A LOCAL CHARACTERIZATION OF VC-MINIMALITY
|
|
- Candice Harper
- 5 years ago
- Views:
Transcription
1 PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 00, Number 0, Pages S (XX) A LOCAL CHARACTERIZATION OF VC-MINIMALITY URI ANDREWS AND VINCENT GUINGONA Abstract. We show VC-minimality is Π 0 4 -complete. In particular, we give a local characterization of VC-minimality. We also show dp-smallness is Π complete. 1. Introduction Motivated by successes in stability theory, model theorists have recently been interested in generalizing results for stable theories to a wider class of theories. The primary candidate of study is the class of NIP theories 1, which includes mathematically important theories left out of the class of stable theories, such as the first order theory of the real field and the p-adic field. To study this large class of theories, it helps to first understand simpler cases, considering theories that are minimal with respect to various notions of dimension (e.g., Vapni-Chervonenis (VC) dimension and dp-ran). VC-minimality is a notion of simplicity for a first-order theory which simultaneously generalizes wea o-minimality and C-minimality. Until now, VC-minimality has been a very difficult notion to wor with. This difficulty is due to the complexity of the definition of VC-minimality. In particular, the definition is Σ 1 1, i.e., it requires an existential quantifier over sets of formulae. As such, it is quite difficult to verify that a theory is not VC-minimal. Instead, most instances of proofs that a theory is not VC-minimal actually show that the theory fails to satisfy one of several weaer principles such as convex orderability, dp-smallness, or dp-minimality. In this paper, we answer the following question: Question 1.1. How hard is it to determine whether or not a theory is VC-minimal? Index sets are a tool used to quantify the complexity of notions. Let P be a property of obects in a class K. Then the index set of P is the set I(P ) := {i i is an index for a recursive C K with the property P }. By restricting to the recursive C K, the complexity of this set comes from the complexity of the notion P, not the inherent complexity in the obect C. The question is formalized as asing to characterize the complexity of the index set I(VC-minimal theories). We show that in fact VC-minimality is far simpler Received by the editors March 20, The first author s research was partially supported by NSF grant DMS The second author s research was supported by NSF grant DMS This material is based upon wor supported by the NSF under Grant No while both authors were in residence at the Mathematical Sciences Research Institute in Bereley, California, during the Spring 2014 semester. 1 NIP stands for Not the Independence Property. 1 c XXXX American Mathematical Society
2 2 ANDREWS AND GUINGONA than expected, being Π 0 4-complete, and we give a simple characterization. This characterization maes VC-minimality far easier wor with. We hope that it will spur further research in the area. Our characterization, which holds for theories in countable languages, is local in the sense that it gives a condition that must hold for each formula. It is nown that VC-minimality implies convex orderability [7], which in turn implies dp-smallness [6] (which implies dp-minimality which implies NIP). So, a question naturally arises: How complex are the definitions of convex orderability and dp-smallness? We show that dp-smallness is, in fact, far more complicated than VC-minimality; it is Π 1 1-complete. In this vain, we also answer a question from [6] by giving examples of dp-small theories in countable languages which are not convexly orderable. We leave the following question open: Open Question 1.2. What is the complexity of convex orderability? 2. Bacground Let X be a set, C P(X). We say C is directed if, for all A, B C, at least one of the following holds: A B, B A, or A B =. For simplicity of notation, for A, B X, we write A B to denote that {A, B} is not directed. That is, A \ B, B \ A, and A B. Remar 2.1 (Swiss Cheese Decomposition). Suppose C P(X) is directed. If A C and B i C for i < n with B i A for all i < n and B i B = for all i, then we call S = A \ (B 0... B n 1 ) a swiss cheese, A is the wheel of S and the B i s are the holes of S. If D X is a (finite) boolean combination of elements of C, then there exists swiss cheeses S 0,..., S m 1 such that S i S = for all i, no wheel of some S i is equal to a hole of some S, and D = S 0... S m 1. We call such S 0,..., S m 1 a swiss cheese decomposition of D. See Lemma 2.1 of [4] for more details. By Theorem 3.1 of [4], there is a means of canonically choosing a decomposition, so we may consider the swiss cheese decomposition of D. Lemma 2.2 (Union of Chains). If C P(X) is directed, C 0 C is a chain, and A := C 0, then C {A} is directed. Proof. Fix B C. We must show that either A B =, A B, or B A. If any B C 0 contains B, then B A, and we are done. Similarly, if every B C 0 is disoint from B, then A B =. The remaining case is where some B C 0 intersects B, but none contains B. Thus this B is contained in B. As C 0 is a chain, every element intersects B and none contains B, so every member of the chain is contained in B. Thus A B.
3 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 3 Definition 2.3. Fix a language L, an L-theory T, and a monster model U = T. For a tuple of variables x, let x denote the length of x and let U x = U x. If ϕ(x; y) is a formula and b U y, then let ϕ(u; b) be the set of all a U x satisfying the formula ϕ(a; b). We say a set of partitioned L-formulae Ψ = {ψ i (x; y i ) i I} is directed if the set C Ψ := {ψ i (U; b) i I, b U yi } is directed (in the ambient set U x ). We say that the theory T is VC-minimal if, there exists a directed set of formulae Ψ (in the free variable x with x = 1) such that, every (parameter) definable set A U is a (finite) boolean combination of elements from C Ψ. In this case, we call Ψ a generating family for T. If Ψ is a generating family for T, then a set ψ i (U, b) for ψ i Ψ is called a ball in Ψ. If C P(X) is directed, then C {{a} a X} is directed. Therefore, without loss of generality, we may assume the formula x = y is in the generating family of any VC-minimal theory. An L-structure M is called convexly orderable if there exists a linear order on M (not necessarily definable) such that, for all L-formulas ϕ(x; y) with x = 1, there exists < ω such that, for all b M y, ϕ(m; b) is a union of at most -convex subsets of M. By Proposition 2.3 of [7], if M is convexly orderable and N M, then N is convexly orderable, so convex orderability is a property of theories. By Theorem 2.4 of [7], any VC-minimal theory is convexly orderable. 3. Devastation and Immortality The following is a technical definition which plays an important role in our local characterization of VC-minimality. Definition 3.1 (Devastation, Immortality). Suppose that ψ(x; y) is a partitioned L-formula and ϕ(x) is an L(U)-formula, both with a common free variable, x. We say that ϕ devastates ψ if there exists a sequence c i : i < ω of elements in U y such that, for all i < < ω, = x(ψ(x; c i ) ψ(x; c ) ϕ(x)), and = x(ψ(x; c i ) ψ(x; c ) ϕ(x)). If there exists no L(U)-formula ϕ(x) which devastates ψ(x; y), then we say that ψ(x; y) is immortal. Remar 3.2. If ϕ(x; y) is an L-formula such that, for all d U y, ϕ(x; d) does not devastate ψ(x; z), then by compactness there exists < ω such that, for all d U y, there does not exist c i : i < from U z so that for all i < <, = x(ψ(x; c i ) ψ(x; c ) ϕ(x, d)), and = x(ψ(x; c i ) ψ(x; c ) ϕ(x, d)). It follows that immortality of ψ(x, y) in a recursive theory T is a Π 0 2 condition defined by ϕ(x; y) T θ where θ is the sentence saying there are no c i : i < as above. If ϕ(x) devastates ψ(x; y) witnessed by c := c i : i < ω, then we may assume c is indiscernible (i.e., any two finite subsequences of c satisfy all the same L-formulae). To see this, use compactness to reduce to only finitely many L-formulae and a finite length sequence, then use Ramsey s Theorem to choose a finite subsequence of c where all the selected L-formulae either hold or fail similarly across the subsequence.
