All those Ramsey classes Ramsey classes with closures and forbidden - - PowerPoint PPT Presentation

all those ramsey classes
SMART_READER_LITE
LIVE PREVIEW

All those Ramsey classes Ramsey classes with closures and forbidden - - PowerPoint PPT Presentation

Ramsey Structural Ramsey Ramsey classes Ramsey classes with constraints All those Ramsey classes Ramsey classes with closures and forbidden homomorphisms Jan Hubi cka Computer Science Institute of Charles University Charles University


slide-1
SLIDE 1

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

All those Ramsey classes

Ramsey classes with closures and forbidden homomorphisms Jan Hubiˇ cka

Computer Science Institute of Charles University Charles University Prague Joint work with Jaroslav Nešetˇ ril

Logic Colloquium 2016, Leeds

slide-2
SLIDE 2

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey Theorem

Theorem (Ramsey Theorem, 1930) ∀n,p,k≥1∃N : N − → (n)p

k.

slide-3
SLIDE 3

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey Theorem

Theorem (Ramsey Theorem, 1930) ∀n,p,k≥1∃N : N − → (n)p

k.

N − → (n)p

k: For every partition of

{1,2,...,N}

p

  • into k classes

(colors) there exists X ⊆ {1, 2, . . . , N}, |X| = n such that X

p

  • belongs to single partition (it is monochromatic)
slide-4
SLIDE 4

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey Theorem

Theorem (Ramsey Theorem, 1930) ∀n,p,k≥1∃N : N − → (n)p

k.

N − → (n)p

k: For every partition of

{1,2,...,N}

p

  • into k classes

(colors) there exists X ⊆ {1, 2, . . . , N}, |X| = n such that X

p

  • belongs to single partition (it is monochromatic)

For p = 2, n = 3, k = 2 put N = 6

slide-5
SLIDE 5

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Many aspects of Ramsey theorem

Ramsey theorem

slide-6
SLIDE 6

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Many aspects of Ramsey theorem

Ramsey theorem Logic

slide-7
SLIDE 7

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Many aspects of Ramsey theorem

Ramsey theorem Logic Combinatorics

slide-8
SLIDE 8

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Many aspects of Ramsey theorem

Ramsey theorem Logic Combinatorics Model Theory

slide-9
SLIDE 9

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Many aspects of Ramsey theorem

Ramsey theorem Logic Combinatorics Model Topological Theory dynamics

slide-10
SLIDE 10

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem for finite relational structures

Let L be a purely relational language with binary relation ≤. Denote by − − → Rel(L) the class of all finite L-structures where ≤ is a linear order. Theorem (Nešetˇ ril-Rödl, 1977; Abramson-Harrington, 1978) ∀A,B∈−

→ Rel(L)∃C∈− → Rel(L) : C −

→ (B)A

2 .

slide-11
SLIDE 11

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem for finite relational structures

Let L be a purely relational language with binary relation ≤. Denote by − − → Rel(L) the class of all finite L-structures where ≤ is a linear order. Theorem (Nešetˇ ril-Rödl, 1977; Abramson-Harrington, 1978) ∀A,B∈−

→ Rel(L)∃C∈− → Rel(L) : C −

→ (B)A

2 .

Theorem (Ramsey Theorem, 1930) ∀n,p,k≥1∃N : N − → (n)p

k.

slide-12
SLIDE 12

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem for finite relational structures

Let L be a purely relational language with binary relation ≤. Denote by − − → Rel(L) the class of all finite L-structures where ≤ is a linear order. Theorem (Nešetˇ ril-Rödl, 1977; Abramson-Harrington, 1978) ∀A,B∈−

→ Rel(L)∃C∈− → Rel(L) : C −

→ (B)A

2 .

B

A

  • is the set of all substructures of B isomorphic to A.

