On locally o-minimal structures Hiroshi Tanaka joint works with - - PowerPoint PPT Presentation

on locally o minimal structures
SMART_READER_LITE
LIVE PREVIEW

On locally o-minimal structures Hiroshi Tanaka joint works with - - PowerPoint PPT Presentation

On locally o-minimal structures Hiroshi Tanaka joint works with Tomohiro Kawakami, Kota Takeuchi and Akito Tsuboi Anan National College of Technology RIMS Model Theory Meeting December 1, 2010 RIMS Hiroshi Tanaka (Anan National College of


slide-1
SLIDE 1

On locally o-minimal structures

Hiroshi Tanaka joint works with Tomohiro Kawakami, Kota Takeuchi and Akito Tsuboi

Anan National College of Technology

RIMS Model Theory Meeting December 1, 2010 RIMS

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 1 / 28

slide-2
SLIDE 2

Outline

1

local o-minimality

2

uniform local o-minimalily and strong local o-minimality

3

local monotonicity

4

local cell decomposition property

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 2 / 28

slide-3
SLIDE 3

Intervals and convex sets

Let L be a language containing <. Let M = (M, <, . . .) be an L-structure expanding a dense linear

  • rdering <.

Definition 1.1 A ⊆ M is said to be convex in M if for any a, b ∈ A, we have (a, b) ⊆ A. If additionally sup A, inf A ∈ M ∪ {−∞, ∞}, then A is called an interval in M.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 3 / 28

slide-4
SLIDE 4

Intervals and convex sets Example 1.2

Let Q1 = (Q, <).

(−1, 1), [−1, 1], [−1, 1), (−1, 1], {1} are intervals. ( − √ 2, √ 2 ) ∩ Q is not an interval but a convex set. Example 1.3

Let Q2 = (Q × Q, <), where < is the lexicographic ordering.

{0} × Q is not an interval but a convex set.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 4 / 28

slide-5
SLIDE 5

O-minimal (weakly o-minimal) structures

Let M = (M, <, . . .) be an L-structure expanding a dense linear

  • rdering <.

Definition 1.4 M is said to be o-minimal if any definable subset of M is a finite union of intervals. M is said to be weakly o-minimal if any definable subset of M is a finite union of convex sets.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 5 / 28

slide-6
SLIDE 6

Locally o-minimal structures

The notion of local o-minmality and strongly local o-minimality was introduced by C. Toffalori and K. Vozoris.

Definition 1.5 M is said to be locally o-minimal if for any a ∈ M and any definable X ⊆ M, there is an open interval I ∋ a such that X ∩ I is a finite union

  • f intervals.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 6 / 28

slide-7
SLIDE 7

Locally o-minimal structures

The notion of local o-minmality and strongly local o-minimality was introduced by C. Toffalori and K. Vozoris.

Definition 1.5 M is said to be locally o-minimal if for any a ∈ M and any definable X ⊆ M, there is an open interval I ∋ a such that X ∩ I is a finite union

  • f intervals.

M is said to be strongly locally o-minimal if for any a ∈ M, there is an

  • pen interval I ∋ a such that for any definable X ⊆ M, X ∩ I is a finite

union of intervals.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 6 / 28

slide-8
SLIDE 8

Locally o-minimal structures

The notion of local o-minmality and strongly local o-minimality was introduced by C. Toffalori and K. Vozoris.

Definition 1.5 M is said to be locally o-minimal if for any a ∈ M and any definable X ⊆ M, there is an open interval I ∋ a such that X ∩ I is a finite union

  • f intervals.

M is said to be strongly locally o-minimal if for any a ∈ M, there is an

  • pen interval I ∋ a such that for any definable X ⊆ M, X ∩ I is a finite

union of intervals. M is said to be uniformly locally o-minimal if for any a ∈ M and any formula ϕ(x, y) ∈ L, there is an open interval I ∋ a such that ϕ(M, b) ∩ I is a finite union of intervals for any b ∈ M.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 6 / 28

slide-9
SLIDE 9

Examples

A typical example of locally o-minimal structures is the following structure.