4 4 ANDREWS AND GUINGONA Lemma 3.3 (Directed and Devastated). Suppose ϕ(x) is an L(U)-formula, and ψ(x; y) is a directed L-formula. Then ϕ(x) devastates ψ(x; y) if and only if there exists an indiscernible sequence c i : i < ω in U y such that one of the following hold: (1) for all i < ω, = x(ψ(x; c i+1 ) ψ(x; c i )), = x(ψ(x; c i ) ψ(x; c i+1 ) ϕ(x)), and = x(ψ(x; c i ) ψ(x; c i+1 ) ϕ(x)); or (2) for all i < ω, = x(ψ(x; c i+1 ) ψ(x; c i )), = x(ψ(x; c i ) ϕ(x)), and = x(ψ(x; c i ) ϕ(x)). Proof. If (1) or (2) hold, then clearly ϕ(x) devastates ψ(x; y). Conversely, if ϕ(x) devastates ψ(x; y), then by Remar 3.2 we can assume the witness c i : i < ω is indiscernible. Therefore, we have either that, for all i < ω, ψ(u; c i+1 ) ψ(u; c i ) or, for all i < < ω, ψ(u; c i ) ψ(u; c ) =. Now (1) or (2) follow from each case. Definition 3.4. If ϕ(x; y) is any formula and a is any parameter, we refer to ϕ(x; a) as an instance of ϕ. Definition 3.5 (Instance Sums). Fix L-formulae ϕ(x; y) and ψ(x; z). Then their instance sum is the following formula (ϕ ψ)(x; y, z, w 0, w 1 ) := (w 0 = w 1 ϕ(x; y)) (w 0 w 1 ψ(x; z)). Remar 3.6 (On Instance Sums). If ϕ(x; y) and ψ(x; z) are L-formulae, then each instance of (ϕ ψ) is T -equivalent to either an instance of ϕ or an instance of ψ. Conversely, each instance of ϕ and each instance of ψ is T -equivalent to an instance of (ϕ ψ). If ϕ(x; y) and ψ(x; z) are immortal L-formulae, then (ϕ ψ) is immortal. If δ(x) devastates (ϕ ψ), then by the pigeonhole principle, either δ devastates ϕ or δ devastates ψ. This contradicts the assumption that both formulae are immortal. If {ϕ(x; y), ψ(x; z)} is directed, then (ϕ ψ) is directed. Lemma 3.7 (Balls are Immortal). If T is VC-minimal and ψ(x; z) is in the generating family of T, then ψ is immortal. Proof. Suppose, by means of contradiction, that ψ(x; z) is in the generating family of T but ψ is not immortal. Therefore, there exists an L(U)-formula ϕ(x) which devastates ψ. Then by Lemma 3.3, there are a i, b i U x and c i U z such that, for all i < < ω, a i ϕ(u) ψ(u; c i ) \ ψ(u; c ), and b i ϕ(u) ψ(u; c i ) \ ψ(u; c ). Since T is VC-minimal, ϕ(u) has a swiss cheese decomposition, namely S 0,..., S m 1 as in Remar 2.1. Therefore, by the pigeonhole principle, for some < m we have infinitely many i < ω such that a i S. Let S = S and, without loss of generality, suppose all a i S. Let A be the wheel and B 0,..., B m 1 be the holes of S (if S has no holes, we get a contradiction, since ψ(u; c i ) S for any i < ω). By the pigeonhole principle again, there exists < m and infinitely many i < ω such that b i B. Let B = B. For each i 1: since b i B, B ψ(u, c i ). Since b i 1 B, B ψ(u, c i ). Thus ψ(u, c i ) B. But now a i B, so a i / S, which contradicts our choice of S.
5 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 5 4. Local characterization of VC-minimality Theorem 4.1 (Local Characterization of VC-Minimality). For a theory T in a countable language L, the following are equivalent: (1) T is VC-minimal, (2) for all L-formulae ϕ(x; y), there exists an immortal directed L-formula ψ(x; z) such that each instance of ϕ is T -equivalent to a (finite) boolean combination of instances of ψ. Since compactness shows that if every instance of ϕ is equivalent to a boolean combination of instances of ψ, then there is an n so that every instance of ϕ is a boolean combination of n instances of ψ, this shows that the index set of VC-minimal theories is Π 0 4. Remar 4.2. Our restriction to a countable language is necessary. Consider the example in the language L = {P i i < ω 1 } with ℵ 1 -many unary predicates and let T be the L-theory which says that, for all finite disoint I, J ω 1, there are infinitely many x such that P (x). i I P i (x) J This theory has quantifier elimination and is superstable. One can easily chec it satisfies condition (2) of Theorem 4.1, but this is not VC-minimal (see Example 2.10 of [7] for more details). Lemma 4.3 (Main Construction Lemma). If ϕ(x; y) and ψ(x; z) are each a directed immortal formula (not assuming {ϕ, ψ} is directed), then there exists δ(x; w) an immortal formula such that {ψ, δ} is directed, and each instance of ϕ is a finite boolean combination of instances of ψ and δ. As the proof of the Main Construction Lemma is somewhat involved and combinatorial, we leave it to Section 8. We now consider the proof of Theorem 4.1, given the Main Construction Lemma. Proof of Theorem 4.1. (1) (2): Suppose T is VC-minimal and fix a L-formula ϕ(x; y). By compactness, there exists a directed family of finitely many L-formulae {ψ i (x; z i ) i < } such that each instance of ϕ is T -equivalent to a boolean combination of instances of the ψ i s. By taing instance sums, we may assume that = 1. By Lemma 3.7, ψ is immortal. (2) (1): We construct Ψ the generating family by induction. First, since L is countable, there exists an enumeration {ϕ i (x; y i ) i < ω} of the L-formulae with x (where x = 1) as a free variable. Let Ψ 0 = and suppose that we have Ψ i a finite directed set of immortal L-formulae constructed so that, for all < i, each instance of ϕ is T -equivalent to a boolean combination of instances of elements from Ψ +1. Suppose further that Ψ Ψ +1 for all < i. Now consider ϕ i (x; y i ) and let ψ(x; z) be given as in (2) (hence ψ is immortal and directed). Let ψ (x; z ) be the instance sum of Ψ i, which is immortal and directed by Remar 3.6. By Lemma 4.3, there exists δ(x; w) an immortal L-formula such that {δ, ψ } is directed and each instance of ψ is T -equivalent to a boolean combination of instances of ψ and δ. Therefore, each instance of ϕ is T -equivalent to a boolean combination
6 6 ANDREWS AND GUINGONA of instances of ψ and δ. Let Ψ i+1 := Ψ i {δ}, which is a finite directed set of immortal L-formulae. Finally, let Ψ = i Ψ i. 5. Stable VC-Minimal Theories A formula ϕ(x; y) has the order property if there exists elements a i U x for i < ω and b U y for < ω such that ϕ(a i ; b ) if and only if i <. A theory T is stable if no formula has the order property. Lemma 5.1. Suppose T is VC-minimal and stable. Then, there exists Ψ := {E i (x, y) i I} a directed set of equivalence relations (on x with x = 1) that is a generating family for T. Proof. Since T is VC-minimal, let Ψ be a generating family for T. Now fix ψ(x; y) Ψ and p(y) S y ( ) (i.e., p is a maximally consistent set of L-formulae with free variable y). Suppose, by means of contradiction, that the type p(y 0 ) p(y 1 ) {ψ(u; y 0 ) ψ(u; y 1 )} is consistent. Tae b 0, b 1 a witness to this and tae σ Aut(U) sending b 0 to b 1. Let b n = σ n (b 0 ) (in particular, this is consistent with the naming of b 1 ). Then, b i : i < ω and ψ is a witness to the (strict) order property, a contradiction to the fact that T is stable. Therefore, there exists δ(y) p(y) such that, for all b 0, b 1 U yi with = δ(b 0 ) δ(b 1 ), either ψ(u; b 0 ) = ψ(u; b 1 ) or ψ(u; b 0 ) ψ(u; b 1 ) =. In other words, the formula E ψ,p (x 0, x 1 ) := ( y)(δ(y) ψ(x 0 ; y) ψ(x 1 ; y)) (x 0 = x 1 ) is a -definable equivalence relation. Now tae Ψ := {E ψ,p ψ(x; y) Ψ, p S y ( )}. We claim that Ψ is a generating family for T. To show this, we simply show C Ψ = C Ψ. For A C Ψ, A = ψ(u; b) for some ψ(x; y) Ψ, b U y. Then, for any a A, one can chec that A = E ψ,tp(b) (U; a). Conversely, tae A C Ψ, so A = E ψ,p (U, a) for some ψ(x; y) Ψ, p S( ), and a U. Let δ(y) p(y) be the associated formula. If there exists b U y such that = δ(b) ψ(a; b), then ψ(u; b) = A, hence A C Ψ. On the other hand, if there exists no such b, then A = {a} so, since (x = y) Ψ, A C Ψ. So, without loss of generality, when dealing with a VC-minimal stable theory, we may assume the generating family is a set of equivalence relations on the home sort. As a corollary of Theorem 4.1, we get the following characterization of stable VC-minimal theories. Theorem 5.2. Suppose T is a stable theory in a countable language. The following are equivalent: (1) T is VC-minimal, (2) For each formula ϕ(x; y), there exist finitely many refining definable equivalence relations {E (x 0, x 1 ) < m}, each of which is immortal, such that, for all b U y, ϕ(u; b) is a (finite) boolean combination of instances of the E s.
7 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 7 Figure 1. Example construction where 1 enters W g(i,) at stage 3. V0 Z0 V1 Z1 V2 Z2 E 6. Π 0 4-completeness of VC-minimality We now show that the characterization of VC-minimality given in Theorem 4.1 is the simplest possible. Theorem 6.1. The index set of VC-minimal theories is Π 0 4-hard. Proof. We describe a recursive function f, which, on a given input i, outputs a theory T i so that T i is always ℵ 0 -stable, and T i is VC-minimal if and only if i S for a Π 0 4-complete set S. We have S written as (W g(i,) is co-finite) for a fixed recursive function g. Our theory will be in the language L := {E ω} {U ω} {V, Z, ω} where each E is binary and all other relations are unary. T i begins with the following axioms: The U s define disoint infinite sets. Each E is an equivalence relation on U with infinitely many infinite classes. The V s define disoint subsets of U. If x V, y V l for l, then E (x, y). For each, ω: There are infinitely many E -classes which do not intersect V. For each, ω: For each E -class A which intersects V, both A V and A V are infinite. For each, ω: x Z if and only if x / V E (x, y) y V. At stage s, for each s, we add the following axioms to T i : and there is a y so that If / Wg(i,) s, then add an axiom stating that there are at least s E -classes which intersect V. If enters W g(i,) at stage s, add an axiom stating that there are exactly s E -classes which intersect V. See Figure 1 for details. Lemma 6.2. For every i, T i is a complete ℵ 0 -stable theory with quantifier elimination.