C − → (B)A

2 : For every 2-colouring of

C

A

  • there exists

B ∈ C

B

  • such

that

  • B

A

  • is monochromatic.
slide-13
SLIDE 13

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem for finite relational structures

Let L be a purely relational language with binary relation ≤. Denote by − − → Rel(L) the class of all finite L-structures where ≤ is a linear order. Theorem (Nešetˇ ril-Rödl, 1977; Abramson-Harrington, 1978) ∀A,B∈−

→ Rel(L)∃C∈− → Rel(L) : C −

→ (B)A

2 .

B

A

  • is the set of all substructures of B isomorphic to A.

C − → (B)A

2 : For every 2-colouring of

C

A

  • there exists

B ∈ C

B

  • such

that

  • B

A

  • is monochromatic.

A B C

slide-14
SLIDE 14

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem for finite relational structures

Let L be a purely relational language with binary relation ≤. Denote by − − → Rel(L) the class of all finite L-structures where ≤ is a linear order. Theorem (Nešetˇ ril-Rödl, 1977; Abramson-Harrington, 1978) ∀A,B∈−

→ Rel(L)∃C∈− → Rel(L) : C −

→ (B)A

2 .

B

A

  • is the set of all substructures of B isomorphic to A.

C − → (B)A

2 : For every 2-colouring of

C

A

  • there exists

B ∈ C

B

  • such

that

  • B

A

  • is monochromatic.

A B C

slide-15
SLIDE 15

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Order is necessary

A B

slide-16
SLIDE 16

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Order is necessary

A B

Vertices of C can be linearly ordered and edges coloured accordingly: If edge is goes forward in linear order it is red blue otherwise.

slide-17
SLIDE 17

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Structural extensions

Ramsey theorem for finite relational structures

slide-18
SLIDE 18

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Structural extensions

Ramsey theorem for finite relational structures Structures with functions (Models)

slide-19
SLIDE 19

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Structural extensions

Ramsey theorem for finite relational structures Structures Structures with with functions axioms (Models)

slide-20
SLIDE 20

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Structural extensions

Ramsey theorem for finite relational structures Structures Structures Infinite with with structures functions axioms (Models)

slide-21
SLIDE 21

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Structural extensions

Ramsey theorem for finite relational structures Structures Structures Infinite Categories with with structures functions axioms (Models)

slide-22
SLIDE 22

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem for finite models

Let L be a language with both relations and functions. Assume that L contains binary relation ≤. Denote by − − → Mod(L) the class of all finite L-structures where ≤ is a linear order. Theorem (H.-Nešetˇ ril, 2016) ∀A,B∈−

− → Mod(L)∃C∈− − → Mod(L) : C −

→ (B)A

2 .

slide-23
SLIDE 23

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes

Definition A class C of finite L-structures is Ramsey iff ∀A,B∈C∃C∈C : C − → (B)A

2 .

slide-24
SLIDE 24

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes

Definition A class C of finite L-structures is Ramsey iff ∀A,B∈C∃C∈C : C − → (B)A

2 .

Example (Linear orders — Ramsey Theorem, 1930) The class of all finite linear orders is a Ramsey class.

slide-25
SLIDE 25

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes

Definition A class C of finite L-structures is Ramsey iff ∀A,B∈C∃C∈C : C − → (B)A

2 .

Example (Linear orders — Ramsey Theorem, 1930) The class of all finite linear orders is a Ramsey class. Example (Structures — Nešetˇ ril-Rödl, 76; Abramson-Harrington, 78) For every relational language L, − − → Rel(L) is a Ramsey class.

slide-26
SLIDE 26

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes

Definition A class C of finite L-structures is Ramsey iff ∀A,B∈C∃C∈C : C − → (B)A

2 .

Example (Linear orders — Ramsey Theorem, 1930) The class of all finite linear orders is a Ramsey class. Example (Structures — Nešetˇ ril-Rödl, 76; Abramson-Harrington, 78) For every relational language L, − − → Rel(L) is a Ramsey class. Example (Partial orders — Nešetˇ ril-Rödl, 84; Paoli-Trotter-Walker, 85) The class of all finite partial orders with linear extension is Ramsey.

slide-27
SLIDE 27

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes

Definition A class C of finite L-structures is Ramsey iff ∀A,B∈C∃C∈C : C − → (B)A

2 .

