometric. • If parameters of a PG ( R, K, T ) satisfy (1), then T ∈ { R, R − 1 } .

  15. The claw bound The condition r ≤ − 1 + 1 2 s ( s + 1)( µ + 1) is known as the claw bound . From now on we will consider only SRGs for which the claw bound is attained, that is, for which the condition r = − 1 + 1 2 s ( s + 1)( µ + 1) (3) holds.

  16. Graphs attaining the claw bound: parameters Lemma. Let Γ be a SRG for which the claw bound is attained. Then, 1 2( µ ( − s 3 − s 2 + 2) − s 3 − s 2 + 2 s ) , k = 1 2( µ ( s 2 + s + 2) + s 2 + 3 s ) − 1 , = λ

  17. Graphs attaining the claw bound: s = − 2 • (10 , 3 , 0 , 1) – Petersen, • (16 , 6 , 2 , 2) – Shrikhande + 1 × geometric, • (28 , 12 , 6 , 4) – 3 × Chang + T (8), • (64 , 30 , 18 , 10) – absolute bound.

  18. Graphs attaining the claw bound: s ≤ − 3 Theorem. Let Γ be a SRG ( n, k, λ, µ ) with eigenvalues k > r > s . If s ≤ − 3 and r = − 1 + 1 2 s ( s + 1)( µ + 1) , then Γ is geometric and µ ∈ { s ( s + 1) , s 2 } . Corollary. Any SRG with parameters of the block graph of a STS (67) (that is, (737 , 96 , 35 , 9) ) is geometric.

  19. Overview of the proof • Claw of size − s + 1 may appear in unique way. • If STC 2 ∗ are satisfied and the size of the largest claw with root v is − s + 1, then the neighborhood of v can be decom- posed into − s + 1 maximal cliques of size k/ ( − s + 1). • Moreover, if the size of the largest claw with root v is − s + 1 for all vertices in Γ, then the above cliques form blocks of a PG ( R ′ , K ′ , T ′ ) with R ′ = − s + 1. ( STC 2 is another set of technical conditions.)

  20. Overview of the proof (cont.) • STC 1 are satisfied. • If s ≤ − 4, then STC 2 are satisfied. • Either the size of the largest claw with root v is − s for all vertices in Γ or the size of the largest claw with root v is − s + 1 for all vertices in Γ. • There are no claws of size − s + 1.

  21. Overview of the proof: s = − 3 • Feasible parameters are obtained only for µ ∈ { 1 , 6 , 9 } . If µ = 1, then STC 2 are satisfied. For the remaining values of µ we show that STC 2 are not necessary. • For µ = 6 it follows from an elementary counting argument. • For µ = 9 we use the following. Proposition. Let ∆ be a triangle-free 12 -valent graph on at most 32 vertices such that the second largest eigenvalue of ∆ is ≤ 2 . Then, ∆ is bipartite.

  22. Infinite geometric family: µ = s 2 1 2( − s 5 − s 4 − s 3 + s 2 + 2 s ) , k = 1 2( s 4 + s 3 + 3 s 2 + 3 s ) − 1 . λ = • Known for s ≤ − 13 (Metsch 1995). • New for s ≥ − 11.

  23. Infinite geometric family: µ = s 2 + s 1 2( − s 5 − 2 s 4 − 2 s 3 + s 2 + 4 s ) , k = 1 2( s 4 + 2 s 3 + 4 s 2 + 5 s ) − 1 . λ = • Known (Metsch 1991).

  24. Infinite non-geometric family: µ = 1 − s 3 − s 2 + s + 1 , k = s 2 + 2 s. λ = • Known for s � = − 4 (Bose and Dowling 1971). • Known for s = − 4 (Bagchi 2006) (parameters are (1666 , 45 , 8 , 1)).

  25. Infinite non-geometric family: µ = − s − 2 1 2( s 4 + 2 s 3 + s 2 − 4) , k = 1 2( − s 3 − 2 s 2 − s − 6) . λ = • Known for s = − 3 (Bose and Dowling 1971) ( µ = 1). • Known for s = − 4 (Brouwer and Neumaier 1988) (parame- ters are (1961 , 70 , 15 , 2)). • New for s ≤ − 5.

  26. Infinite non-geometric family: s = − µ 5 − 3 µ 4 − µ 3 + 11 µ 2 + 10 µ − 8 . 8 • Known for µ = 2 (Brouwer and Neumaier 1988) (parameters are (1961 , 70 , 15 , 2)). • New for µ ≥ 3.

  27. Sporadic examples • There are 71 other parameter sets with s ≥ − 750000, • none is geometric, • only five of them have µ ≥ − s . µ s 36 − 30 176 − 34 154 − 43 1905 − 254 850 − 390

  28. n k λ µ r s 32180016 8246 442 2 458 − 18 165989881 41790 2079 10 2089 − 20 560562256 59130 2254 6 2274 − 26 830690560 92934 3558 10 3574 − 26 622523049 56840 2083 5 2105 − 27 1487052121 117720 4039 9 4059 − 29 6260912087 482856 16100 36 16094 − 30 2206382116 95340 2774 4 2804 − 34 62862845606 3376240 99438 176 99296 − 34 13027595348 517465 13968 20 13985 − 37 229749888277 6018606 140075 154 139964 − 43 3 ∗ − 49 ∗ 17344819251 230450 4657 4703 95539730521 1430520 28019 21 28049 − 51 1453875051457 2969136 31165 6 31254 − 95

  29. Thank You

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