8 8 ANDREWS AND GUINGONA Proof. For quantifier elimination, it suffices to show that we can eliminate ( x) from a conunction of formulae of the form U (x), ±E (x, y l ), ±V (x), and ±Z (x) for some fixed. As any E -class which intersects V is contained in V Z, this is straightforward. Moreover, a simple type-counting argument shows that T i is ℵ 0 -stable. Lemma 6.3. If (W g(i,) is co-finite), then T i is VC-minimal. Proof. For each, let S be the set ω W g(i,). Each S is finite, by assumption. Define X to be the set of elements in U, but not in any V or Z for S. Let Φ be the family composed of the following families of definable sets: {U ω} {V, Z S, ω} {X ω} {E (x, y) x V S } {E (x, y) x Z S } {E (x, y) x X } {E (x, y) V l (x) l / S } {E (x, y) Z l (x) l / S } It is immediate that Φ is directed. For l / S, V l is a finite union of instances of {E (x, y) V l (x)}. A similar condition holds for Z l. Each E -class is the union of elements of Φ given by the fourth, fifth, and sixth lines. By quantifier elimination, every definable set is a boolean combination of instances from Φ. Thus Φ witnesses VC-minimality of T i. Lemma 6.4. If (W g(i,) is co-infinite), then T i is non-vc-minimal. Proof. Fix so W g(i,) is co-infinite. Let ψ be a directed formula so that every instance of E is a boolean combination of instances of ψ. By Lemma 5.1, ψ can be assumed to be comprised of equivalence relations. By quantifier elimination, instances of ψ are E -classes away from finitely many exceptional V and Z. Let / W g(i,) not be one of those finitely many exceptional. Thus V intersects infinitely many E -classes. This shows that V devastates ψ. Thus ψ cannot be contained in any family witnessing VC-minimality of T i by Lemma 3.7, and thus T i is non-vc-minimal. Corollary 6.5. The index set of VC-minimal theories is Π 0 4-complete. Remar 6.6. One should note that all the theories T i constructed in Theorem 6.1 are, in fact, convexly orderable. This gives us a large list of examples of theories that are ℵ 0 -stable and convexly orderable but not VC-minimal. 7. Complexity of dp-smallness Definition 7.1. We say a theory T is dp-small if there does not exist an L-formula ϕ(x; y), a sequence b i : i < ω, and L(U)-formulae ψ (x) (where x = 1) such that, for all i, < ω, the following partial type is consistent with T : {ϕ(x; b i ), ψ (x)} { ϕ(x; b i ) i i} { ψ (x) }.
9 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 9 Figure 2. Example construction of a particular tree. <1,2> U<0> E U<> <0> <1> <> U<1,2> U<1> By Proposition 1.5 of [6], if a theory is convexly orderable, then it is dp-small. In particular, all VC-minimal theories are dp-small. A theory being not dp-small is clearly a Σ 1 1 condition, hence the index of dp-small theories is Π 1 1. Theorem 7.2. The index set of dp-small theories is Π 1 1-complete. Proof. We use the fact that {T ω <ω T is a recursive tree with no path} is Π 1 1- complete (see Theorem 5.14 of [3]). Given a (recursive index for a) tree T ω <ω, we produce a theory so that the tree T has a path if and only if the theory is not dp-small. We fix the language L := {E} {U σ σ ω <ω } where E is binary and each U σ is unary. The theory is axiomatized as follows: E is an equivalence relation with infinitely many infinite classes. xu (x) If σ and τ are incomparable, then x U σ and y U τ implies E(x, y). If σ τ, then U τ U σ. If τ / T, then U τ = If τ = σ i, and τ T, then there is an infinite set S of E-equivalence classes so that for each E-equivalence class A S, U τ A is an infinite coinfinite subset of U σ A. Further, there are infinitely many E-equivalence classes which intersect U σ which do not intersect U τ. See Figure 2 for an example. It is straightforward to verify that the theory produced is complete for any T and is dp-small if and only if T has no infinite path. For example, if there is an infinite path, say σ 0 σ 1..., tae ϕ(x; y) := E(x, y), ψ (x) := U σ (x) U σ+1 (x) for ω, and pairwise non-e-related b i s outside of U σ0, each E-related to elements in U σ for all. This is a witness to the non-dp-smallness of the theory. To conclude this section, we use the ideas behind the construction in Theorem 6.1 to provide an answer to a question from [6]. Example 7.3. We give an example of a theory in a countable language that is dp-small but not convexly orderable, answering a question from [6]. This theory happens to be ℵ 0 -stable. Let L = {E} {U i, i < ω}, where E is a binary relation and each U i, is a unary relation. Let T be the L-theory which says
10 10 ANDREWS AND GUINGONA E is an equivalence relation with infinitely many infinite classes; the U i, are pairwise disoint; for all i < ω, U i,0... U i,i is a union of infinitely many E-classes; if an E-class intersects U i,, it does so with infinitely many points and it intersects each U i, for i; and if an E-class intersects U i,, it does not intersect U i, for i i. This is ℵ 0 -stable and has quantifier elimination. Suppose, by means of contradiction, that it were convexly orderable, say with on M = T. Then, there exists < ω such that, for all a M, E(M; a) is a union of at most -convex sets. Loo at U 2, for 2. Again, by convex orderability, there exists l < ω such that each U 2, (M) is a union of a most l -convex sets. Let B,m for m < l be the mth -convex component of U 2, (M) (some may be empty). By the pigeonhole principle, there exists m 0,..., m < l and an infinite collection of E-classes A 0, A 1,... such that B,m A t for all and t < ω. As the B,m are -convex and pairwise disoint, and each intersect A 0 and M \ A 0, we must have that A 0 is a union of at least + 1 -convex sets. This is a contradiction. However, this theory is dp-small. Suppose, by means of contradiction, that ϕ(x; y) together with ψ l (x) for l < ω is a witness to non-dp-smallness. That is, there exists b i : i < ω such that, for all i, l < ω, the partial type {ϕ(x; b i ), ψ l (x)} { ϕ(x; b i ) i i} { ψ l (x) l l} is consistent. By quantifier elimination, we may assume ϕ is E with perhaps a restriction to some U i, and that the ψ l (x) are of the form U i, perhaps restricted to an E-class. One checs such formulae cannot mae the above partial type consistent. 8. The Main Construction Lemma Suppose M is a countable model of a theory T in a countable language. In this section, for simplicity of exposition, for a formula ϕ(x; y) and b M y, we will write ϕ b to mean ϕ(m; b). Lemma 8.1 (Unions and Intersections of chains). Suppose ρ(x; y) and τ(x; z) are so that for any y, z, ρ y τ z. Let χ be any union of a chain of instances of ρ or intersection of a chain of instances of ρ. Then for every z, τ z χ. Proof. We first suppose χ is a union of a chain of instances of ρ. Suppose z is so that τ z χ. Let a be in the intersection and b be in χ τ z. Let ρ w be in the chain so that it contains a and b. Then ρ w τ z, which is a contradiction. Now suppose χ is an intersection of a chain of instances of ρ. Suppose z is so that τ z χ. Let a be any element of τ z χ and choose w so ρ w does not contain a. Then ρ w τ z, which is a contradiction. Lemma 8.2 (Main Construction Lemma). If ϕ(x; y) and ψ(x; z) are each a directed immortal formula (not assuming {ϕ, ψ} is directed), then there exists δ(x; w) an immortal formula such that {ψ, δ} is directed, and each instance of ϕ is a finite boolean combination of instances of ψ and δ. Proof. We begin by defining the following formulae: θ 0 a := {x ϕ a z(x ψ z ψ z ϕ a )}