Example (Linear orders — Ramsey Theorem, 1930) The class of all finite linear orders is a Ramsey class. Example (Structures — Nešetˇ ril-Rödl, 76; Abramson-Harrington, 78) For every relational language L, − − → Rel(L) is a Ramsey class. Example (Partial orders — Nešetˇ ril-Rödl, 84; Paoli-Trotter-Walker, 85) The class of all finite partial orders with linear extension is Ramsey. Example (Models — H.-Nešetˇ ril, 2016) For every language L, − − → Mod(L) is a Ramsey class.

slide-28
SLIDE 28

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes are amalgamation classes

Definition (Amalgamation property of class K)

A B B′ C

slide-29
SLIDE 29

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey classes are amalgamation classes

Definition (Amalgamation property of class K)

A B B′ C

Nešetˇ ril, 80’s: Under mild assumptions Ramsey classes have amalgamation property.

A A B C

slide-30
SLIDE 30

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes = ⇒ amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures

slide-31
SLIDE 31

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes = ⇒ amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures Kechris, Pestov, Todorˇ cevi` c: Fraïssé Limits, Ramsey Theory, and topological dynamics of automorphism groups (2005)

slide-32
SLIDE 32

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes = ⇒ amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures Kechris, Pestov, Todorˇ cevi` c: Fraïssé Limits, Ramsey Theory, and topological dynamics of automorphism groups (2005) Definition Let L′ be language containing language L. A lift (or expansion) of L-structure A is L′-structure A′ on the same vertex set such that all relations/functions in L ∩ L′ are identical.

slide-33
SLIDE 33

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes = ⇒ amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures Kechris, Pestov, Todorˇ cevi` c: Fraïssé Limits, Ramsey Theory, and topological dynamics of automorphism groups (2005) Definition Let L′ be language containing language L. A lift (or expansion) of L-structure A is L′-structure A′ on the same vertex set such that all relations/functions in L ∩ L′ are identical. Theorem (Nešetˇ ril, 1989) All homogeneous graphs have Ramsey lift.

slide-34
SLIDE 34

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme amalgamation classes Example

1

The class of finite graphs G is an amalgamation class

slide-35
SLIDE 35

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme amalgamation classes ⇓ homogeneous structures Example

1

The class of finite graphs G is an amalgamation class

2

Fraïssé limit of G is the Rado graph R

slide-36
SLIDE 36

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme amalgamation classes ⇓ lifts of homogeneous ⇐ = homogeneous structures Example

1

The class of finite graphs G is an amalgamation class

2

Fraïssé limit of G is the Rado graph R

3

The lift R′ of R adds generic linear order

slide-37
SLIDE 37

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures Example

1

The class of finite graphs G is an amalgamation class

2

Fraïssé limit of G is the Rado graph R

3

The lift R′ of R adds generic linear order

4

Age(R′) (the class of linearly ordered finite graphs) is Ramsey

slide-38
SLIDE 38

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes = ⇒ amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures Example

1

The class of finite graphs G is an amalgamation class

2

Fraïssé limit of G is the Rado graph R

3

The lift R′ of R adds generic linear order

4

Age(R′) (the class of linearly ordered finite graphs) is Ramsey

slide-39
SLIDE 39

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Nešetˇ ril’s Classification Programme, 2005

Classification Programme Ramsey classes = ⇒ amalgamation classes ⇑ ⇓ lifts of homogeneous ⇐ = homogeneous structures Example

1

The class of finite graphs G is an amalgamation class

2

Fraïssé limit of G is the Rado graph R

3

The lift R′ of R adds generic linear order

4

Age(R′) (the class of linearly ordered finite graphs) is Ramsey Theorem (Jasi´ nski, Laflamme, Nguyen Van Thé, Woodrow, 2014) All homogeneous digraphs have Ramsey lift.

slide-40
SLIDE 40

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Does every amalgamation class have a Ramsey lift?