Example 1.6 (Marker and Steinhorn) R = (R, <, +, sin(x)) is strongly locally o-minimal.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 7 / 28

slide-10
SLIDE 10

Facts

In locally o-minimal structures, the following are known.

Proposition 1.7 (Toffalori and Vozoris)

Any weakly o-minimal structure is locally o-minimal.

Proposition 1.8 (Toffalori and Vozoris)

Local o-minimality is preserved under elementary equivalence.

Remark 1.9 (Toffalori and Vozoris)

Strong local o-minimality is not preserved under elementary equivalence.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 8 / 28

slide-11
SLIDE 11

Uniformly locally o-minimal structures Proposition 2.1 (Kawakami, Takeuchi, Tsuboi, and T.)

Let M be a uniformly locally o-minimal structure. Suppose that M is

ω-saturated. Then, M is strongly locally o-minimal.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 9 / 28

slide-12
SLIDE 12

Uniformly locally o-minimal structures Proposition 2.1 (Kawakami, Takeuchi, Tsuboi, and T.)

Let M be a uniformly locally o-minimal structure. Suppose that M is

ω-saturated. Then, M is strongly locally o-minimal. Proof.

Let a ∈ M and ϕ(x, y) ∈ L. By the uniformity of M, there is an open interval I ∋ a such that for any b ∈ M, we can take nb ∈ N so that ϕ(M, b) ∩ I is a union

  • f nb many intervals.

By the saturation of M, the set {nb : b ∈ M} is uniformly bounded, denoted by nϕ ∈ N.

  • Hiroshi Tanaka (Anan National College of Technology)

Locally o-minimal structures December 1, 2010 9 / 28

slide-13
SLIDE 13

Uniformly locally o-minimal structures Proof.

Let θϕ(u, v) ≡ for any z ∈ M, the set {x ∈ (u, v) : M |

= ϕ(x, z)} is a

union of at most nϕ many intervals. Let Γ(u, v) ≡ {u < a < v} ∪ {θϕ(u, v) : ϕ ∈ L}.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 10 / 28

slide-14
SLIDE 14

Uniformly locally o-minimal structures Proof.

Let θϕ(u, v) ≡ for any z ∈ M, the set {x ∈ (u, v) : M |

= ϕ(x, z)} is a

union of at most nϕ many intervals. Let Γ(u, v) ≡ {u < a < v} ∪ {θϕ(u, v) : ϕ ∈ L}. By compactness, Γ(u, v) is consistent. By the saturation of M, there are c, d ∈ M such that M |

= Γ(c, d).

The open interval (c, d) witnesses to the strong local

  • -minimality of M.
  • Hiroshi Tanaka (Anan National College of Technology)

Locally o-minimal structures December 1, 2010 10 / 28

slide-15
SLIDE 15

Uniformly locally o-minimal structures

We show that there is an ω-saturated locally o-minimal structure that is not uniformly locally o-minimal.

Example 2.2

Let L = {<, Pq}q∈Q+ and M := (Q, <, Pq)q∈Q+. Here, Pq(a, b) ⇐

⇒ a + √ 2 · q ≤ b in R.

Let M∗ ≻ M be ω-saturated. Then, M∗ is locally o-minimal but not uniformly locally o-minimal. For example, when a = 1, q = 2,

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 11 / 28

slide-16
SLIDE 16

Uniformly locally o-minimal structures Example 2.3

Let L = {<, Pq}q∈Q+ and M := (Q, <, Pq)q∈Q+. Here, Pq(a, b) ⇐

⇒ a + √ 2 · q ≤ b in R.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 12 / 28

slide-17
SLIDE 17

Uniformly locally o-minimal structures Example 2.3

Let L = {<, Pq}q∈Q+ and M := (Q, <, Pq)q∈Q+. Here, Pq(a, b) ⇐

⇒ a + √ 2 · q ≤ b in R. Proof. Th(M) admits elimination of quantifiers.