11 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 11 φa Figure 3. Example of θ 0, θ 1, θ 2, and θ 3. 1 θa,z ψz θa 0 ψz' ψz'' 3 θa,z'' 2 θa,z' If ϕ a ψ z, then define θ 1 a,z := {ψ y ψ y ϕ a = ψ z ϕ a }. Otherwise, θ 1 a,z :=. If ϕ a ψ z, then define θ 2 a,z := {ψ y ψ y ϕ a = ψ z ϕ a }. Otherwise, θ 2 a,z :=. θ 3 a,z := (θ 2 a,z ϕ a ) {ψ z (ψ z ϕ a ) ψ z ϕ a θ 2 a,z ϕ a } See Figure 3 for an example. We intend to show that {ψ(x; y), θ 0 (x; a), θ 1 (x; a, z), θ 2 (x; a, z), θ 3 (x; a, z)} is directed, each θ i is immortal, and that each instance of ϕ is a boolean combination of instances from this family. Lemma 8.3. {ψ(x; y), θ 0 (x; a), θ 1 (x; a, z), θ 2 (x; a, z), θ 3 (x; a, z)} is directed Proof. For each pair of formulae from {ψ(x; y), θ 0 (x; a), θ 1 (x; a, z), θ 2 (x; a, z), θ 3 (x; a, z)}, we argue that no two instances can be. ψ, θ 0 : If x ψ y θa, 0 then ψ y ϕ a. If ψ y ϕ a, then θa 0 ϕ a ψ y. So we suppose ψ y ϕ a. Tae any x ψ y. If x were in some ψ y where ψ y ϕ a, then x ψ y ψ y, contradicting x θa. 0 Thus ψ y θa. 0 ψ, θ 1 : This follows from Lemma 8.1 since any instance of θ 1 is a union of a chain of instances of ψ and ψ is directed. ψ, θ 2 : This follows from Lemma 8.1 since any instance of θ 2 is an intersection of a chain of instances of ψ and ψ is directed. ψ, θ 3 : Let ψ y intersect θa,z. 3 If ψ y ψ z ϕ a, then ψ y θa,z 2 θa,z. 3 So, we may assume ψ y ϕ a ψ z ϕ a. If for some w, ψ w ϕ a and ψ w ψ y and ψ w θa,z, 2 then ψ y is explicitly excluded from θa,z 3 and the intersection is empty. Otherwise, ψ y θa,z 2 ϕ a and it is contained in θa,z. 3 θ 0, θ 0 : We may assume ϕ a ϕ a. If there is no x θa 0 and y so x ψ y and ψ y ϕ a, then θa 0 θ0 a. Otherwise, this ψ y must contain ϕ a, since ψ y ϕ a. Thus θa 0 ϕ a and ϕ a θ0 a =. θ 0, θ 3 : Let x θa 0 θa 3,v. Let S be the set of v so that ψ v ϕ a = ψ v ϕ a. Then x ψ v for every v S. So ψ v ϕ a for each v S. If ψ v ϕ a for any v S, then ψ v θa 0 (see ψ, θ 0 ), so θa 3,v ψ v θa. 0 So we assume
12 12 ANDREWS AND GUINGONA ϕ a ψ v for each such v. Thus ϕ a θa 2,v. Thus ϕ a ϕ a. If there is a z so that θa 0 ψ z and ψ z ϕ a and ψ z ϕ a θa 2,v ϕ a, then θ0 a θa 3,v =. Otherwise, θa 0 θa 3,v. θ 1, anything: Since no instance of ψ is to any instance of a θ i, this follows by Lemma 8.1. θ 2, anything: Since no instance of ψ is to any instance of a θ i, this follows by Lemma 8.1. θa,z, 3 θa 3,z : As θ3 a,z ϕ a, we may assume ϕ a ϕ a. Similarly, we may assume either θa,z 2 θa 2,z or vice versa. We start with the first case: θ2 a,z θa 2,z. Let S be the set of w so that ψ w ϕ a = ψ z ϕ a. Then for every w S, since ψ w ϕ a, it follows that ψ w ϕ a. Since {ψ, θ 2 } is directed, either ψ w is a proper subset of θa 2,z and is thus excluded from θ3 a,z or ψ w contains θa 2,z. In the first case, θ3 a,z is disoint from θa 3,z, so we suppose the second case holds for every w S. Thus θa,z 2 θa 2,z. It remains to chec that any ψ y contained in θa,z 2 excluded from θa,z 3 is also excluded from θa 3,z. If ψ y ϕ a, then ψ y ϕ a and if it defines a proper subset of θa 2,z ϕ a, then it defines a proper subset of θa,z 2 ϕ a, as needed. Now we consider the second case: θa 2,z θ2 a,z. If θa 2,z ϕ a θa,z 2 ϕ a, then using a small enough instance of ψ w where ψ w ϕ a = ψ z ϕ a, we see that θa 2,z is excluded from θ3 a,z. Thus we may assume θa 2,z ϕ a = θa,z 2 ϕ a. It remains to see that any instance of ψ y omitted from θa,z 3 is also omitted from θa 3,z. Since ϕ a ϕ a, if ψ y intersects θa 2,z and ψ y ϕ a, then ψ y ϕ a. Thus if ψ y is omitted in the definition of θa,z, 3 it is also omitted in the definition of θa 3,z. Thus θ3 a,z θa 3,z. Lemma 8.4. Suppose ρ(x; y) is an immortal formula, and that each instance of χ(x; z) is a union of a chain of instances of ρ. Then χ is immortal. Suppose ρ(x; y) is an immortal formula, and that each instance of χ(x; z) is an intersection of a chain of instances of ρ. Then χ is immortal. Proof. First we consider the case where every instance of χ(x; z) is a union of a chain of instances of ρ. Suppose towards a contradiction that γ(x) devastates χ witnessed by the indiscernible c i : i < ω. For all i < < ω, χ ci χ c intersects both γ and γ. Since each χ instance is a union of a chain of instances of ρ, there exists d i for each i < ω so that ρ di χ ci and for all i <, ρ di χ c intersects both γ and γ. This witnesses that γ devastates ρ, contrary to the assumption of ρ s immortality. Now we consider the case where every instance of χ(x; z) is an intersection of a chain of instances of ρ. Suppose towards a contradiction that γ(x) devastates χ witnessed by the indiscernible c i : i < ω. For all i < < ω, χ ci χ c intersects both γ and γ. Since each χ instance is an intersection of a chain of instances of ρ, there exists d i for each i < ω so that χ ci ρ di and for all i <, ρ di ρ d intersects both γ and γ. This witnesses again that γ devastates ρ, contrary to the assumption of ρ s immortality. Lemma 8.5. θ 0 is immortal.
13 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 13 Proof. Towards a contradiction, suppose γ devastates θ 0 and consider the indiscernible sequence a i : i < ω witnessing this as in Lemma 3.3. If ϕ a0 ϕ a1 =, then γ devastates ϕ by indiscernibility, contradicting the immortality of ϕ. If ϕ a0 ϕ a1, then one of two cases holds: (i) There exists z such that ϕ a0 ψ z and ψ z ϕ a1. In this case, θa 0 1 ϕ a0 =, hence θa 0 0 and θa 0 1 are disoint and θa 0 1 (ϕ a1 \ ϕ a0 ). Therefore, by indiscernibility, γ devastates ϕ, contrary to assumption. (ii) There exists no such z. Then ϕ a0 θa 0 1, hence θa 0 0 θa 0 1, but this contradicts the choice of the a i s in Lemma 3.3. Similarly, if ϕ a1 ϕ a0 and there exists z such that ϕ a1 ψ z and ψ z ϕ a0, then this contradicts the immortality of ϕ. Therefore, we must have that ϕ a1 ϕ a0 and no such z exists. Hence θa 0 1 θa 0 0. As γ does not devastate ϕ, we must have that (ϕ a0 \ ϕ a1 ) is contained in either γ or γ. Without loss of generality, suppose it is contained in γ. Then, by indiscernibility, (ϕ ai \ ϕ ai+1 ) γ for all i < ω. Notice that γ (θa 0 0 \ θa 0 1 ) by assumption, so choose x in this set. As x / θa 0 1, there exists z such that x ψ z and ψ z ϕ a1, hence ψ z θa 0 1 =. However, since x θa 0 0 and {ψ, θ 0 } is directed, we must have that ψ z θa 0 0. Therefore, ψ z (θa 0 0 \ θa 0 1 ). Moreover, as ψ z ϕ a1 and ψ z ϕ a0, we have that ψ z (ϕ a0 \ ϕ a1 ). Hence, ψ z γ. By indiscernibility, there are z i such that ψ zi γ, ψ zi γ, and ψ zi (θa 0 i \ θa 0 i+1 ). In particular, the ψ zi s are disoint. Hence, γ devastates ψ, contrary to immortality of ψ. Lemma 8.6. θ 1 is immortal Proof. This follows from Lemma 8.4. Lemma 8.7. θ 2 is immortal Proof. This follows from Lemma 8.4. Lemma 8.8. θ 3 is immortal. Proof. For this proof, let θ 4 a,b,z := (θ 2 a,z ϕ b ) {ψ z (ψ z ϕ b ) ψ z ϕ b θ 2 a,z ϕ b }. In particular, θa,a,z 4 = θa,z, 3 so it suffices to show θ 4 is immortal. By means of contradiction, suppose γ devastates θ 4, and consider the indiscernible sequence a i, b i, z i : i < ω witnessing this as in Lemma 3.3. Fix any i. Since θ 2 is directed, we have three cases: (i) θa 2 i,z i θa 2,z =, (ii) θa 2 i,z i θa 2,z, or (iii) θa 2,z θa 2 i,z i. For Case (i), since γ devastates θ 4 and θa 4 i,b i,z i θa 2 i,z i, we have that γ devastates θ 2 by indiscernibility. Case (ii) and (iii) are symmetric, so let us suppose that Case (ii) holds. In almost the exact same way as one shows that ψ y θa,z 3 for any
