Forbidden Conjectures David Sumner, Professor Emeritus University - - PowerPoint PPT Presentation

forbidden conjectures
SMART_READER_LITE
LIVE PREVIEW

Forbidden Conjectures David Sumner, Professor Emeritus University - - PowerPoint PPT Presentation

Forbidden Conjectures David Sumner, Professor Emeritus University of South Carolina Graph Theory app for iPad coming soon to the Apple app store. ! Email david@dpsumner.com for more information or to request to be a beta-tester. See more


slide-1
SLIDE 1

Forbidden Conjectures

David Sumner, Professor Emeritus University of South Carolina

slide-2
SLIDE 2

See more details at http://www.dpsumner.com Graph Theory app for iPad coming soon to the Apple app store.

!

Email david@dpsumner.com for more information or to request to be a beta-tester.

slide-3
SLIDE 3

Conjecture (Matthews, S, 1984). Every 4-connected, claw-free graph is Hamiltonian.

Conjecture (Gyárfás, S). For every tree T, there exists a function fT such that for every T-free graph G, χ(G) ≤ fT ω(G)

( ).

slide-4
SLIDE 4

50 Years Earlier

slide-5
SLIDE 5

The earliest work that I am aware of for graphs forbidding paths

  • n four vertices is due to Dave Foulis and Charlie Randall at the

University of Massachusetts in their work on Empirical Logic in the early to mid 1960’s.

Graphs with no Induced P

4

So, in particular is not a Dacey graph because

P

4

v u K

They defined a graph to be a Dacey Graph if whenever two vertices dominate a maximal clique, then they must be adjacent.

slide-6
SLIDE 6

The most common terms for these graphs today seems to be either cographs or .

P

4-free

Foulis - Randall defined a graph G to be hereditary Dacey if every induced subgraph of G is a Dacey Graph. It turns out that the hereditary Dacey graphs are precisely those graphs that have no induced path on four vertices.

Graphs with no Induced P

4

P

4 is self-complementary.

slide-7
SLIDE 7

The most common terms for these graphs today seems to be either cographs or .

P

4-free

Foulis - Randall defined a graph G to be hereditary Dacey if every induced subgraph of G is a Dacey Graph. It turns out that the hereditary Dacey graphs are precisely those graphs that have no induced path on four vertices.

Graphs with no Induced P

4

P

4 is self-complementary.

The complement of a cograph is again a cograph.

slide-8
SLIDE 8

Theorem (Foulis - Randall). If G is a cograph, then there exist cographs A and B such that (i). If G is connected, then G = A + B. (ii). If G is not connected, then G = A∪ B.

A B A B

G is connected G is not connected G is the join of A and B. G is the union of A and B.

Graphs with no Induced P4

slide-9
SLIDE 9

A B A B

G is connected G is not connected G is the join of A and B. G is the union of A and B.

Graphs with no Induced P4

Theorem (S, 1971). If S is a maximal independent set of the connected cograph G, then N(S) ≠ ∅ and G = N(S)+ G − N(S)

[ ].

Theorem (S, 1971). If M is a minimal separating set of vertices,

  • f the connected cograph G, then G = M + G − M

[ ].

slide-10
SLIDE 10

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


Graphs with no Induced P4

He did not, of course, use the term cograph or P

4-free.

Theorem (Wolk 1965). Every cograph is a comparability graph.

A B A B A B

Orient each of A and B transitively. Orient edges between A and B from A towards B.

slide-11
SLIDE 11

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


Theorem (Seinsche 1974). Every cograph is perfect. Arditti and DeWerra pointed out in a note in JCT(B) 1976 that Seinsche’s result was really a special case of Wolk’s.

Graphs with no Induced P4

He did not, of course, use the term cograph or P

4-free.

Theorem If G is P

4-free, then χ(G) = ω(G).

Theorem (Wolk 1965). Every cograph is a comparability graph.

slide-12
SLIDE 12

Graphs with no Induced K1,3

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


The other tree on four vertices is the claw and the graphs that do not contain an induced claw are even more interesting than the cographs. The first mention of the claw-free graphs that I saw was in Beineke’s forbidden subgraph characterization of line graphs.

slide-13
SLIDE 13

Graphs with no Induced K1,3

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


Theorem If G is P

4-free, then χ(G) = ω(G).

Theorem If G is a graph, then χ(G) ≤ Δ +1. In general, there is no upper bound on the chromatic number

  • f a graph in terms of ω(G).

Theorem (Erdös) There exist graphs with arbitrarily high girth and chromatic number. Ramsey Number: For positive integers n,m every graph with more than r(n,m) vertices has an independent set on n vertices or a complete subgraph on m vertices.

slide-14
SLIDE 14

An independent set of order 3 would produce a K1, 3 with center v.

Graphs with no Induced K1,3

Theorem If G is P

4-free, then χ(G) = ω(G).

Theorem If G is K1,3-free, then Δ(G) < r(3, ω(G))< ω(G)2 Theorem If G is a graph, then χ(G) ≤ Δ +1.

v

A complete set of order ω would produce a clique on ω +1 vertices.

slide-15
SLIDE 15

Theorem If G is P

4-free, then χ(G) = ω(G).

Theorem If G is K1,3-free, then Δ(G) < r(3, ω(G))< ω(G)2 Theorem If G is a graph, then χ(G) ≤ Δ +1. Theorem If G is P

5-free and triangle-free, then G is 3-colorable.

