FilaProtaAGT.dvi @ Applied General Topology c© Universidad Politécnica de Valencia Volume 9, No. 2, 2008 pp. 169-175 Spread of balleans M. Filali and I. V. Protasov Abstract. A ballean is a set endowed with some family of balls in such a way that a ballean can be considered as an asymptotic coun- terpart of a uniform topological space. We introduce and study a new cardinal invariant called the spread of a ballean. In particular, we show that, for every ordinal ballean B, spread of B coincides with density of B. 2000 AMS Classification: 54A25, 54E25, 05A18. Keywords: ballean, pseudodiscrete subset, density, cellularity, spread. 1. Introduction A ball structure is a triplet B = (X, P, B), where X, P are non-empty sets and, for any x ∈ X and α ∈ P , B(x, α) is a subset of X which is called the ball of radius α around x. It is supposed that x ∈ B(x, α) for all x ∈ X, α ∈ P . The set X is called the support of B, P is called the set of radiuses. Given any x ∈ X, A ⊆ X, α ∈ P , we put B⋆(x, α) = {y ∈ X : x ∈ B(y, α)}, B(A, α) = ⋃ a∈A B(a, α). A ball structure B is called a ballean (or a coarse structure) if • for any α, β ∈ P , there exist α′, β′ ∈ P such that, for every x ∈ X, B(x, α) ⊆ B⋆(x, α′), B⋆(x, β) ⊆ B(x, β′); • for any α, β ∈ P , there exists γ ∈ P such that, for every x ∈ X, B(B(x, α), β) ⊆ B(x, γ); • for any x, y ∈ X, there exists α ∈ P such that y ∈ B(x, α). Let B1 = (X1, P1, B1), B2 = (X2, P2, B2) be balleans. A mapping f : X1 −→ X2 is called a ≺- mapping if, for every α ∈ P1, there exists β ∈ P2 such that f (B1(x, α)) ⊆ B2(f (x), β) for every x ∈ X. If f is a bijection such that f 170 M. Filali and I. V. Protasov and f −1 are ≺- mappings, we say that f is an asymorphism. If X1 = X2 and the identity mapping id : X1 −→ X2 is an asymorphism, we identify B1 and B2, and write B1 = B2. For each ballean B = (X, P, B), replacing every ball B(x, α) with B(x, α) ⋂ B∗(x, α), we obtain the same ballean, so every ballean can be determined in such a way that B(x, α) = B∗(x, α) for all x ∈ X, α ∈ P . For motivations for the study of balleans as the asymptotic counterparts of the uniform topological spaces see [1, 5, 6]. 2. Spread and density Let B = (X, P, B) be a ballean. A subset V ⊆ X is called bounded if there exist x ∈ X and α ∈ P such that V ⊆ B(x, α). A ballean is called bounded if its support X is bounded. Given a ballean B = (X, P, B), we say that a subset Y ⊆ X is pseudodis- crete, if for every α ∈ P , there exists a bounded subset V of X such that B(y, α) ⋂ Y = {y} for every y ∈ Y \ V . A ballean B is called pseudodiscrete if its support X is pseudodiscrete. For every subset Y ⊆ X, we put | Y |B= min{|Y \V | : V is a bounded subset of X}, and introduce a new cardinal invariant spread(B) = sup{|Y |B : Y is a pseudodiscrete subset of X}. We note that |Y |B = 0 if and only if Y is bounded, so spread(B) = 0 for every bounded ballean. A subset L of X is called large if there exists α ∈ P such that X = B(L, α). The density of B is defined in [4] by den(B) = min{|L| : L is a large subset of X}. Clearly, den(B) = 1 for every bounded ballean B, and den(B) is an infinite cardinal for every unbounded ballean B. Proposition 2.1. For every ballean B, we have spread(B) ≤ den(B). Proof. Let L be a large subset of X and Y be a pseudodiscrete subset of X. It suffices to show that |L| ≥ |Y \V | for some bounded subset V of X. We may suppose that B(x, α) = B∗(x, α) for all x ∈ X, α ∈ P . We take β ∈ P such that X = B(L, β), and choose γ ∈ P such that B(B(x, β), β) ⊆ B(x, γ) for each x ∈ X. Since Y is pseudodiscrete, there