14 14 ANDREWS AND GUINGONA a, y, z, one can show that ψ y θa,b,z 4 for any a, b, y, z. Hence, θ2 c,y θa,b,z 4 for any a, b, c, y, z by Lemma 8.1. Thus, there are three subcases: (a) θa 2 i,z i θa 4,b,z =, (b) θa 2 i,z i θa 4,b,z, or (c) θa 4,b,z θa 2 i,z i. In Case (a), (θ 4 a,b,z \ θ 4 a i,b i,z i ) (θ 2 a,z \ θ 2 a i,z i ), therefore γ devastates θ 2 by indiscernibility. In Case (b), fix < i < or > i >. Then, by indiscernibility, θ 2 a,d θ 4 a i,b i,z i and, by definition, θ 4 a,b,z θ 2 a,z. Hence, (θ 4 a,b,z \ θ 4 a i,b i,z i ) (θ 2 a,z \ θ 2 a,d ). Therefore, γ devastates θ 2 by indiscernibility. Hence Case (c) must hold. Together, (ii) and (c) imply θa 4 i,b,z i = θa 4,b,z. Hence, by indiscernibility, we may assume there are a and z such that, for all i < ω, θa,b 4 = i,z θ4 a i,b i,z i. We now consider the sequence a, b i, z : i < ω which witnesses that θ 4 is devastated by γ. If ϕ b0 ϕ b1 =, then, as θa,b 4 ϕ i,z b i for all i, γ devastates ϕ. So we may assume that ϕ b1 ϕ b0 (note that, if ϕ b0 ϕ b1, then θa,b 4 0,z θ4 a,b 1,z, contrary to this sequence witnessing devastation of θ 4 ). If both γ and γ intersect ϕ b0 \ ϕ b1, then γ devastates ϕ. So, without loss of generality (and by indiscernibility), we may assume ϕ bi \ ϕ bi+1 γ for all i < ω. In particular, note that γ must intersect i<ω ϕ b i. Since γ intersects θa,b 4 \ 0,z θ4 a,b 1,z, there exists w such that ψ w intersects γ, ψ w θa,z, 2 ψ w ϕ b1, and ψ w ϕ b0. In particular, ψ w intersects (ϕ b0 \ ϕ b1 ), hence also γ. For all i < ω, ψ w does not contain (ϕ bi \ ϕ bi+1 ) as otherwise θ 4 a,b i,z = θ4 a,b i+1,z, contrary to the choice of b i. On the other hand, for all but finitely many i, ψ w does not intersect (ϕ bi \ ϕ bi+1 ), as otherwise ψ w would devastate ϕ. By removing finitely many and reindexing, we may assume ψ w is disoint from (ϕ bi \ ϕ bi+1 ) for all i > 1. By indiscernibility, for each i < ω, there exists w i such that ψ wi intersects γ and γ, ψ wi intersects (ϕ b2i \ ϕ b2i+1 ), and ψ wi is disoint from (ϕ b2 \ ϕ b2+1 ) for all i (the last condition is clear for > i and, for < i, note that ψ wi ϕ b2i, hence ψ wi is disoint from (ϕ b2 \ ϕ b2+1 )). In particular, since ψ is directed, the last two conditions imply that the ψ wi s are disoint. Hence, by the first condition, γ devastates ψ, contrary to immortality of ψ. Lemma 8.9. For any c, ϕ c is a boolean combination of instances from {ψ, θ 0, θ 1, θ 2, θ 3 }. Proof. Every element in ϕ c is either in θc 0 or is in some θc,z. 1 We first note that there is a finite set of instances of θc,z 1 which suffices to cover (ϕ c θc). 0 Otherwise, we could choose more and more instances of θc,z 1 which would witness that ϕ c devastates θ 1.
15 A LOCAL CHARACTERIZATION OF VC-MINIMALITY 15 We now define a sequence of sets whose union will be ϕ c. Set Y 0 = θc. 0 Suppose we have defined the sets Y for < i. Suppose further that there is a finite set S i 1 of elements so that ( w S i 1 θc,w 1 ϕ c ) = (ϕ c <i Y ). Now we define Y i := w S i 1 ((θ 1 c,w θ 2 c,w) θ 3 c,w). To complete the recursive definition of the sequence of sets Y i for i < ω, we need to see that there is a finite set S i so that ( w S i θc,w 1 ϕ c ) = (ϕ c i Y ). We build S i as follows: Having selected elements a 0,... a 1 so that ( < θ1 c,a ϕ c ) (ϕ c i Y ), we need to select an element a. Fix an element x ϕ c ( i Y < θ1 c,a ) and let a be an element so x θc,a 1. By directedness of {θ 0, θ 1, θ 3 }, θc,a 1 ϕ c ( i Y < θ1 c,a ). This process must stop, yielding a finite set S i, as otherwise ϕ c devastates θ 1. It remains to see that for some i, i Y = ϕ c. Otherwise there is an infinite sequence of b i for i ω so that b i S i for each i and θc,b 1 i+1 θc,b 1 i, and this sequence witnesses that ϕ devastates θ 1. As each Y is a boolean combination of instances from {ψ, θ 0, θ 1, θ 2, θ 3 }, ϕ c is a boolean combination of instances from {ψ, θ 0, θ 1, θ 2, θ 3 }. References [1] H. Adler, Theories controlled by formulas of Vapni-Chervonenis codimension 1 (2008). Preprint. [2] U. Andrews, S. Cotter, J. Freitag, and A. Medvedev, VC-minimality: examples and observations (2014). Preprint. [3] C. J. Ash and J. Knight, Computable structures and the hyperarithmetical hierarchy, Studies in Logic and the Foundations of Mathematics, vol. 144, North-Holland Publishing Co., Amsterdam, [4] J. Flenner and V. Guingona, Canonical forests in directed families, Proc. Am. Math. Soc. 142 (2014), [5], Convexly orderable groups and valued fields, J. Symbolic Logic 79 (2014), [6] V. Guingona, On VC-minimal fields and dp-smallness, Arch. Math. Logic (2013). to appear. [7] V. Guingona and M. C. Lasowsi, On VC-Minimal Theories and Variants, Archive for Mathematical Logic (2011). to appear. address: andrews@math.wisc.edu URL: Department of Mathematics, University of Wisconsin, Madison, WI , USA address: guingona.1@nd.edu URL: University of Notre Dame, Department of Mathematics, 255 Hurley, Notre Dame, IN 46556, USA
ON VC-MINIMAL THEORIES AND VARIANTS. 1. Introduction
ON VC-MINIMAL THEORIES AND VARIANTS VINCENT GUINGONA AND MICHAEL C. LASKOWSKI Abstract. In this paper, we study VC-minimal theories and explore related concepts. We first define the notion of convex orderablity
More informationVC-MINIMALITY: EXAMPLES AND OBSERVATIONS
VC-MINIMALITY: EXAMPLES AND OBSERVATIONS URI ANDREWS, SARAH COTTER BLANSET, JAMES FREITAG, AND ALICE MEDVEDEV Abstract. VC-minimality is a recent notion in model theory which generalizes strong minimality,
More informationCONVEXLY ORDERABLE GROUPS AND VALUED FIELDS
CONVEXLY ORDERABLE GROUPS AND VALUED FIELDS JOSEPH FLENNER AND VINCENT GUINGONA Abstract. We consider the model theoretic notion of convex orderability, which fits strictly between the notions of VC-minimality
More informationOn VC-Minimality in Algebraic Structures
On VC-Minimality in Algebraic Structures Vincent Guingona University of Notre Dame 3 April 2012 For the 2012 ASL Annual North American Meeting, Madison, Wisconsin Outline 1 VC-Minimality Convex Orderability
More informationHence C has VC-dimension d iff. π C (d) = 2 d. (4) C has VC-density l if there exists K R such that, for all
1. Computational Learning Abstract. I discuss the basics of computational learning theory, including concept classes, VC-dimension, and VC-density. Next, I touch on the VC-Theorem, ɛ-nets, and the (p,
More informationON VC-MINIMAL FIELDS AND DP-SMALLNESS
ON VC-MINIMAL FIELDS AND DP-SMALLNESS VINCENT GUINGONA Abstract. In this paper, we show that VC-minimal ordered fields are real closed. We introduce a notion, strictly between convexly orderable and dp-minimal,
More informationVC-DENSITY FOR TREES
VC-DENSITY FOR TREES ANTON BOBKOV Abstract. We show that for the theory of infinite trees we have vc(n) = n for all n. VC density was introduced in [1] by Aschenbrenner, Dolich, Haskell, MacPherson, and
More informationProperly forking formulas in Urysohn spaces
Properly forking formulas in Urysohn spaces Gabriel Conant University of Notre Dame gconant@nd.edu March 4, 2016 Abstract In this informal note, we demonstrate the existence of forking and nondividing
More informationAMS regional meeting Bloomington, IN April 1, 2017
Joint work with: W. Boney, S. Friedman, C. Laskowski, M. Koerwien, S. Shelah, I. Souldatos University of Illinois at Chicago AMS regional meeting Bloomington, IN April 1, 2017 Cantor s Middle Attic Uncountable
More informationSPECTRA OF ATOMIC THEORIES
SPECTRA OF ATOMIC THEORIES URI ANDREWS AND JULIA F. KNIGHT Abstract. For a countable structure B, the spectrum is the set of Turing degrees of isomorphic copies of B. For complete elementary first order
More informationStrong theories, burden, and weight
Strong theories, burden, and weight Hans Adler 13th March 2007 Abstract We introduce the notion of the burden of a partial type in a complete first-order theory and call a theory strong if all types have
More informationON VC-DENSITY IN VC-MINIMAL THEORIES
ON VC-DENSITY IN VC-MINIMAL THEORIES VINCENT GUINGONA Abstract. We show that any formula with two free variables in a VC-minimal theory has VC-codensity at most two. Modifying the argument slightly, we