Theorem If G is K1,3-free, then χ(G) < ω(G)2.

Graphs with no Induced K1,3

Theorem If G is P

6-free, C6-free and triangle-free, then G is 3-colorable.

slide-16
SLIDE 16
  • Theorem. (Scott). If T is any tree, then there exists a function fT

such that every graph that does not contain any subdivision of T as an induced subgraph satisfies χ(G) ≤ fT (ω(G)). Conjecture (Gyárfás, S). For every tree T, there exists a function fT such that for every T-free graph G, χ(G) ≤ fT ω(G)

( ).

  • Theorem. (Gyárfás). The conjecture is true for stars and paths.
  • Theorem. (Kierstead and Penrice). The conjecture is true for trees of radius 2.
  • Theorem. (Kierstead and Zhu). Proved the conjecture for a special

class of radius-three trees.

slide-17
SLIDE 17

Conjecture (Gyárfás, S). For every tree T, there exists a function fT such that for every T-free graph G, χ(G) ≤ fT ω(G)

( ).

If we don't care about the trees being induced, then much stronger results are true. Theorem (Gyárfás, S). Every coloring of a k-chromatic graph using the labels {1,2,…,k} contains a copy of every labelled tree on {1,2,…,k}.

slide-18
SLIDE 18

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


  • Theorem. Every even-order, connected, claw-free graph

has a 1-factor.

Graphs with no Induced K1,3

This can be proved directly or as a corollary to even stronger results.

  • Theorem. If G is an connected claw-free graph, then it is possible

to produce a maximum matching for G by sequentially removing adjacent pairs of vertices so that the graph remains connected after each deletion.

slide-19
SLIDE 19

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


  • Theorem. Every even-order, connected, claw-free graph

has a 1-factor.

Graphs with no Induced K1,3

  • Theorem. Every even-order, k-connected, claw-free graph

with no induced K1,k+1 has a 1-factor. A referee report from hell! After a year in review...

More generally,

slide-20
SLIDE 20

For$any$cograph$G,$at$most$one$of$G$and$$$is$point$determining.


  • Theorem. Every even-order, connected, claw-free graph

has a 1-factor.

Graphs with no Induced K1,3

  • Theorem. If G is connected of even order and every induced

connected subgraph of order 4 has a 1-factor, then so does G.

  • Theorem. For any integer k >1, if G is connected of even order

and every induced connected subgraph of order 2k has a 1-factor, then so does G.

More generally,

K1,3 is the only connected graph on four vertices with no 1-factor.

slide-21
SLIDE 21

Graphs with no Induced K1,3

A graph is locally-connected if the open neighborhood, N(v), of each vertex induces a connected subgraph. Theorem (Oberly, S). If G is a connected, locally-connected, claw-free graph, then G is Hamiltonian. Maybe there is more true than the result states. Maybe a slight strengthening of the condition will imply a stronger result. The Oliver Twist Syndrome.

slide-22
SLIDE 22

Graphs with no Induced K1,3

A non-complete graph G is t-tough if for every separating set S of vertices,

t ≤ S κ G − S

( )

Clearly, every Hamiltonian graph is 1-tough. Conjecture (Chvátal ). There exists t > 1, such that if G is t-tough, then G is Hamiltonian. Conjecture (Chvátal ). There exists t > 2, such that if G is t-tough, then G is Hamiltonian.

slide-23
SLIDE 23

Graphs with no Induced K1,3

Conjecture (Chvátal ). There exists t > 2, such that if G is t-tough, then G is Hamiltonian.

In general, κ ≥ 2τ for any graph.

  • Theorem. (Matthews, S). If G is a claw-free graph, then κ (G) = 2τ(G).

Theorem (Tutte). Every 4-connected, planar graph is Hamiltonian. Theorem (Tutte). If G is planar with τ(G) > 32, then G is Hamiltonian.

Moreover, if the 2-tough conjecture were true, then 4-connected would be enough to guarantee claw-free graphs are Hamiltonian.

For claw-free graphs, equality holds.

slide-24
SLIDE 24

Conjecture (Matthews, S, 1984). Every 4-connected, claw-free graph is Hamiltonian. Conjecture (Fleischner, 1996). These two conjectures are equivalent. Theorem ( , 1997). The two conjectures are equivalent. He proved this result using a new closure concept. Conjecture (Thomassen, 1985). Every 4-connected line graph is Hamiltonian.

Theorem (Zhan). Every 7-connected, line graph is Hamiltonian. Theorem (Zhan). ) Every 7-connected, claw-free graph is Hamiltonian.

slide-25
SLIDE 25

Theorem ( , 1997). The two conjectures are equivalent. Conjecture (Thomassen, 1985). Every 4-connected line graph is Hamiltonian. Conjecture ( , Saburov, Vana, 2011). Every 4-connected, claw-free graph is 1-Hamiltonian-connected. There are quite a few other equivalent versions of the conjecture. 
 Including

Theorem Kaiser, Vrána, Rajacek, 2014

( ). Every 5-connected, claw-free

graph with minimum degree at least 6 is 1-Hamiltonian-connected.

Conjecture (Matthews, S, 1984). Every 4-connected, claw-free graph is Hamiltonian.

slide-26
SLIDE 26

Graph Theory App for the iPad.

!

For more information email me at david@dpsumner.com Check out the details at http://www.dpsumner.com