exists a bounded subset V of X such that B(y, γ) ⋂ Y = {y} for each y ∈ Y \ V. By the choice of γ, the family {B(y, β) : y ∈ Y \ V } is disjoint. Since B(x, β) ⋂ L 6= ∅ for each x ∈ X, we have L ⋂ B(y, β) 6= ∅ for each y ∈ Y \ V. Hence |L| ≥ |Y \ V |, as required. � In the next example, for every infinite cardinal γ, we construct a ballean B such that den(B) = γ but spread(B) = 0. Spread of balleans 171 Example 2.2. Let X be a set of cardinality γ, κ be an infinite regular cardinal such that κ ≤ γ. We denote by F the family of all subsets of X of cardinality < κ. Let P be the set of all mappings f : X −→ F such that, for every x ∈ X, we have x ∈ f (x) and |{y ∈ X : x ∈ f (y)}| < κ. Given any x ∈ X and α ∈ P, we put B(x, f ) = f (x) and note that the ball structure B = (X, P, B) is a ballean. We need the regularity of κ to state that |B(B(x, f ), g)| < κ for all x ∈ X and f, g ∈ P. Note that a subset V of X is bounded if and only if |V | < κ; and a subset L of X is large if and only if |L| = γ implying that den(B) = γ. Now we check that spread(B) = 0. To this end, we take an arbitrary subset Y of X such that |Y | ≥ κ, write it as Y = {yλ : λ ∈ |Y |}, and define a mapping f : X −→ F by the rule: f (yλ) = {yλ, yλ+1} for each λ < |Y |, and f (x) = {x} for each x ∈ X \ V . Then f ∈ P and | B(y, f ) ⋂ Y |= 2 for every y ∈ Y , so Y is not pseudodiscrete and spread(B) = 0. For every ballean B = (X, P, B), we use the preodering ≤ on X defined by the rule: α ≤ β if and only if B(x, α) ⊆ B(x, β) for every x ∈ X. A subset P ′ ⊆ P is called cofinal if, for every α ∈ P , there exists α′ ∈ P ′ such that α′ ≥ α. The cofinality cf (B) is the minimal cardinality of the cofinal subsets of P . A ballean B is called ordinal if P contains a cofinal subset of P ′ which is well-ordered by ≤. Replacing P ′ with its minimal cofinal subset, we get the same ballean. Hence, we can write B as (X, p, B), where p is a regular cardinal (considered as a set of ordinals). Let (X, d) be a metric space. For all x ∈ X and r ∈ R+, we put Bd(x, r) = {y ∈ X : d(x, y) ≤ r} and get the metric ballean B(X, d) = (X, R+, Bd). Clearly, every metric ballean is ordinal. We shall show in Theorem 2.3 that spread(B) = den(B) for every unbounded ordinal ballean B. To this end we use another cardinal invariant of a ballean B = (X, P, B), the cellularity of B, defined in [4]. A subset Y of X is called thick if, for every α ∈ P, there exists y ∈ Y such that B(y, α) ⊆ Y . The cellularity of B is the cardinal cell(B) = sup{|F| : F is a disjoint family of thick subsets of X}. By [4, Theorem 1], for every ordinal ballean B, we have cell(B) = den(B) and there exists a disjoint family F of cardinality den(B) consisting of thick subsets of X. For every infinite cardinal κ, there exists a metric ballean B with den(B) = κ (see [4, Example 1]). Theorem 2.3. For every unbounded ordinal ballean B with support X, we have spread(B) = den(B) and there exists a subset Y of X such that |Y |B = |Y | = den(B). 172 M. Filali and I. V. Protasov Proof. Let B = (X, ρ, B) where ρ is an infinite regular cardinal, κ = den(B) and cf (κ) be the cofinality of κ. Let {Fλ : λ ∈ κ} be a disjoint family of thick subsets of X and put F = ⋃ λ∈κ Fλ. We fix some element x0 ∈ X and consider four cases. Case ρ < cf (κ): We prove the following auxiliary statement. For every α ∈ ρ, there exists a bounded subset Z of F such that |Z| = κ and the family {B(z, α) : z ∈ Z} is disjoint. For every λ ∈ κ, we take y(λ) ∈ Fλ such that B(y(λ), λ) ⊆ Fλ, and pick f (λ) ∈ ρ such that y(λ) ∈ B(x0, f (λ)). Since f maps κ to