More informationω-stable Theories: Introduction
ω-stable Theories: Introduction 1 ω - Stable/Totally Transcendental Theories Throughout let T be a complete theory in a countable language L having infinite models. For an L-structure M and A M let Sn
More informationVAUGHT S THEOREM: THE FINITE SPECTRUM OF COMPLETE THEORIES IN ℵ 0. Contents
VAUGHT S THEOREM: THE FINITE SPECTRUM OF COMPLETE THEORIES IN ℵ 0 BENJAMIN LEDEAUX Abstract. This expository paper introduces model theory with a focus on countable models of complete theories. Vaught
More informationAmalgamation and the finite model property
Amalgamation and the finite model property Alex Kruckman University of California, Berkeley March 25, 2015 Alex Kruckman (UC Berkeley) Amalgamation and the FMP March 25, 2015 1 / 16 The motivating phenomenon
More informationDisjoint n-amalgamation
October 13, 2015 Varieties of background theme: the role of infinitary logic Goals 1 study n- toward 1 existence/ of atomic models in uncountable cardinals. 2 0-1-laws 2 History, aec, and Neo-stability
More informationGödel s Completeness Theorem
A.Miller M571 Spring 2002 Gödel s Completeness Theorem We only consider countable languages L for first order logic with equality which have only predicate symbols and constant symbols. We regard the symbols
More informationLöwenheim-Skolem Theorems, Countable Approximations, and L ω. David W. Kueker (Lecture Notes, Fall 2007)
Löwenheim-Skolem Theorems, Countable Approximations, and L ω 0. Introduction David W. Kueker (Lecture Notes, Fall 2007) In its simplest form the Löwenheim-Skolem Theorem for L ω1 ω states that if σ L ω1
More informationMarch 3, The large and small in model theory: What are the amalgamation spectra of. infinitary classes? John T. Baldwin
large and large and March 3, 2015 Characterizing cardinals by L ω1,ω large and L ω1,ω satisfies downward Lowenheim Skolem to ℵ 0 for sentences. It does not satisfy upward Lowenheim Skolem. Definition sentence
More informationStable embeddedness and N IP
Stable embeddedness and N IP Anand Pillay University of Leeds January 14, 2010 Abstract We give some sufficient conditions for a predicate P in a complete theory T to be stably embedded. Let P be P with
More informationHINDMAN S THEOREM AND IDEMPOTENT TYPES. 1. Introduction
HINDMAN S THEOREM AND IDEMPOTENT TYPES URI ANDREWS AND ISAAC GOLDBRING Abstract. Motivated by a question of Di Nasso, we show that Hindman s Theorem is equivalent to the existence of idempotent types in
More informationA trichotomy of countable, stable, unsuperstable theories
A trichotomy of countable, stable, unsuperstable theories Michael C. Laskowski Department of Mathematics University of Maryland S. Shelah Department of Mathematics Hebrew University of Jerusalem Department
More informationDIVIDING AND WEAK QUASI-DIMENSIONS IN ARBITRARY THEORIES
DIVIDING AND WEAK QUASI-DIMENSIONS IN ARBITRARY THEORIES ISAAC GODBRING AND HENRY TOWSNER Abstract. We show that any countable model of a model complete theory has an elementary extension with a pseudofinite-like
More informationLINDSTRÖM S THEOREM SALMAN SIDDIQI
LINDSTRÖM S THEOREM SALMAN SIDDIQI Abstract. This paper attempts to serve as an introduction to abstract model theory. We introduce the notion of abstract logics, explore first-order logic as an instance
More informationARTEM CHERNIKOV AND SERGEI STARCHENKO
A NOTE ON THE ERDŐS-HAJNAL PROPERTY FOR STABLE GRAPHS arxiv:1504.08252v2 [math.lo] 11 Apr 2016 ARTEM CHERNIKOV AND SERGEI STARCHENKO Abstract. In this short note we provide a relatively simple proof of
More informationScott Sentences in Uncountable Structures
Rose-Hulman Undergraduate Mathematics Journal Volume 18 Issue 1 Article 14 Scott Sentences in Uncountable Structures Brian Tyrrell Trinity College Dublin Follow this and additional works at: http://scholar.rose-hulman.edu/rhumj
More informationHrushovski s Fusion. A. Baudisch, A. Martin-Pizarro, M. Ziegler March 4, 2007
Hrushovski s Fusion A. Baudisch, A. Martin-Pizarro, M. Ziegler March 4, 2007 Abstract We present a detailed and simplified exposition of Hrushovki s fusion of two strongly minimal theories. 1 Introduction
More informationNIP THEORIES AND SHELAH S THEOREM
NIP THEORIES AND SHELAH S THEOREM These are the informal and expanded notes of a mini-course given in the Universidad Autónoma de Madrid, 29-30 of May 2018 (6 hours). The mini-course was given at the end
More informationPETER A. CHOLAK, PETER GERDES, AND KAREN LANGE
D-MAXIMAL SETS PETER A. CHOLAK, PETER GERDES, AND KAREN LANGE Abstract. Soare [20] proved that the maximal sets form an orbit in E. We consider here D-maximal sets, generalizations of maximal sets introduced
More informationThe Canonical Ramsey Theorem and Computability Theory
The Canonical Ramsey Theorem and Computability Theory Joseph R. Mileti August 29, 2005 Abstract Using the tools of computability theory and reverse mathematics, we study the complexity of two partition
More informationQualifying Exam Logic August 2005
Instructions: Qualifying Exam Logic August 2005 If you signed up for Computability Theory, do two E and two C problems. If you signed up for Model Theory, do two E and two M problems. If you signed up
More informationForcing in Lukasiewicz logic
Forcing in Lukasiewicz logic a joint work with Antonio Di Nola and George Georgescu Luca Spada lspada@unisa.it Department of Mathematics University of Salerno 3 rd MATHLOGAPS Workshop Aussois, 24 th 30
More informationWeight and measure in NIP theories
Weight and measure in NIP theories Anand Pillay University of Leeds September 18, 2011 Abstract We initiate an account of Shelah s notion of strong dependence in terms of generically stable measures, proving
More informationIntroduction to Model Theory
Introduction to Model Theory Charles Steinhorn, Vassar College Katrin Tent, University of Münster CIRM, January 8, 2018 The three lectures Introduction to basic model theory Focus on Definability More
More informationGREGORY TREES, THE CONTINUUM, AND MARTIN S AXIOM
The Journal of Symbolic Logic Volume 00, Number 0, XXX 0000 GREGORY TREES, THE CONTINUUM, AND MARTIN S AXIOM KENNETH KUNEN AND DILIP RAGHAVAN Abstract. We continue the investigation of Gregory trees and
More informationPETER A. CHOLAK, PETER GERDES, AND KAREN LANGE
D-MAXIMAL SETS PETER A. CHOLAK, PETER GERDES, AND KAREN LANGE Abstract. Soare [23] proved that the maximal sets form an orbit in E. We consider here D-maximal sets, generalizations of maximal sets introduced
More informationFirst-Order Logic. 1 Syntax. Domain of Discourse. FO Vocabulary. Terms
First-Order Logic 1 Syntax Domain of Discourse The domain of discourse for first order logic is FO structures or models. A FO structure contains Relations Functions Constants (functions of arity 0) FO
More informationA Local Normal Form Theorem for Infinitary Logic with Unary Quantifiers
mlq header will be provided by the publisher Local Normal Form Theorem for Infinitary Logic with Unary Quantifiers H. Jerome Keisler 1 and Wafik Boulos Lotfallah 2 1 Department of Mathematics University
More informationDENSELY k-separable COMPACTA ARE DENSELY SEPARABLE
DENSELY k-separable COMPACTA ARE DENSELY SEPARABLE ALAN DOW AND ISTVÁN JUHÁSZ Abstract. A space has σ-compact tightness if the closures of σ-compact subsets determines the topology. We consider a dense
More informationDefinable henselian valuation rings
Definable henselian valuation rings Alexander Prestel Abstract We give model theoretic criteria for the existence of and - formulas in the ring language to define uniformly the valuation rings O of models
More informationThe Vaught Conjecture Do uncountable models count?