A question asked by Bodirsky, Nešetˇ ril, Nguyen van Thé, Pinsker, Tsankov around 2010

slide-41
SLIDE 41

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Does every amalgamation class have a Ramsey lift?

A question asked by Bodirsky, Nešetˇ ril, Nguyen van Thé, Pinsker, Tsankov around 2010 Yes: extend language by infinitely many unary relations; assign every vertex to unique relation.

slide-42
SLIDE 42

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Does every amalgamation class have a Ramsey lift?

A question asked by Bodirsky, Nešetˇ ril, Nguyen van Thé, Pinsker, Tsankov around 2010 Yes: extend language by infinitely many unary relations; assign every vertex to unique relation. Definition (Nguyen van Thé) Let K be class of L-structures and K′ be class of lifts of K. K is precompact if for every A ∈ K there are only finitely many lifts of A in K′.

slide-43
SLIDE 43

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Does every amalgamation class have a Ramsey lift?

A question asked by Bodirsky, Nešetˇ ril, Nguyen van Thé, Pinsker, Tsankov around 2010 Yes: extend language by infinitely many unary relations; assign every vertex to unique relation. Definition (Nguyen van Thé) Let K be class of L-structures and K′ be class of lifts of K. K is precompact if for every A ∈ K there are only finitely many lifts of A in K′. K has lift property if for every A ∈ K there exists B ∈ K such that every lift of B in K′ contains every lift of A in K′.

slide-44
SLIDE 44

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Does every amalgamation class have a Ramsey lift?

A question asked by Bodirsky, Nešetˇ ril, Nguyen van Thé, Pinsker, Tsankov around 2010 Yes: extend language by infinitely many unary relations; assign every vertex to unique relation. Definition (Nguyen van Thé) Let K be class of L-structures and K′ be class of lifts of K. K is precompact if for every A ∈ K there are only finitely many lifts of A in K′. K has lift property if for every A ∈ K there exists B ∈ K such that every lift of B in K′ contains every lift of A in K′. Theorem (Kechris, Pestov, Todorˇ cevi` c 2005, Nguyen van Thé 2012) For every amalgamation class K there exists, up to bi-definability, at most one Ramsey class K′ of lifts of K with lift property.

slide-45
SLIDE 45

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Existence of precompact lifts

Question: Does every amalgamation class have a precompact Ramsey lift? No: Consider Z seen as a metric space.

slide-46
SLIDE 46

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Existence of precompact lifts

Question: Does every amalgamation class have a precompact Ramsey lift? No: Consider Z seen as a metric space. Theorem (Evans, 2015) There exists a countable, ω-categorical structure M which that property that its age has no precompact Ramsey lift.

slide-47
SLIDE 47

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Existence of precompact lifts

Question: Does every amalgamation class have a precompact Ramsey lift? No: Consider Z seen as a metric space. Theorem (Evans, 2015) There exists a countable, ω-categorical structure M which that property that its age has no precompact Ramsey lift. Theorem (Evans, Hubiˇ cka, Nešetˇ ril, 2016+) There still exists minimal (non-precompact) Ramsey lift of M with lift property.

slide-48
SLIDE 48

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Map of Ramsey Classes

free restricted

slide-49
SLIDE 49

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Map of Ramsey Classes

free restricted linear orders cyclic orders unions of complete graphs interval graphs permutations

slide-50
SLIDE 50

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Map of Ramsey Classes

free restricted linear orders cyclic orders graphs unions of complete graphs interval graphs permutations Kn-free graphs partial orders acyclic graphs metric spaces

slide-51
SLIDE 51

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Map of Ramsey Classes

free restricted linear orders cyclic orders graphs unions of complete graphs interval graphs permutations Kn-free graphs partial orders acyclic graphs metric spaces

slide-52
SLIDE 52

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

From an amalgamation class to a Ramsey class

The Nešetˇ ril-Rödl partite construction of Ramsey object demands more complicated (multi)amalgamations.