ρ, by assumption, there exist a subset ∧ ⊆ κ and β ∈ ρ such that | ∧ | = κ and f (λ) = β for every λ ∈ ∧ . Put Z = {y(λ) : λ ∈ ∧ }. Using the auxiliary statement, we can construct inductively a family {Yα : α ∈ ρ} of bounded subsets of F such that the family {B(y, α) : y ∈ Yα} is disjoint for each α ∈ ρ, B(Yα, α) ⋂ B(Yα′ , α ′) = ∅ for all distinct α, α′ ∈ ρ, and |Yα| = κ for every α ∈ ρ. Put Y = ⋃ α∈ρ Yα. Case cf (κ) ≤ ρ < κ: Using the assumption and repeating the arguments from the previous case, we get the following auxiliary statement. For every α ∈ ρ and every cardinal κ′ < κ, there exists a bounded subset Z of F such that |Z| > κ′ and the family {B(z, α) : z ∈ Z} is disjoint. Using the auxiliary statement, we can construct inductively a family {Yα : α ∈ ρ} of bounded subsets of F and an increasing sequence (κα)α∈ρ of cardinals such that κ = sup{κα : α ∈ ρ}, the family {B(y, α) : y ∈ Yα} is disjoint for each α ∈ ρ, B(Yα, α) ⋂ B(Yα′ , α ′) = ∅ for all distinct α, α′ ∈ ρ, and |Yα| ≥ κα for every α ∈ ρ. We put Y = ⋃ α∈ρ Yα. Case ρ = κ: For every α ∈ ρ, we take yα ∈ Fα such that B(yα, α) ⊆ Fα. Put Y = {yα : α ∈ ρ}. Case ρ > κ: This variant is impossible, see Case ρ > κ in the proof of Theorem 1 from [4]. � By definition, a G−space is a set X endowed with a (left) action G × X −→ X, (g, x) 7−→ g(x) of a group G with identity e such that e(x) = x and g(h(x)) = (gh)(x) for all x ∈ X and g, h ∈ G. Now let X be an infinite transitive G−space, i.e., for any x, y ∈ X, there exists g ∈ G such that g(x) = y. Let κ be an infinite cardinal such that κ ≤ |X|, and consider Fκ = {F ⊆ G : |F | < κ, e ∈ A}. For any x ∈ X and F ∈ Fκ, we put B(x, F ) = F (x) = {f (x) : f ∈ F }, and get the ballean B(X, κ) = (X, Fκ, B). Let L be a large subset of X. We take F ∈ Fκ such that B(L, F ) = X. Since |F | < |X| and |B(L, F )| ≤ |L||F |, we have |L| = |X|, so den(B(X, κ)) = |X|. Spread of balleans 173 Theorem 2.4. Let X be an infinite transitive G-space, κ be an infinite cardinal such that κ ≤| X |. Then the following statements hold. (1) If G = X and g(x) = gx, then spread(B(X, κ)) = |X|. (2) If G is the group of all permutations of X, then spread(B(X, κ)) = 0. (3) Let ρ be an infinite cardinal such that ρ ≤ |X|, and let G be the group of all permutations of X with supp(g) < ρ, g ∈ G where supp(g) = {x ∈ X : g(x) 6= x}. If either ρ < κ, or ρ = κ and κ is regular, then spread(B(X, κ)) = |X|. If either ρ > κ, or ρ = κ and κ is singular, then spread(B(X, κ)) = 0. Proof. (1) By [2, Proposition 4.1], there exists a subset Y of G such that |Y | = |G| and |gY ⋂ Y | ≤ 3 for every g ∈ G, g 6= e. We take an arbitrary F ∈ Fκ, and let Z = F Y ∩ Y . Since |F | < κ, the subset Z of Y satisfies |Z| ≤ 3|F | < κ, i.e., Z is bounded, and B(y, F ) ⋂ Y = F y ⋂ Y = {y} for every y ∈ Y \Z. It follows that Y is pseudodiscrete. Since |Y |B = |Y |, we have spread(B(X, κ)) = |X|. (2) It suffices to show that every subset Y = {yα : α ∈ λ} of X of cardinality λ ≥ κ is not pseudodiscrete. We say that an ordinal α ∈ λ is even if either α is a limit ordinal or α = β + n for some limit ordinal β and some even natural number n. Otherwise, we say that α is odd. Then we define a permutation f of X by the rule: f (x) =      yα+1 if x = yα and α is even, yα−1 if x = yα and α is odd, x for each x ∈ X\Y. We put F = {f, e}. Clearly, |B(yα, F ) ⋂ Y | = 2 for every α ∈ λ. Since Y is not bounded, Y is not pseudodiscrete. (3) If either ρ < κ, or ρ = κ and κ is regular, we take an arbitrary subset F ∈ Fκ and put Z = ⋃ {supp(g) : g ∈ F }. Then |Z| < κ, so it is bounded, and clearly