The Vaught Conjecture Do uncountable models count? John T. Baldwin Department of Mathematics, Statistics and Computer Science University of Illinois at Chicago May 22, 2005 1 Is the Vaught Conjecture model
More informationLimit computable integer parts
Limit computable integer parts Paola D Aquino, Julia Knight, and Karen Lange July 12, 2010 Abstract Let R be a real closed field. An integer part I for R is a discretely ordered subring such that for every
More informationarxiv:math/ v1 [math.lo] 5 Mar 2007
Topological Semantics and Decidability Dmitry Sustretov arxiv:math/0703106v1 [math.lo] 5 Mar 2007 March 6, 2008 Abstract It is well-known that the basic modal logic of all topological spaces is S4. However,
More informationMath 225A Model Theory. Speirs, Martin
Math 5A Model Theory Speirs, Martin Autumn 013 General Information These notes are based on a course in Metamathematics taught by Professor Thomas Scanlon at UC Berkeley in the Autumn of 013. The course
More informationRepresenting Scott Sets in Algebraic Settings
Representing Scott Sets in Algebraic Settings Alf Dolich Kingsborough Community College Julia F. Knight University of Notre Dame Karen Lange Wellesley College David Marker University of Illinois at Chicago
More informationThe complexity of recursive constraint satisfaction problems.
The complexity of recursive constraint satisfaction problems. Victor W. Marek Department of Computer Science University of Kentucky Lexington, KY 40506, USA marek@cs.uky.edu Jeffrey B. Remmel Department
More informationCategoricity Without Equality
Categoricity Without Equality H. Jerome Keisler and Arnold W. Miller Abstract We study categoricity in power for reduced models of first order logic without equality. 1 Introduction The object of this
More informationRepresenting Scott Sets in Algebraic Settings
Wellesley College Wellesley College Digital Scholarship and Archive Faculty Research and Scholarship 8-2015 Representing Scott Sets in Algebraic Settings Alf Dolich Julia F. Knight Karen Lange klange2@wellesley.edu
More informationGEOMETRIC STRUCTURES WITH A DENSE INDEPENDENT SUBSET
GEOMETRIC STRUCTURES WITH A DENSE INDEPENDENT SUBSET ALEXANDER BERENSTEIN AND EVGUENI VASSILIEV Abstract. We generalize the work of [13] on expansions of o-minimal structures with dense independent subsets,
More informationSTABILITY AND POSETS
STABILITY AND POSETS CARL G. JOCKUSCH, JR., BART KASTERMANS, STEFFEN LEMPP, MANUEL LERMAN, AND REED SOLOMON Abstract. Hirschfeldt and Shore have introduced a notion of stability for infinite posets. We
More informationMODEL THEORY FOR ALGEBRAIC GEOMETRY
MODEL THEORY FOR ALGEBRAIC GEOMETRY VICTOR ZHANG Abstract. We demonstrate how several problems of algebraic geometry, i.e. Ax-Grothendieck, Hilbert s Nullstellensatz, Noether- Ostrowski, and Hilbert s
More informationA NEW LINDELOF SPACE WITH POINTS G δ
A NEW LINDELOF SPACE WITH POINTS G δ ALAN DOW Abstract. We prove that implies there is a zero-dimensional Hausdorff Lindelöf space of cardinality 2 ℵ1 which has points G δ. In addition, this space has
More informationarxiv:math.lo/ v1 28 Nov 2004
HILBERT SPACES WITH GENERIC GROUPS OF AUTOMORPHISMS arxiv:math.lo/0411625 v1 28 Nov 2004 ALEXANDER BERENSTEIN Abstract. Let G be a countable group. We proof that there is a model companion for the approximate
More informationPosets, homomorphisms and homogeneity
Posets, homomorphisms and homogeneity Peter J. Cameron and D. Lockett School of Mathematical Sciences Queen Mary, University of London Mile End Road London E1 4NS, U.K. Abstract Jarik Nešetřil suggested
More informationShort notes on Axioms of set theory, Well orderings and Ordinal Numbers
Short notes on Axioms of set theory, Well orderings and Ordinal Numbers August 29, 2013 1 Logic and Notation Any formula in Mathematics can be stated using the symbols and the variables,,,, =, (, ) v j
More informationEXCURSIONS IN MODEL THEORY
EXCURSIONS IN MODEL THEORY RAFAEL WINGESTER RIBEIRO DE OLIVEIRA Abstract. This paper aims to introduce the reader familiar with undergraduate level logic to some fundamental constructions in Model Theory.
More informationMore Model Theory Notes
More Model Theory Notes Miscellaneous information, loosely organized. 1. Kinds of Models A countable homogeneous model M is one such that, for any partial elementary map f : A M with A M finite, and any
More informationNOTES ON WELL ORDERING AND ORDINAL NUMBERS. 1. Logic and Notation Any formula in Mathematics can be stated using the symbols
NOTES ON WELL ORDERING AND ORDINAL NUMBERS TH. SCHLUMPRECHT 1. Logic and Notation Any formula in Mathematics can be stated using the symbols,,,, =, (, ) and the variables v j : where j is a natural number.
More informationModel theory of bounded arithmetic with applications to independence results. Morteza Moniri
Model theory of bounded arithmetic with applications to independence results Morteza Moniri Abstract In this paper we apply some new and some old methods in order to construct classical and intuitionistic
More informationAN INTRODUCTION TO GEOMETRIC STABILITY THEORY
AN INTRODUCTION TO GEOMETRIC STABILITY THEORY SALMAN SIDDIQI Abstract. In this paper, we will introduce some of the most basic concepts in geometric stability theory, and attempt to state a dichotomy theorem
More informationHerbrand Theorem, Equality, and Compactness
CSC 438F/2404F Notes (S. Cook and T. Pitassi) Fall, 2014 Herbrand Theorem, Equality, and Compactness The Herbrand Theorem We now consider a complete method for proving the unsatisfiability of sets of first-order
More informationA NOTE ON THE EIGHTFOLD WAY
A NOTE ON THE EIGHTFOLD WAY THOMAS GILTON AND JOHN KRUEGER Abstract. Assuming the existence of a Mahlo cardinal, we construct a model in which there exists an ω 2 -Aronszajn tree, the ω 1 -approachability
More informationPRESERVATION THEOREMS IN LUKASIEWICZ MODEL THEORY
Iranian Journal of Fuzzy Systems Vol. 10, No. 3, (2013) pp. 103-113 103 PRESERVATION THEOREMS IN LUKASIEWICZ MODEL THEORY S. M. BAGHERI AND M. MONIRI Abstract. We present some model theoretic results for
More informationThe Countable Henkin Principle
The Countable Henkin Principle Robert Goldblatt Abstract. This is a revised and extended version of an article which encapsulates a key aspect of the Henkin method in a general result about the existence
More informationA Hanf number for saturation and omission: the superstable case
A Hanf number for saturation and omission: the superstable case John T. Baldwin University of Illinois at Chicago Saharon Shelah The Hebrew University of Jerusalem Rutgers University April 29, 2013 Abstract
More informationMathematics 114L Spring 2018 D.A. Martin. Mathematical Logic
Mathematics 114L Spring 2018 D.A. Martin Mathematical Logic 1 First-Order Languages. Symbols. All first-order languages we consider will have the following symbols: (i) variables v 1, v 2, v 3,... ; (ii)
More informationReverse mathematics and uniformity in proofs without excluded middle
Reverse mathematics and uniformity in proofs without excluded middle Jeffry L. Hirst jlh@math.appstate.edu Carl Mummert mummertcb@appstate.edu Appalachian State University Submitted for publication: 5/3/2006
More informationTHE CHARACTERISTIC SEQUENCE OF A FIRST-ORDER FORMULA
THE CHARACTERISTIC SEQUENCE OF A FIRST-ORDER FORMULA M. E. MALLIARIS Abstract. For a first-order formula ϕ(x; y) we introduce and study the characteristic sequence P n : n < ω of hypergraphs defined by
More informationTRIVIAL, STRONGLY MINIMAL THEORIES ARE MODEL COMPLETE AFTER NAMING CONSTANTS
TRIVIAL, STRONGLY MINIMAL THEORIES ARE MODEL COMPLETE AFTER NAMING CONSTANTS SERGEY S. GONCHAROV, VALENTINA S. HARIZANOV, MICHAEL C. LASKOWSKI, STEFFEN LEMPP, AND CHARLES F. D. MCCOY Abstract. We prove
More informationFINITE MODEL THEORY (MATH 285D, UCLA, WINTER 2017) LECTURE NOTES IN PROGRESS
FINITE MODEL THEORY (MATH 285D, UCLA, WINTER 2017) LECTURE NOTES IN PROGRESS ARTEM CHERNIKOV 1. Intro Motivated by connections with computational complexity (mostly a part of computer scientice today).