A B C

slide-53
SLIDE 53

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

A, B, C-hypergraphs

A B C

Definition Given structures A, B, C the A, B, C-hypergraph is a hypergraph whose vertices are copies of A in C and hyper-edges are corresponds to copies of B: M is an hyper-edge if M = B for some B ∈ C

B

slide-54
SLIDE 54

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

A, B, C-hypergraphs

A B C

Definition Given structures A, B, C the A, B, C-hypergraph is a hypergraph whose vertices are copies of A in C and hyper-edges are corresponds to copies of B: M is an hyper-edge if M = B for some B ∈ C

B

  • If the A, B, C-hypegraph has chromatic number 3, then C −

→ (B)A

2

slide-55
SLIDE 55

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

A, B, C-hypergraphs

A B C

Definition Given structures A, B, C the A, B, C-hypergraph is a hypergraph whose vertices are copies of A in C and hyper-edges are corresponds to copies of B: M is an hyper-edge if M = B for some B ∈ C

B

  • If the A, B, C-hypegraph has chromatic number 3, then C −

→ (B)A

2

Hypergraphs of large chromatic number are known to exists, but typically they do not correspond to A, B, C-hypergraph

slide-56
SLIDE 56

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

Structure is irreducible if every pair of vertices is contained in some tuple of some relation We consider classes of irreducible structures and split amalgamation into two steps:

slide-57
SLIDE 57

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

Structure is irreducible if every pair of vertices is contained in some tuple of some relation We consider classes of irreducible structures and split amalgamation into two steps:

1

free amalgamation,

slide-58
SLIDE 58

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

Structure is irreducible if every pair of vertices is contained in some tuple of some relation We consider classes of irreducible structures and split amalgamation into two steps:

1

free amalgamation,

2

completion.

slide-59
SLIDE 59

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

Structure is irreducible if every pair of vertices is contained in some tuple of some relation We consider classes of irreducible structures and split amalgamation into two steps:

1

free amalgamation,

2

completion. Definition Irreducible structure C′ is a strong completion of C if it has the same vertex set and every irreducible substructure of C is also (induced) substructure of C′.

slide-60
SLIDE 60

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

Theorem (H.-Nešetˇ ril, 2016) Let R be a Ramsey class of irreducible finite structures and let K be a hereditary locally finite subclass of R with strong amalgamation. Then K is Ramsey.

slide-61
SLIDE 61

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

Theorem (H.-Nešetˇ ril, 2016) Let R be a Ramsey class of irreducible finite structures and let K be a hereditary locally finite subclass of R with strong amalgamation. Then K is Ramsey. Schematically Ramsey = ⇒ amalgamation amalgamation + order + local finiteness = ⇒ Ramsey What is local finiteness?

slide-62
SLIDE 62

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

K is locally finite subclass of R if for every C0 in R there exists a finite bound on minimal set of obstacles which prevents a structure with homomorphism to C0 from being completed to K.

slide-63
SLIDE 63

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Our approach

K is locally finite subclass of R if for every C0 in R there exists a finite bound on minimal set of obstacles which prevents a structure with homomorphism to C0 from being completed to K. Definition Let R be a class of finite irreducible structures and K a subclass of R. We say that the class K is locally finite subclass of R if for every C0 ∈ R there is n = n(C0) such that every structure C has strong K-completion providing that it satisfies the following:

1

there is a homomorphism-embedding from C to C0

2

every substructure of C with at most n vertices has a strong K-completion.

slide-64
SLIDE 64

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Locally finite subclass, an example

Example Consider class of metric spaces with distances {1, 2, 3, 4}. Graph with edges labelled by {1, 2, 3, 4} can be completed to a metric space if and only if it does not contain one of:

1 1 3 1 1 4 1 2 4 4 1 1 1

slide-65
SLIDE 65

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

S-metric-spaces

Theorem (H.-Nešetˇ ril, 2016) Let S be set of positive reals. The class MS of all metric spaces with distances restricted to S has precompact Ramsey lift iff MS is an amalgamation class. Consider only finite S Identify minimal obstacles for completion of S-graphs. Show that they are all cycles of bounded size Apply previous theorem.

