() @ Applied General Topology c© Universidad Politécnica de Valencia Volume 12, no. 1, 2011 pp. 17-25 p-Compact, p-Bounded and p-Complete Rigoberto Vera Mendoza ∗ Abstract In this paper the nonstandard theory of uniform topological spaces is applied with two main objectives: (1) to give a nonstandard treatment of Bernstein’s concept of p-compactness with additional results, (2) to introduce three new concepts (p,q)-compactness, p-totally boundedness and p-completeness. I prove some facts about them and how these three concepts are related with p-compactness. I also give a partial answer to the open question stated in [3] 2010 MSC: Primary 46A04, 46A20; Secondary 46B25. Keywords: Monad, p-compact, p-totally bounded and p-complete space 1. Introduction It was in 1966 that Abraham Robinson’s book [4] on nonstandard analysis appeared. His methods were based on the theory of models and in particu- lar on the Lowenheim-Skolem theorem. He introduced extension ∗X (in a superstructure no standard) of any set X (in a standard superstructure) by looking at nonstandard models of their respectives theories. ”Infinitely close” or ”infinitely large” were to be found in the enlargement ∗X . In this way he was able to justify proofs using ”infinitesimals” or ”monads” and that was not possible before his discovery, more over, he showed that these methods were able to produce original solutions to unsolved mathematical questions as well. This approach to the nonstandard analysis is based on the axiomatic set theory called ZFC (Zermelo-Fraenkel with Axiom Choice), all theorems of con- ventional mathematics remain valid. We start with a superstructure S , the ∗Proyecto 9.4 Coordinación de la Investigación Cient́ıfica. 18 R. Vera-Mendoza sets A, B, X , etc. are set of individuals in S. The extensions or enlargements ∗A, ∗B, ∗X , etc. are sets in the superstructure Ŝ. Nonstandard analysis is a technique rather than a subject. Aside from the- orems that tell us that nonstandard notions are equivalent to corresponding standard notions, all the results we obtain can be proved by standard meth- ods. Therefore, the subject can only be claimed to be of importance insofar as it leads to simpler, more accesible expositions, or, more important, to mathe- matical discoveries. In writing formulas we use conventional symbols such that ∈, =, ∀, ∃, ⇒, ∨, ∧, ∼, () etc. 2. Some Notions of Nonstandard Theory The nonstandard enlargements satisfies the following properties: A ⊂ ∗A , ∗(A ∩ B) = ∗A ∩ ∗B , ∗(A ∪ B) = ∗A ∪ ∗B , A ⊂ B ⇒ ∗A ⊂ ∗B , ∗(A × B) = ∗A × ∗B , ∗(A\ B) = ∗A\ ∗B , A = ∗A ⇔ A is a finite set. Therefore, ∗N \ N, ∗R \ R are non empty sets whose elements are called ”infinite large numbers”. In general, for any infinite set X, ∗X \ X 6=φ Each function f : A → B has an ”extension” ∗f : ∗A → ∗B . For this we observe that f ⊂ A × B , so that, ∗f ⊂ ∗A × ∗B Given a topological space (X, t) and z ∈ ∗X , the monad of z is µ(z) = ⋂ {∗O | O ∈ t and z ∈ ∗O} The space (X, t) is Hausdorff ⇔ ∀ x, y ∈ X , µ(x) ∩ µ(y) = φ We denote by X̂ = ⋃ {µ(x) | x ∈ X} Let L denote a language in the standard superstructure and by L̂ we denote