More informationThe Relation Reflection Scheme
The Relation Reflection Scheme Peter Aczel petera@cs.man.ac.uk Schools of Mathematics and Computer Science The University of Manchester September 14, 2007 1 Introduction In this paper we introduce a new
More informationPartial cubes: structures, characterizations, and constructions
Partial cubes: structures, characterizations, and constructions Sergei Ovchinnikov San Francisco State University, Mathematics Department, 1600 Holloway Ave., San Francisco, CA 94132 Abstract Partial cubes
More informationCOMPUTING AND DOMINATING THE RYLL-NARDZEWSKI FUNCTION
COMPUTING AND DOMINATING THE RYLL-NARDZEWSKI FUNCTION URI ANDREWS AND ASHER M. KACH Abstract. We study, for a countably categorical theory T, the complexity of computing and the complexity of dominating
More informationINTRODUCTION TO CARDINAL NUMBERS
INTRODUCTION TO CARDINAL NUMBERS TOM CUCHTA 1. Introduction This paper was written as a final project for the 2013 Summer Session of Mathematical Logic 1 at Missouri S&T. We intend to present a short discussion
More informationTRANSFERING SATURATION, THE FINITE COVER PROPERTY, AND STABILITY
TRANSFERING SATURATION, THE FINITE COVER PROPERTY, AND STABILITY John T. Baldwin Department of Mathematics University of Illinois at Chicago Chicago, IL 60680 Rami Grossberg Department of Mathematics Carnegie
More informationLecture 2: Syntax. January 24, 2018
Lecture 2: Syntax January 24, 2018 We now review the basic definitions of first-order logic in more detail. Recall that a language consists of a collection of symbols {P i }, each of which has some specified
More informationProof Theory and Subsystems of Second-Order Arithmetic
Proof Theory and Subsystems of Second-Order Arithmetic 1. Background and Motivation Why use proof theory to study theories of arithmetic? 2. Conservation Results Showing that if a theory T 1 proves ϕ,
More informationOctober 12, Complexity and Absoluteness in L ω1,ω. John T. Baldwin. Measuring complexity. Complexity of. concepts. to first order.
October 12, 2010 Sacks Dicta... the central notions of model theory are absolute absoluteness, unlike cardinality, is a logical concept. That is why model theory does not founder on that rock of undecidability,
More informationMeta-logic derivation rules
Meta-logic derivation rules Hans Halvorson February 19, 2013 Recall that the goal of this course is to learn how to prove things about (as opposed to by means of ) classical first-order logic. So, we will
More informationIntroduction. Itaï Ben-Yaacov C. Ward Henson. September American Institute of Mathematics Workshop. Continuous logic Continuous model theory
Itaï Ben-Yaacov C. Ward Henson American Institute of Mathematics Workshop September 2006 Outline Continuous logic 1 Continuous logic 2 The metric on S n (T ) Origins Continuous logic Many classes of (complete)
More informationA generalization of modal definability
A generalization of modal definability Tin Perkov Polytechnic of Zagreb Abstract. Known results on global definability in basic modal logic are generalized in the following sense. A class of Kripke models
More informationPart II. Logic and Set Theory. Year
Part II Year 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2018 60 Paper 4, Section II 16G State and prove the ǫ-recursion Theorem. [You may assume the Principle of ǫ- Induction.]
More informationtp(c/a) tp(c/ab) T h(m M ) is assumed in the background.
Model Theory II. 80824 22.10.2006-22.01-2007 (not: 17.12) Time: The first meeting will be on SUNDAY, OCT. 22, 10-12, room 209. We will try to make this time change permanent. Please write ehud@math.huji.ac.il
More informationKRIPKE S THEORY OF TRUTH 1. INTRODUCTION
KRIPKE S THEORY OF TRUTH RICHARD G HECK, JR 1. INTRODUCTION The purpose of this note is to give a simple, easily accessible proof of the existence of the minimal fixed point, and of various maximal fixed
More informationIntroduction to Model Theory
Introduction to Model Theory Jouko Väänänen 1,2 1 Department of Mathematics and Statistics, University of Helsinki 2 Institute for Logic, Language and Computation, University of Amsterdam Beijing, June
More informationINTERPRETING HASSON S EXAMPLE
INTERPRETING HASSON S EXAMPLE CHARLES K. SMART Abstract. We generalize Ziegler s fusion result [8] by relaxing the definability of degree requirement. As an application, we show that an example proposed
More informationA computably stable structure with no Scott family of finitary formulas
A computably stable structure with no Scott family of finitary formulas Peter Cholak Department of Mathematics, University of Notre Dame Richard A. Shore Department of Mathematics, Cornell University Reed
More informationStable embeddedness and N IP
Stable embeddedness and N IP arxiv:1001.0515v1 [math.lo] 4 Jan 2010 Anand Pillay University of Leeds January 4, 2010 Abstract We give some sufficient conditions for a predicate P in a complete theory T
More informationContinuum Harvard. April 11, Constructing Borel Models in the. Continuum Harvard. John T. Baldwin. University of Illinois at Chicago
April 11, 2013 Today s Topics 1 2 3 4 5 6 Pseudo-minimal 7 Further Applications Section 1: { Models in L ω1,ω L ω1,ω satisfies downward Löwenheim Skolem to ℵ 0 for sentences. It does not satisfy upward
More informationPreliminaries. Introduction to EF-games. Inexpressivity results for first-order logic. Normal forms for first-order logic
Introduction to EF-games Inexpressivity results for first-order logic Normal forms for first-order logic Algorithms and complexity for specific classes of structures General complexity bounds Preliminaries
More information6 CARDINALITY OF SETS
6 CARDINALITY OF SETS MATH10111 - Foundations of Pure Mathematics We all have an idea of what it means to count a finite collection of objects, but we must be careful to define rigorously what it means
More informationMORLEY S CATEGORICITY THEOREM
MORLEY S CATEGORICITY THEOREM NICK RAMSEY Abstract. A theory is called κ-categorical, or categorical in power κ, if it has one model up to isomorphism of cardinality κ. Morley s Categoricity Theorem states
More informationAnnals of Pure and Applied Logic
Annals of Pure and Applied Logic 161 (2010) 944 955 Contents lists available at ScienceDirect Annals of Pure and Applied Logic journal homepage: www.elsevier.com/locate/apal Types directed by constants
More informationMathematical Logic. Reasoning in First Order Logic. Chiara Ghidini. FBK-IRST, Trento, Italy
Reasoning in First Order Logic FBK-IRST, Trento, Italy April 12, 2013 Reasoning tasks in FOL Model checking Question: Is φ true in the interpretation I with the assignment a? Answer: Yes if I = φ[a]. No
More informationSUBSPACES OF COMPUTABLE VECTOR SPACES
SUBSPACES OF COMPUTABLE VECTOR SPACES RODNEY G. DOWNEY, DENIS R. HIRSCHFELDT, ASHER M. KACH, STEFFEN LEMPP, JOSEPH R. MILETI, AND ANTONIO MONTALBÁN Abstract. We show that the existence of a nontrivial
More informationA BOREL SOLUTION TO THE HORN-TARSKI PROBLEM. MSC 2000: 03E05, 03E20, 06A10 Keywords: Chain Conditions, Boolean Algebras.
A BOREL SOLUTION TO THE HORN-TARSKI PROBLEM STEVO TODORCEVIC Abstract. We describe a Borel poset satisfying the σ-finite chain condition but failing to satisfy the σ-bounded chain condition. MSC 2000:
More informationTR : Possible World Semantics for First Order LP
City University of New York (CUNY) CUNY Academic Works Computer Science Technical Reports Graduate Center 2011 TR-2011010: Possible World Semantics for First Order LP Melvin Fitting Follow this and additional
More informationForking and Dividing in Random Graphs
Forking and Dividing in Random Graphs Gabriel Conant UIC Graduate Student Conference in Logic University of Notre Dame April 28-29, 2012 Gabriel Conant (UIC) Forking and Dividing in Random Graphs April
More informationINDEPENDENCE RELATIONS IN RANDOMIZATIONS
INDEPENDENE RELATIONS IN RANDOMIZATIONS URI ANDREWS, ISAA GOLDBRING, AND H. JEROME KEISLER Abstract. The randomization of a complete first order theory T is the complete continuous theory T R with two
More informationPropositional and Predicate Logic - VII
Propositional and Predicate Logic - VII Petr Gregor KTIML MFF UK WS 2015/2016 Petr Gregor (KTIML MFF UK) Propositional and Predicate Logic - VII WS 2015/2016 1 / 11 Theory Validity in a theory A theory
More information