slide-66
SLIDE 66

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

S-metric-spaces

Theorem (H.-Nešetˇ ril, 2016) Let S be set of positive reals with no jump numbers. The class MS of all metric spaces with distances restricted to S has precompact Ramsey lift iff MS is an amalgamation class. Consider only finite S Identify minimal obstacles for completion of S-graphs. Show that they are all cycles of bounded size Apply previous theorem.

slide-67
SLIDE 67

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

S-metric-spaces

Theorem (H.-Nešetˇ ril, 2016) Let S be set of positive reals with no jump numbers. The class MS of all metric spaces with distances restricted to S has precompact Ramsey lift iff MS is an amalgamation class. Consider only finite S Identify minimal obstacles for completion of S-graphs. Show that they are all cycles of bounded size Apply previous theorem. Consider S = {1, 3} and cycles:

3 1 1 1 1 all 1s

slide-68
SLIDE 68

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

S-metric-spaces

Definition (Delhommé, Laflamme, Pouzet, Sauer, 2007) a ∈ S is jump number if a < max(S) and 2a ≤ minb∈S (b > a).

slide-69
SLIDE 69

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

S-metric-spaces

Definition (Delhommé, Laflamme, Pouzet, Sauer, 2007) a ∈ S is jump number if a < max(S) and 2a ≤ minb∈S (b > a). Let A be an S-metric space, j jump number. Then the following is an equivalence: u ∼j v whenever d(u, v) ≤ j

slide-70
SLIDE 70

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

S-metric-spaces

Definition (Delhommé, Laflamme, Pouzet, Sauer, 2007) a ∈ S is jump number if a < max(S) and 2a ≤ minb∈S (b > a). Let A be an S-metric space, j jump number. Then the following is an equivalence: u ∼j v whenever d(u, v) ≤ j Lemma If K is a subclass of R and define more equivalences then R, then K is not locally finite subclass.

3 1 1 1 1 all 1s = = = = =

slide-71
SLIDE 71

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Extended approach

Represent equivalences with infinitely many classes by additional vertices and functions

slide-72
SLIDE 72

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Extended approach

Represent equivalences with infinitely many classes by additional vertices and functions

slide-73
SLIDE 73

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Extended approach

Represent equivalences with infinitely many classes by additional vertices and functions Show local finiteness in this extended language. Forbidden configurations are now finite: = = = = = = =

slide-74
SLIDE 74

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Extended approach

Represent equivalences with infinitely many classes by additional vertices and functions Show local finiteness in this extended language. Forbidden configurations are now finite: = = = = = = = We need Ramsey theorem for amalgamation classes of ordered models!

slide-75
SLIDE 75

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem with closures

Definition

Let L be a language, R be a Ramsey class of finite irreducible L-structures and U be a closure description. We say that a subclass K of R is an (R, U)-multiamalgamation class iff:

1

K is hereditary subclass of R consiting of U-closed structures.

2

K is closed for strong amalgamation over U-closed substructures

slide-76
SLIDE 76

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem with closures

Definition

Let L be a language, R be a Ramsey class of finite irreducible L-structures and U be a closure description. We say that a subclass K of R is an (R, U)-multiamalgamation class iff:

1

K is hereditary subclass of R consiting of U-closed structures.

2

K is closed for strong amalgamation over U-closed substructures

3

Let B ∈ K and C0 ∈ R. Then there exists n = n(B, C0) such that if U-closed L-structure C satisfies the following:

1

C0 is a completion of C), and,

2

every substructure of C, |C| ≤ n has a K-completion. Then there exists C′ ∈ K that is a completion of C with respect to copies of B.

slide-77
SLIDE 77

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem with closures

Theorem (H.-Nešetˇ ril, 2016) Every (R, U)-multiamalgamation class K is Ramsey. Again it is the only non-trivial step is to verify local finiteness!

slide-78
SLIDE 78

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey theorem with closures

Theorem (H.-Nešetˇ ril, 2016) Every (R, U)-multiamalgamation class K is Ramsey. Again it is the only non-trivial step is to verify local finiteness! The theorem is proved by a variation and strengthening of the Partite Constructions (Nešetˇ ril-Rödl).