So, M is weakly o-minimal and hence locally o-minimal.

  • Hiroshi Tanaka (Anan National College of Technology)

Locally o-minimal structures December 1, 2010 12 / 28

slide-18
SLIDE 18

Uniformly locally o-minimal structures Proof.

However, M is not uniformly locally o-minimal.

∵) For any open interval I = (b, c) ∋ 0,

we have P1(b, M) ∧ P1(c, M) ∅.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 13 / 28

slide-19
SLIDE 19

Uniformly locally o-minimal structures Proof.

However, M is not uniformly locally o-minimal.

∵) For any open interval I = (b, c) ∋ 0,

we have P1(b, M) ∧ P1(c, M) ∅. We take u ∈ M such that P1(b, u) ∧ P1(c, u). Then, the set P1(M, u) divedes into two convexes C1 and C2. Neither C1 nor C2 are intervals.

  • Hiroshi Tanaka (Anan National College of Technology)

Locally o-minimal structures December 1, 2010 13 / 28

slide-20
SLIDE 20

Uniformly locally o-minimal structures

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 14 / 28

slide-21
SLIDE 21

Local monotonicity Definition 3.1 A local o-minimal strucuture M = (M, <, . . .) have local monotonicity if for any definable X ⊆ M, any definable f : X → M and any a ∈ M there are an open interval I ∋ a and intervals X0, X1, . . ., Xn such that any f|Xi is constant, strictly increasing, or strictly decreasing. If additionally any f|Xi is continuous, we have local monotonicity with continuity.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 15 / 28

slide-22
SLIDE 22

Local monotonicity

In general, locally o-minimal structures do not have local

  • monotonicity. However, the following holds.

Fact 1 (Toffalori and Vozoris)

Any strongly locally o-minimal structure satisfies local monotonicity.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 16 / 28

slide-23
SLIDE 23

Local monotonicity

In general, locally o-minimal structures do not have local

  • monotonicity. However, the following holds.

Fact 1 (Toffalori and Vozoris)

Any strongly locally o-minimal structure satisfies local monotonicity.

Fact 2 (Toffalori and Vozoris)

Any locally o-minimal expansion (R, <, . . .) of (R, <) is strongly locally

  • -minimal.

Fact 3 (Kawakami and T.)

Any locally o-minimal expansion (R, <, . . .) of (R, <) satisfies local monotonicity with continuity.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 16 / 28

slide-24
SLIDE 24

Local monotonicity

In general, a strongly locally o-minimal structure dose not satisfy local monotonicity with continuity.

Example 3.2 M = (Q × Q, <, 0, f(x), E(x, y)).

Here, < is the lexicograhic ordering, 0 := (0, 0), and for any (p, q), (p1, q1), (p2, q2) ∈ Q × Q,

f((p, q)) := (q, 0) and E((p1, q1), (p2, q2)) ⇐ ⇒ p1 = p2.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 17 / 28

slide-25
SLIDE 25

Local monotonicity

In general, a strongly locally o-minimal structure dose not satisfy local monotonicity with continuity.

Example 3.2 M = (Q × Q, <, 0, f(x), E(x, y)).

Here, < is the lexicograhic ordering, 0 := (0, 0), and for any (p, q), (p1, q1), (p2, q2) ∈ Q × Q,

f((p, q)) := (q, 0) and E((p1, q1), (p2, q2)) ⇐ ⇒ p1 = p2.

Then, for x = (x1, x2), y = (y1, y2) ∈ M,

x < y ⇐ ⇒ x1 < y1

  • r (x1 = y1 ∧ x2 < y2)

⇐ ⇒ (¬E(x, y) ∧ x < y) or (E(x, y) ∧ f(x) < f(y)).