B(x, F ) ⋂ X = {x} for every x ∈ X \ Z. It follows that X is pseudodiscrete. Since |X|B = |X| we conclude that spread(B(X, κ)) = |X|. If ρ > κ we can use the arguments proving (2) to show that every subset of X of cardinality ≥ κ is not pseudodiscrete. If ρ = κ and κ is singular, we fix an arbitrary subset Y of X with |Y | = κ and partition Y = ⋃ {Yβ : β ∈ cf (κ)} so that |Yβ| < κ for every β ∈ cf (κ). For every β ∈ cf (κ), we fix a permutation fβ of X such that supp(fβ ) = Yβ , f (Yβ ) = Yβ . Then we put F = {e} ⋃ {fβ : β ∈ cf (κ)} 174 M. Filali and I. V. Protasov and note that F ∈ Fκ and |B(y, F ) ⋂ Y | ≥ 2 for every y ∈ Y . It follows that Y is not pseudodiscrete and spread(B(X, κ)) = 0. � Theorem 2.5. For every unbounded pseudodiscrete ballean B, we have den(B) = spread(B) and cell(B) = 1. Proof. Let X be the support of B. By [5, Theorem 3.6], there exists a filter ϕ on X such that ⋂ ϕ = ∅ and B = (X, ϕ, B) where B(x, F ) = { {x}, if x ∈ F ; X \ F, if x /∈ F for all x ∈ X and F ∈ ϕ. We put κ = min{|F | : F ∈ ϕ} and note that a subset L of X is large if and only if L ∈ ϕ so den(B) = κ. On the other hand, a subset V of X is bounded if and only if X\V ∈ ϕ. Hence, |X|B = κ, and so spread(B) = κ. We note also that every unbounded subset of X is thick. Hence, if ϕ is an ultrafilter then cell(B) = 1. � For every pair γ, λ of infinite cardinals with γ < λ, we construct next a ballean B such that den(B) = γ and spread(B) = λ. Example 2.6. We take a ballean B1 = (X1, P1, B1) such that spread(B1) = 0, |X1| = den(B1) = γ and each ball B1(x1, α1) is finite (see Example 2.2). Let B2 = (X2, P2, B2) be a pseudodiscrete ballean such that spread(B2) = |X2| = λ. We consider the ballean B = (X, P, B) with X = X1 × X2, P = P1 × P2 and B((x1, x2), (α1, α2)) = B(x1, α1) × B(x2, α2). Since den(B1) = γ and |X| = γ, we see that den(B) = γ. Since spread(B2) = λ, we see that spread(B) ≥ λ. Let now Z be any subset of X with |Z| > λ. Then there exist an infinite subset Y of X1 and a ∈ X2 such that Y × {a} ⊆ Z. Since Y is not pseudodiscrete in B1, Z is not pseudodiscrete in B. Hence, spread(B) = λ. We conclude the exposition with the following open questions. Problem 2.7. Given a ballean B with support X, does there exist a pseudodis- crete subset Y ⊆ X such that |Y |B = spread(B) ? Problem 2.8. Let B = (X, P, B) be a ballean, |X| = κ and let |P | ≤ κ. Assume that there exists κ′ < κ such that | B(x, α) |≤ κ′ for all x ∈ X, α ∈ P . Is spread(B) = κ? By [4, Theorem 2(i)], den(B) = cell(B) = κ. Problem 2.9. Let B = (X, P, B) be a ballean, |X| = κ and let |P | ≤ κ. Assume that κ is regular and |B(x, α)| < κ for all x ∈ X, α ∈ P . Is spread(B) = κ? By [4, Theorem 2(ii)], den(B) = cell(B) = κ. Spread of balleans 175 References [1] A. Dranishnikov, Asymptotic topology, Russian Math. Surveys 55 (2000), 1085–1129. [2] C. Chou, On the size of the set of left invariant means on a semigroup, Proc. Amer. Math. Soc. 23 (1969), 199–205. [3] I. V. Protasov, Normal ball structures, Math. Stud. 20 (2003), 3–16. [4] I. V. Protasov, Cellularity and density of balleans, Appl. Gen. Topology 8 (2007), 283– 291. [5] I. V. Protasov and M. Zarichnyi, General Asymptology, Math. Stud. Monogr. Ser., VNTL, Lviv (2006). [6] J. Roe, Lectures on Coarse Geometry, AMS University Lecture Ser., 31 (2003). Received October 2006 Accepted February 2007 Mahmoud Filali (Mahmoud.Filali@oulu.fi) Dept of Math. Sciences, University of Oulu, FIN - 90014, Oulu, Finland Igor Protasov (islab@unicyb.kiev.ua) Department of Cybernetics, Kyiv University, Volodimirska 64, Kyiv 01033, Ukraine