slide-79
SLIDE 79

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Hales-Jewett Theorem Ramsey Theorem Partite Lemma Neˇ setˇ ril-R¨

  • dl Theorem

Partite Con- struction Ramsey Property of

  • rdered structures

(Theorem 3.6) Partite Con- struction for U-substructures (Lemma 2.6) U-closed Partite Construction (Lemma 2.5) Partite Lemma with closures (Lemma 2.4) Iterated Partite Construction (Lemmas 2.7 and 2.8) Ramsey property

  • f locally finite

strong amalga- mation classes (Theorem 2.1) Ramsey property

  • f multiamalga-

mation classes (Theorem 2.2) Explicit description

  • f lift of Forbhe(F)

(Theorem 3.3) Ramsey property

  • f lifts of classes

defined by forbidden homomorphism- embeddings (Theorem 3.7) C0 − → (B)A

2

U-closed C1 − → (B)A

2

Iterate n(C0) times

slide-80
SLIDE 80

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden homomorphisms

Theorem (Nešetˇ ril-Rödl Theorem for relational structures, 1977) Let L be a relational language containing binary relation ≤ and E be a (possibly infinite) family of ordered irreducible L-structures.

slide-81
SLIDE 81

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden homomorphisms

Theorem (Nešetˇ ril-Rödl Theorem for relational structures, 1977) Let L be a relational language containing binary relation ≤ and E be a (possibly infinite) family of ordered irreducible L-structures. Then the class of all ordered structures in Forbe(E) is Ramsey.

slide-82
SLIDE 82

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden homomorphisms

Theorem (Nešetˇ ril-Rödl Theorem for relational structures, 1977) Let L be a relational language containing binary relation ≤ and E be a (possibly infinite) family of ordered irreducible L-structures. Then the class of all ordered structures in Forbe(E) is Ramsey. Forbe(E) = class of all finite structures with no embedding of F ∈ E.

slide-83
SLIDE 83

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden homomorphisms

Theorem (Nešetˇ ril-Rödl Theorem for relational structures, 1977) Let L be a relational language containing binary relation ≤ and E be a (possibly infinite) family of ordered irreducible L-structures. Then the class of all ordered structures in Forbe(E) is Ramsey. Forbe(E) = class of all finite structures with no embedding of F ∈ E. Forbhe(E) = class of all finite structures with no homomorphism-embedding image of some F ∈ E. Theorem (H.-Nešetˇ ril, 2016) Let L be a relational language containing binary relation ≤, and F be a regular family of finite connected weakly ordered L-structures.

slide-84
SLIDE 84

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden homomorphisms

Theorem (Nešetˇ ril-Rödl Theorem for relational structures, 1977) Let L be a relational language containing binary relation ≤ and E be a (possibly infinite) family of ordered irreducible L-structures. Then the class of all ordered structures in Forbe(E) is Ramsey. Forbe(E) = class of all finite structures with no embedding of F ∈ E. Forbhe(E) = class of all finite structures with no homomorphism-embedding image of some F ∈ E. Theorem (H.-Nešetˇ ril, 2016) Let L be a relational language containing binary relation ≤, and F be a regular family of finite connected weakly ordered L-structures. Assume that the class of all ordered structures in Forbhe(F) is a locally finite subclass of − − → Rel(L).

slide-85
SLIDE 85

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden homomorphisms

Theorem (Nešetˇ ril-Rödl Theorem for relational structures, 1977) Let L be a relational language containing binary relation ≤ and E be a (possibly infinite) family of ordered irreducible L-structures. Then the class of all ordered structures in Forbe(E) is Ramsey. Forbe(E) = class of all finite structures with no embedding of F ∈ E. Forbhe(E) = class of all finite structures with no homomorphism-embedding image of some F ∈ E. Theorem (H.-Nešetˇ ril, 2016) Let L be a relational language containing binary relation ≤, and F be a regular family of finite connected weakly ordered L-structures. Assume that the class of all ordered structures in Forbhe(F) is a locally finite subclass of − − → Rel(L). Then the class K of all ordered structures in Forbhe(F) has a precompact Ramsey lift.