Hence, Th(M) admits elimination of quantifiers.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 17 / 28

slide-26
SLIDE 26

Local monotonicity Example 3.3 M = (Q × Q, <, 0, f(x), E(x, y)). f((p, q)) := (q, 0) and E((p1, q1), (p2, q2)) ⇐ ⇒ p1 = p2.

Let a = (a1, a2), b = (b1, b2) ∈ M. The set {x ∈ M : f(x) = (a1, 0)} = Q × {a1} is discrete. The set {x ∈ M : E(x, a)} = {a1} × Q is convex. Hence, M is strongly locally o-minimal.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 18 / 28

slide-27
SLIDE 27

Local monotonicity Example 3.3 M = (Q × Q, <, 0, f(x), E(x, y)). f((p, q)) := (q, 0) and E((p1, q1), (p2, q2)) ⇐ ⇒ p1 = p2.

Let a = (a1, a2), b = (b1, b2) ∈ M. The set {x ∈ M : f(x) = (a1, 0)} = Q × {a1} is discrete. The set {x ∈ M : E(x, a)} = {a1} × Q is convex. Hence, M is strongly locally o-minimal. However, for open interval (a, b), the set f((a, b)) = (a1, b1) × {0} is discrete. Therefore, for any c ∈ (a, b), f(c) is not continuous.

M dose not satisfy local monotonicity with continuity.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 18 / 28

slide-28
SLIDE 28

Local cell decomposition property

For every n ∈ N, we inductively introduce cells in Mn. The definiton of cells in locally o-minimal structures is the same as that of cells in o-minimal structures.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 19 / 28

slide-29
SLIDE 29

Local cell decomposition property

For every n ∈ N, we inductively introduce cells in Mn. The definiton of cells in locally o-minimal structures is the same as that of cells in o-minimal structures.

Definition 4.1 Let C ⊆ M. C = {a}, where a ∈ M, is called a 0-cell and dim C := 0 If C is an open interval, the C is called a 1-cell and dim C := 1.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 19 / 28

slide-30
SLIDE 30

Local cell decomposition property Definition 4.2 Suppose that C ⊆ Mn is a cell with dim C = k. Let f : C → M be definable and continuous. Then C1 := graph( f) := {(x, f(x)) : x ∈ C} is a k-cell in Mn+1 and we put dim C1 = k. Let g, h be definable continuous functions from C to M ∪ {±∞} with g < h on C. Then C2 := (g, h)C := {(x, y) : x ∈ C, g(x) < y < h(x)} is a (k + 1)-cell in Mn+1 and we put dim C2 = k + 1. h(x) g(x) (g, h)C

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 20 / 28

slide-31
SLIDE 31

Local cell decomposition property Definition 4.3

1

M is said to have the cell decomposition property if for any n ∈ N and any definable X ⊆ Mn, there is a finite partition of X into cells.

2

M is said to have the local cell decomposition property if for any n ∈ N, any a ∈ Mn, and any definable X ⊆ Mn, there is an open box B ⊆ Mn containing a and a finite partition of X ∩ B into cells.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 21 / 28

slide-32
SLIDE 32

Local cell decomposition property Theorem 4.4 (Knight, Pillay and Steinhorn)

Any o-mininal structure has the cell decomposition property.

Theorem 4.5 (Kawakami and T.)

Let R = (R, <, · · · ) be a locally o-minimal expansion of (R, <).

1

R has the local cell decomposition property.

2

Let n ∈ N. Let X ⊆ Rn be a cell, f : X → R a definable function and

a ∈ Rn. Then, there exists an open box B ⊆ Rn containing a and a

finite partition P of X ∩ B into cells such that for any Y ∈ P, f|Y is continuous.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 22 / 28

slide-33
SLIDE 33

Local cell decomposition property

Actually, Theorem 4.5(1) holds in strongly locally o-minimal structures.

Proposition 4.6 (Kawakami, Takeuchi, Tsuboi, and T.)

