Searching for template structures for the class of ...omer/BPGMT18Presentation.pdf · Intuition...
Transcript of Searching for template structures for the class of ...omer/BPGMT18Presentation.pdf · Intuition...
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Searching for template structures for theclass of Hrushovski ab initio geometries
Omer MermelsteinBen-Gurion University of the Negev
BPGMTC 2018
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Intuition
δ(A) = |A| − |E |
d(A,B) = inf {δ(B ′) | A ⊆ B ′ ⊆fin B}
A 6 B ⇐⇒ δ(A) = d(A,B)
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Intuition
δ(A) = |A| − |E |
d(A,B) = inf {δ(B ′) | A ⊆ B ′ ⊆fin B}
A 6 B ⇐⇒ δ(A) = d(A,B)
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Intuition
δ(A) = |A| − |E |
d(A,B) = inf {δ(B ′) | A ⊆ B ′ ⊆fin B}
A 6 B ⇐⇒ δ(A) = d(A,B)
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Intuition
δ(A) = |A| − |E |
d(A,B) = inf {δ(B ′) | A ⊆ B ′ ⊆fin B}
A 6 B ⇐⇒ δ(A) = d(A,B)
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Pregeometry - definition
Examples: cardinality, linear dimension, transcendence degree.
DefinitionA combinatorial pregeometry is a (assume countable) set X with adimension function d : P(X )→ N ∪ {∞} such that for allY1,Y2 ∈ P(X ):
1. d(Y1) ≤ |Y1|2. d(Y1) ≤ d(Y1 ∪ Y2) ≤ d(Y1) + d(Y2)
3. d(Y1 ∪ Y2) ≤ d(Y1) + d(Y2)− d(Y1 ∩ Y2) (submodular)
*4. d(Y1) = sup {d(Y0) | Y0 ∈ Fin(Y1)} (finitary)
More definitionsA finite set Y is independent if d(Y ) = |Y |.(X , d) is n-pure if all Y ∈ [X ]≤n are independent.A pregeometry is called a geometry if it is 2-pure.
Each pregeometry has a canonically associated geometry.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Pregeometry - definition
Examples: cardinality, linear dimension, transcendence degree.
DefinitionA combinatorial pregeometry is a (assume countable) set X with adimension function d : P(X )→ N ∪ {∞} such that for allY1,Y2 ∈ P(X ):
1. d(Y1) ≤ |Y1|2. d(Y1) ≤ d(Y1 ∪ Y2) ≤ d(Y1) + d(Y2)
3. d(Y1 ∪ Y2) ≤ d(Y1) + d(Y2)− d(Y1 ∩ Y2) (submodular)
*4. d(Y1) = sup {d(Y0) | Y0 ∈ Fin(Y1)} (finitary)
More definitionsA finite set Y is independent if d(Y ) = |Y |.(X , d) is n-pure if all Y ∈ [X ]≤n are independent.A pregeometry is called a geometry if it is 2-pure.
Each pregeometry has a canonically associated geometry.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Fraısse’s TheoremLet C be a countable (up to isomorphisms) class of finitelygenerated structures, closed under isomorphisms. Let 6 be adistinguished notion of embedding, preserved under isomorphism.Assume
HP A 6 B, B ∈ C =⇒ A ∈ C.
JEP A,B ∈ C =⇒ ∃D ∈ C s.t. A,B 6 D.
AP ∀A,B1,B2 ∈ C
B1
A D
B2
66
6 6
Then there exists a unique (up to isomorphism) countable genericstructure M with age6(M) = C such that ∀A,B ∈ C
B
A M
66
6
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
DefinitionFor any finite hypergraph A = (V ,E ) define
δ(A) = |V | − |E |
and for any induced subgraph B ⊆ A define
d(B,A) = min {δ(B ′) | B ⊆ B ′ ⊆fin A}B 6 A ⇐⇒ δ(B) = d(B,A)
The function d(·,A) defines a pregeometry whenever ∅ 6 A.
DefinitionSay that B strongly embeds into A if there exists an embeddingf : B → A such that f [B] 6 A.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Hrushovski’s constructionLet C = {A | ∅ 6 A}. The class C is a Fraısse amalgamation classwith respect to 6-embeddings. We denote its generic structure Mand its associated geometry G.
Variations
I Restrict C to n-uniform hypergraphs — Mn
Fact: n 6= k =⇒ Gn 6∼= Gk (Evans and Ferreira)
I Consider directed hypergraphs — M6∼Fact: G 6∼ ∼= G
I Restrict to hypergraphs whose geometries are n-pure — Mn
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Hrushovski’s constructionLet C = {A | ∅ 6 A}. The class C is a Fraısse amalgamation classwith respect to 6-embeddings. We denote its generic structure Mand its associated geometry G.
Variations
I Restrict C to n-uniform hypergraphs — Mn
Fact: n 6= k =⇒ Gn 6∼= Gk (Evans and Ferreira)
I Consider directed hypergraphs — M6∼Fact: G 6∼ ∼= G
I Restrict to hypergraphs whose geometries are n-pure — Mn
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Hrushovski’s constructionLet C = {A | ∅ 6 A}. The class C is a Fraısse amalgamation classwith respect to 6-embeddings. We denote its generic structure Mand its associated geometry G.