slide-86
SLIDE 86

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Example

Theorem (H., Nešetˇ ril, 2014) The class of graphs not containing bow-tie as non-induced subgraph have Ramsey lift. Bow-tie graph:

slide-87
SLIDE 87

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Example

Theorem (H., Nešetˇ ril, 2014) The class of graphs not containing bow-tie as non-induced subgraph have Ramsey lift. Bow-tie graph: Amalgamation of two triangles must unify vertices.

Wrong!

slide-88
SLIDE 88

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Example

Structure of bow-tie-free graphs Edges in no triangles Edges in 1 triangle Edges in > 1 triangles

slide-89
SLIDE 89

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Example

Structure of bow-tie-free graphs Edges in no triangles Edges in 1 triangle Edges in > 1 triangles Lemma Bow-tie-free graphs have free amalgamation over closed structures.

slide-90
SLIDE 90

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Example

Closures in the class of bowtie-free graphs:

slide-91
SLIDE 91

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden monomorphisms

Theorem (H.-Nešetˇ ril, 2016) Let M be a set of finite connected structures such that Forbm(M) has an ω-categorical universal structure U.

slide-92
SLIDE 92

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden monomorphisms

Theorem (H.-Nešetˇ ril, 2016) Let M be a set of finite connected structures such that Forbm(M) has an ω-categorical universal structure U. Further assume that for every M ∈ M at least one of the conditions holds:

1

there is no homomorphism-embedding of M to U, or,

2

M can be constructed from irreducible structures by a series of free amalgamations over irreducible substructures.

slide-93
SLIDE 93

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden monomorphisms

Theorem (H.-Nešetˇ ril, 2016) Let M be a set of finite connected structures such that Forbm(M) has an ω-categorical universal structure U. Further assume that for every M ∈ M at least one of the conditions holds:

1

there is no homomorphism-embedding of M to U, or,

2

M can be constructed from irreducible structures by a series of free amalgamations over irreducible substructures. Then the class of finite algebraically closed substructures of U has precompact Ramsey lift.

slide-94
SLIDE 94

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden monomorphisms

Theorem (H.-Nešetˇ ril, 2016) Let M be a set of finite connected structures such that Forbm(M) has an ω-categorical universal structure U. Further assume that for every M ∈ M at least one of the conditions holds:

1

there is no homomorphism-embedding of M to U, or,

2

M can be constructed from irreducible structures by a series of free amalgamations over irreducible substructures. Then the class of finite algebraically closed substructures of U has precompact Ramsey lift.

slide-95
SLIDE 95

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Ramsey lifts for forbidden monomorphisms

Theorem (H.-Nešetˇ ril, 2016) Let M be a set of finite connected structures such that Forbm(M) has an ω-categorical universal structure U. Further assume that for every M ∈ M at least one of the conditions holds:

1

there is no homomorphism-embedding of M to U, or,

2

M can be constructed from irreducible structures by a series of free amalgamations over irreducible substructures. Then the class of finite algebraically closed substructures of U has precompact Ramsey lift.

slide-96
SLIDE 96

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Further applications

Many classical Ramsey classes follows as corollaries of

  • ur results:

Partial orders Metric spaces H-colourable graphs Graphs with metric embeddings Steiner systems graphs with no short odd cycles . . .

S-metric spaces and generalizations Classes defining multiple orders Totally ordered structures “Fat” structures . . .

slide-97
SLIDE 97

Ramsey → Structural Ramsey Ramsey classes Ramsey classes with constraints

Thank you for the attention

J.H., J. Nešetˇ ril: Bowtie-free graphs have a Ramsey lift. Submitted (arXiv:1402.2700), 2014, 22 pages. J.H., J. Nešetˇ ril: All those Ramsey classes (Ramsey classes with closures and forbidden homomorphisms). Submitted (arXiv:1606.07979), 2016, 59 pages.