Any strongly locally o-minimal structure M has the local cell decomposition property.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 23 / 28

slide-34
SLIDE 34

Strong local o-minimality

To show Proposition 4.6, we first give a characterization of strong local o-minimality.

Definition 4.7 Let M be an L-structure and A ⊆ M.

1

Defn(A, M) := {D ∩ An : D ⊆ Mn is M-definable}. Def(A, M) := ∪

n∈ω Defn(A, M).

2

The structure Adef := (A, (X)X∈Def(A,M)).

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 24 / 28

slide-35
SLIDE 35

Strong local o-minimality

To show Proposition 4.6, we first give a characterization of strong local o-minimality.

Definition 4.7 Let M be an L-structure and A ⊆ M.

1

Defn(A, M) := {D ∩ An : D ⊆ Mn is M-definable}. Def(A, M) := ∪

n∈ω Defn(A, M).

2

The structure Adef := (A, (X)X∈Def(A,M)). Remark 4.8

If A ⊆ M is definable, then Def(Adef) = Def(A, M), that is, for any X ⊆ A,

X is definable in Adef ⇐ ⇒ X is definable in M.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 24 / 28

slide-36
SLIDE 36

Strong local o-minimality

Tsuboi gave the following characterization of strong local

  • -minimality at Model theory summer school 2010 (August).

Proposition 4.9 The following are equivalent.

1

M is strongly locally o-minimal.

2

For any a1, . . . , an ∈ M, there are intervals I1 = (b1, c1], . . ., In = (bn, cn] with ai ∈ (bi, ci) such that, by putting I := ∪

i=1,...,n Ii,

Idef is o-minimal.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 25 / 28

slide-37
SLIDE 37

Local cell decomposition property Proposition 4.10 (Kawakami, Takeuchi, Tsuboi, and T.)

Any strongly locally o-minimal structure M has the local cell decomposition property.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 26 / 28

slide-38
SLIDE 38

Local cell decomposition property Proof.

Let a = (a1, a2, . . . , an) ∈ Mn and X ⊆ Mn definable. By the strong local o-minimality of M, there are intervals I1 = (b1, c1], . . ., In = (bn, cn] with ai ∈ (bi, ci) such that Idef := (I1 ∪ I2 ∪ · · · ∪ In)def is o-minimal.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 27 / 28

slide-39
SLIDE 39

Local cell decomposition property Proof.

Let a = (a1, a2, . . . , an) ∈ Mn and X ⊆ Mn definable. By the strong local o-minimality of M, there are intervals I1 = (b1, c1], . . ., In = (bn, cn] with ai ∈ (bi, ci) such that Idef := (I1 ∪ I2 ∪ · · · ∪ In)def is o-minimal. We put B := (b1, c1) × · · · × (bn, cn) (⊆ In). Then, (a1, . . . , an) ∈ B, and B is definable in Idef because B is definable in M.

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 27 / 28

slide-40
SLIDE 40

Local cell decomposition property Proof.

Let a = (a1, a2, . . . , an) ∈ Mn and X ⊆ Mn definable. By the strong local o-minimality of M, there are intervals I1 = (b1, c1], . . ., In = (bn, cn] with ai ∈ (bi, ci) such that Idef := (I1 ∪ I2 ∪ · · · ∪ In)def is o-minimal. We put B := (b1, c1) × · · · × (bn, cn) (⊆ In). Then, (a1, . . . , an) ∈ B, and B is definable in Idef because B is definable in M. Hence, by the o-minimality of Idef, there is a finite partition P of X ∩ B into cells in Idef. Since B is an open box, any Y ∈ P is also a cell in M. This proprosition is proved.

  • Hiroshi Tanaka (Anan National College of Technology)

Locally o-minimal structures December 1, 2010 27 / 28

slide-41
SLIDE 41

A diagram

Hiroshi Tanaka (Anan National College of Technology) Locally o-minimal structures December 1, 2010 28 / 28