Variations
I Restrict C to n-uniform hypergraphs — Mn
Fact: n 6= k =⇒ Gn 6∼= Gk (Evans and Ferreira)
I Consider directed hypergraphs — M6∼Fact: G 6∼ ∼= G
I Restrict to hypergraphs whose geometries are n-pure — Mn
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
AxiomWe want to classify the geometries of strongly minimal structures(regular types).
(Buzzwords: Zilber’s conjecture, Hrushovski fusion)
FactEvery reduct of a strongly minimal structure is strongly minimal. 1
Questions
I What are the non-trivial reducts of M?
I Is any reduct of an ab initio construction again an ab initioconstruction?
1Disregard the fact that M is not strongly minimal
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
AxiomWe want to classify the geometries of strongly minimal structures(regular types).
(Buzzwords: Zilber’s conjecture, Hrushovski fusion)
FactEvery reduct of a strongly minimal structure is strongly minimal. 1
Questions
I What are the non-trivial reducts of M?
I Is any reduct of an ab initio construction again an ab initioconstruction?
1Disregard the fact that M is not strongly minimal
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
AxiomWe want to classify the geometries of strongly minimal structures(regular types).
(Buzzwords: Zilber’s conjecture, Hrushovski fusion)
FactEvery reduct of a strongly minimal structure is strongly minimal. 1
Questions
I What are the non-trivial reducts of M?
I Is any reduct of an ab initio construction again an ab initioconstruction?
1Disregard the fact that M is not strongly minimal
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
AxiomWe want to classify the geometries of strongly minimal structures(regular types).
(Buzzwords: Zilber’s conjecture, Hrushovski fusion)
FactEvery reduct of a strongly minimal structure is strongly minimal. 1
Questions
I What are the non-trivial reducts of M?
I Is any reduct of an ab initio construction again an ab initioconstruction?
1Disregard the fact that M is not strongly minimal
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Example
Consider the reduct of M to the formula:
ϕ(x1, . . . , xn) := ∃!y1, . . . yn−1
n∧i=1
E (y1, . . . , yn−1, xi )
Where x1, . . . , xn, y1, . . . , yn−1 are distinct elements.
I The reduct 〈M, ϕ(M)〉 is non-trivial.
I The number of ϕ-edges on a set is not linearly bounded – astandard δ-function cannot capture the dimension.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Example
Consider the reduct of M to the formula:
ϕ(x1, . . . , xn) := ∃!y1, . . . yn−1
n∧i=1
E (y1, . . . , yn−1, xi )
Where x1, . . . , xn, y1, . . . , yn−1 are distinct elements.
I The reduct 〈M, ϕ(M)〉 is non-trivial.
I The number of ϕ-edges on a set is not linearly bounded – astandard δ-function cannot capture the dimension.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
SolutionGeneralize the predimension function: make a ϕ-clique havepredimension n − 1.
From here on, we restrict the discussion to n-uniform hypergraphs,for some fixed big n.
The λ function
I K ⊆ A is a clique if [K ]n ⊆ E .
I Define M(A) to be the set of maximal cliques in A.
I s(A) =∑
K∈M(A)(|K | − (n − 1))
I λ(A) = |A| − s(A)
Redefining 6 using λ, the class C expands, but remains a Fraısseamalgamation class (modulo some technicalities swept under therug).
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
SolutionGeneralize the predimension function: make a ϕ-clique havepredimension n − 1.
From here on, we restrict the discussion to n-uniform hypergraphs,for some fixed big n.
The λ function
I K ⊆ A is a clique if [K ]n ⊆ E .
I Define M(A) to be the set of maximal cliques in A.
I s(A) =∑
K∈M(A)(|K | − (n − 1))
I λ(A) = |A| − s(A)
Redefining 6 using λ, the class C expands, but remains a Fraısseamalgamation class (modulo some technicalities swept under therug).
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
SolutionGeneralize the predimension function: make a ϕ-clique havepredimension n − 1.
From here on, we restrict the discussion to n-uniform hypergraphs,for some fixed big n.
The λ function
I K ⊆ A is a clique if [K ]n ⊆ E .
I Define M(A) to be the set of maximal cliques in A.
I s(A) =∑
K∈M(A)(|K | − (n − 1))
I λ(A) = |A| − s(A)
Redefining 6 using λ, the class C expands, but remains a Fraısseamalgamation class (modulo some technicalities swept under therug).
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
DefinitionConsider a pregeometry (X , d) as a hypergraph by defining
E = {Y ⊆fin X | d(Y ) < Y }
Now (X , d) can be seen as a first order structure in the languageL = {En}n∈N.
The λ function estimates the dimension of an n-uniformhypergraph associated to a pregeometry.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
DefinitionDefine Cgeo = age(Gn
n � En).(Mn
n is Hrushovski’s construction for n-uniform hypergraphs whosegeometry is n-pure).
TheoremCgeo is a Fraısse amalgamation class with respect to6-embeddings with a generic structure Mgeo .Moreover, Mgeo ∼= Gn
n � En, and under this identificationGgeo = Gn
n.
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Questions?
Thank you!
Searching for ab initioTemplates
Omer Mermelstein
Preliminaries
Intuition
Combinatorial geometries
Fraısse amalgamation
The ab initioconstruction
Generalizing thepredimension
Motivation
Clique predimension
The geometricconstruction
Questions?
Thank you!