de corresponding language in the nonstandard superstructure. A formula of L in which no variable has a free occurrence is called a sentence of L. For each sentence α ∈ L we have correspondent sentence ∗α ∈ L̂ . α = ∗α ⇔ α contains only constants b ∈ S Let α be a sentence of L, we will denote |= α if ” α is true in S”; we will denote ∗ |= ∗α if ” ∗α is true in Ŝ” Transfer Principle (TP): The admisible proposition α ∈ (Ŝ, £) is true ⇔ ∗α ∈ ( ∗Ŝ, ∗£) is true. |= α ⇔ ∗ |= ∗α For instance, A ∩ B 6= φ ⇔ ∗A ∩ ∗B 6= φ , A ⊂ B ⇔ ∗A ⊂ ∗B , etc. The TP provides one of the basic tools of nonstandard analysis. A mathe- matical theorem that is equivalent to |= α for some sentence α ∈ L can be proved by showing instead that ∗ |= ∗α [1]. A relation R ⊂ A × B is called concurrent if given a1, a2, ..., an ∈ dom(R) there exists b ∈ B such that (ai, b) ∈ R for all i = 1, 2, ..., n p-Compact, p-Bounded and p-Complete 19 Concurrence Theorem (CT): If R is a concurrent relation then there exists b ∈ ∗B such that ( ∗a, b) ∈ ∗R for all a ∈ A We only will consider Hausdorff uniform topological spaces (X, t) , with uniformity U . Then, on ∗X we will consider three non-Hausdorff topologies: a) τU will be the topology generated by the sets ∗O such that O ∈ t. b) τ̂U will be the uniform topology on ∗X given by the uniformity Û = { ∗V | V ∈ U}. c) τi the topology generated by { ∗A | A ⊂ X}. A basis of τU is { ∗V (x) | x ∈ X, V ∈ U}. A basis of τ̂U is { ∗V (z) | z ∈ ∗X, V ∈ U}. For any z ∈ ∗X , µ(z), o(z) and i(z) will denote the monads of z for the topologies a), b) and c) respectively. τU and τ̂U agree on X, thereby µ(x) = o(x) ∀ x ∈ X , in general τU < τ̂U . On the other hand, i(x) = {x} ∀ x ∈ X and τU < τi . So that, i(z) ⊂ µ(z) and o(z) ⊂ µ(z) . The bigger the topology the smaller the monad. If f : (X, t) → (Y, t′) is a continuous function then ∗f : ( ∗X, τU ) → ( ∗Y, τ ′U ) is a continuous function. Any function ∗f : ( ∗X, τi) → ( ∗Y, τ ′i ) is continuous. The topology on the set of natural numbers N always be the discrete topo- logy. For any p ∈ ∗N \ N and any topological space X β′pX = ⋃ {µ(∗f (p)) | f : N → X}. 3. p-Compactness Definition 3.1. Let p ∈ ∗N \ N be and {zn | n ∈ N} ⊂ ∗X . We will say that z ∈ ∗X is a p-limit of the sequence {zn} ( z = p − lim zn ) if z is an accumulation point of the sequence such that for each O ∈ t , with z ∈ ∗O , p ∈ ∗{n ∈ N | zn ∈ ∗O}. Lemma 3.2. If f : X → Y is a continuous function and z ∈ ∗X is a p-lim of {zn} ⊂ ∗X then ∗f (z) = p − lim ∗f (zn) in ∗Y. Proof. Let W ⊂ Y be an open set such that ∗f (z) ∈ ∗W and let O ⊂ X be open such that z ∈ ∗O and ∗f (∗O) ⊂ ∗W . Hence, p ∈ ∗{n ∈ N | zn ∈ ∗O} ⊂ {n ∈ N | ∗f (zn) ∈ ∗W }. � 20 R. Vera-Mendoza Theorem 3.3. z = p − lim xn ⇔ xp ∈ µ(z). Proof. ⇒ ) Let O ∈ t such that z ∈ ∗O , by definition xp ∈ ∗O. ⇐ ) Let O ∈ t be such that z ∈ ∗O . The following sentence is true in the nonstandard structure (∃ n ∈ ∗N )(xn ∈ ∗O) , by the PT (∃ n ∈ N )(xn ∈ O) is true, i.e., A = {n ∈ N | xn ∈ O} 6= ∅ y, p ∈ ∗A. � Proposition 3.4. Let p, q ∈ ∗N \ N . If there exists f : N → N such that p = q − lim{f (n)} then ∗f (q) ∈ β′pN. Proof. p ∈ ∗O ⇒ ∗f (q) ∈ ∗O ⇒ µ(∗f (q)) ⊂ µ(p) . � Proposition 3.5. If for {xn} ⊂ X there exists {xnk } → x ∈ X such that p ∈ ∗{nk} , then x = p − lim xn. Definition 3.6 ([3]). A topological space (X, t) is p-compact if every sequence in X has a p-limit in ∗X X̂ = ⋃ {µ(x) | x ∈ X}. X ⊂ X̂ ⊂ β′pX taking for each x ∈ X f : N → X, f (n) = x Theorem 3.7. (X, t) is p-compact ⇔ X̂ = β′pX. Proof. ⇒ ) The contention left to proof is X̂ ⊃ β′pX Let {xn} ⊂ X be such that q ∈ µ(xp) . Since xp ∈ µ(x) for some x ∈ X , this implies that q ∈ µ(xp) ⊂ µ(x) ⇐ ) Nothing left to prove. � Corollary 3.8. (X, t) compact ⇒ p-compact for all p ∈ ∗N \ N. Proof. X̂ ⊂ β′pX ⊂ ∗X = X̂. � Definition 3.9. Let tp be the topology generated by the set {O ∈ t | ∗f (p)6∈ ∗O for some f : N → X} tp is closed under intersections. Lemma 3.10. X tp-compact implies p-compact. Proposition 3.11. (X, t) p-compact implies for each numerable tp-cover of X has a finite sub-cover. Corollary 3.12. If X is p-compact and tp-Lindelof then it is tp-compact. Definition 3.13 (Comfort Order). p ≤C q (in ∗N \ N )if (X, t) q-compact implies p-compact [2]. Corollary 3.14. If β′pX ⊂ β ′ qX then X q-compact implies p-compact. p-Compact, p-Bounded and p-Complete 21 4. (p, q)-Compactness Definition 4.1. p ⊗ q = {A ⊂ N × N | (p, q) ∈ ∗A}. Definition 4.2. A topological space (X, t) is (p, q)-compact if for each f : N × N → X there exists x ∈ X such that for every x ∈ O ∈ t , ∗ f (p, q) ∈ µ(x) y (p, q) ∈ ∗{(m, n) | f (m, n) ∈ O}. Theorem 4.3. (X, t) is (p, q)-compact if and only if it is p and q compact. Proof. ⇒ ) Let us prove that X is p-compact. Let f : N → X and π1 : N × N → N be the projection in the first coordinate, then f ◦ π1 : N × N → X : By hypothesis there exists x ∈ X such that ∗f (p) = (∗f ◦ ∗π1)(p, q) = ∗(f ◦ π1)(p, q) ∈ µ(x) and for each x ∈ O ∈ t (p, q) ∈ ∗{(m, n) | f (m) = (f ◦ π1)(m, n) ∈ O} i.e., {(m, n) | f (m) = (f ◦ π1)(m, n) ∈ O} ∈ p ⊗ q . This implies that p = ∗π1(p, q) ∈ ∗π∗1 ({(m, n) | f (m) = (f ◦ π1)(m, n) ∈ O}) = =∗ {m ∈ N | f (m) ∈ O}. Analogous proof for (f ◦ π2) : N × N → X. Since X is q-compact, there exists y ∈ X such that ∗f (q) = ∗(f ◦ π2)(p, q) ∈ µ(y) and for each y ∈ U ∈ t , (p, q) ∈ ∗{(m, n) | (f ◦ π2)(m, n) ∈ U} , i.e., {(m, n) | f (n) = (f ◦ π2)(m, n) ∈ U} ∈ p ⊗ q . This implies that q = ∗π2(p, q) ∈ ∗π2( ∗{(m, n) | f (n) = (f ◦ π2)(m, n) ∈ U}) ∈ ∗{n ∈ N | f (n) ∈ U}. ⇐ ) Let us consider f : N × N → X . To prove that there exists x ∈ X such that ∗f (p, q) ∈ µ(x) . For each n ∈ N we define fn : N → X as fn(m) = f (m, n) . By hypothesis, there exists xn ∈ X such that ∗f (p, n) = ∗fn(p) ∈ µ(xn) and for each xn ∈ O ∈ t , p ∈ ∗{m ∈ N | fn(m) ∈ O}. Let us define gp : N → X as gp(n) = xn ∼ ∗f (p, n) . Hence ([1, page 81]) gp ∼ ∗f (p, −) ∈ ∗( ∏ N X) . The set {∗W | W is an open basic set of ∏ N X} is a basis of the topology on ∗( ∏ N X) . By the q-compacity of X , there exists y ∈ X such that ∗gp(q) ∈ µ(y) and for each y ∈ U ∈ t , q ∈ ∗A = ∗{n ∈ N | gp(n) ∈ U}. Since gp(n) ∼ ∗f (p, n) , ∗f (p, n) ∈ ∗U ∀ n ∈ A , i.e., the following formula is true (∀ n ∈ N )(n ∈ A ⇒ ∗f (p, n) ∈ ∗U ) . By the PT, the following formula is true (∀ n ∈ ∗N )(n ∈ ∗A ⇒ ∗f (p, n) ∈ ∗U ) . q ∈ ∗A ⇒ ∗f (p, q) ∈ ∗U ⇒ ∗f (p, q) ∈ ⋂ { ∗U | y ∈ U ∈ t} = µ(y). � 22 R. Vera-Mendoza 5. p-totally bounded and p-complete Definition 5.1. All uniform space (X, t) is completely regular, let us denote by {ρj}J the saturated family of pseudometrics that define the topology t given by the uniformity U . We recall that for each z ∈ ∗X o(z) = {y ∈ ∗X | ρj(y, z) ≈ 0 ∀ j ∈ J} [1, 5] µ(z) = ⋂ { ∗V (x) | V ∈ U, x ∈ X, z ∈ ∗V (x)} Every V ∈ U is V = {(x, y) ∈ X × X | ρj (x, y) < ǫ} for some ǫ > 0 and some pseudometric ρj , also denoted by V = V ǫ j . For all z ∈ ∗X , µ(z) ⊃ o(z) and for all x ∈ X , µ(x) = o(x) For (X, t) we define the set pns ∗X = {z ∈ ∗X | µ(z) = o(z)}. Proposition 5.2. z ∈ pns ∗X ⇒ µ(z) ⊂ pns ∗X. Proof. If z ∈ ∗X and w ∈ µ(z) then o(w) ⊂ µ(w) ⊂ µ(z) and o(w) = o(z) If z ∈ pns ∗X then µ(z) = o(z) = o(w) ⊂ µ(w) ⊂ µ(z) , i.e., o(w) = µ(w) thereby w ∈ pns ∗X. � Proposition 5.3. z ∈ pns ∗X ⇔ given ǫ > 0 and any pseudometric ρj there exists xj ∈ X such that ∗ρj (z, xj ) < ǫ. Proof. ⇒) Let z ∈ pns ∗X be given, then, o(z) = µ(z) . For any ǫ > 0 , a pseudometric ρj and V = V ǫ j ∈ U . Since o(z) × o(z) ⊂ V , µ(z) × µ(z) ⊂ V , PC tells us that there exists D ∈ ∗Fµ(z) such that D ⊂ µ(z) , there exists D ∈ ∗Fµ(z) such that D × D ⊂ ∗V , hence, by PT, there exists D ∈ Fµ(z) such that D × D ⊂ V . D ∈ Fµ(z) ⇒ µ(z) ⊂ ∗D ⇒ {z} × ∗D ⊂ ∗V ⇒ (z, x) ∈ ∗V for all x ∈ D ⊂ X , i.e., ρ(z, x) < ǫ for all x ∈ D ⇐) Let w ∈ µ(z) , ǫ > 0 be given and (z, xj ) ∈ ∗V ǫ 2 j = {(x, y) | ρj (x, y) < ǫ 2 } . Since w ∈ ∗V ǫ 2 j (xj ) and ∗ρj (w, z) ≤ ∗ρj (w, xj ) + ∗ρj (xj , z) < ǫ 2 + ǫ 2 = ǫ . This implies that w ∈ ∗V ǫj (z) and therefore w ∈ o(z) , i.e., µ(z) ⊂ o(z) ⊂ µ(z). � Corollary 5.4. z ∈ pns ∗X ⇔ Fµ(z) is a Cauchy filter Remark 5.5. X̂ ⊂ pns ∗X ⊂ ∗X X̂ = pns ∗X ⇔ X is complete pns ∗X = ∗X ⇔ X is totally bounded Definition 5.6. (X, t) is p-totally bounded if for each function f : N → X , ∗f (p) ∈ pns ∗X p-Compact, p-Bounded and p-Complete 23 Lemma 5.7. p-compact ⇒ p-totally bounded. Proof. X̂ ⊂ pns ∗X ⊂ ∗X. � Lemma 5.8. X is p-totally bounded ⇔ β′pX ⊂ pns ∗X. Proof. ⇒ ) X p-totally bounded ⇒ ∗f (p) ∈ pns ∗X ⇒ µ( ∗f (p)) ⊂ pns ∗X. � Corollary 5.9. X is p-totally bounded for all p ∗N \ N ⇔ ∪pβ ′ pX ⊂ pns ∗X. Lemma 5.10. (X, t) totally bounded ⇒ p-totally bounded for all p ∗N \ N. Proof. ⇒) Totally bounded implies pns ∗X = ∗X. � Lemma 5.11. If (X, t) is a complete space then X p-compact ⇔ p-totally bounded. Proof. Complete implies X̂ = pns ∗X. � ∗f : ∗N → ∗X always is a τU -continuous function. Proposition 5.12. X is p-totally bounded ⇔ every f : N → X is a contin- uous function in p with the τ̂U topology. Proof. ⇒ ) Since ∗f (p) ∈ pns ∗X , µ(∗f (p)) = o(∗f (p)) , since ∗f is a conti- nuous function with τU , ∗f (o(p)) = ∗f (µ(p)) ⊂ µ(∗f (p)) = o(∗f (p)) then it also is a continuous function in p with τ̂U (Cauchy’s Principle, Theorem 8.1.4, [5]) ⇐ ) Let f : N → X be a continuous function in p ∈ N with τU . To prove that ∗f (p) ∈ pns ∗X , that is, µ(∗f (p)) = o(∗f (p)) , for this only left to prove that µ(∗f (p)) ⊂ o(∗f (p)) We recall that Fp = {A ⊂ N | p ∈ ∗A} ⊂ P (N ) is a Cauchy filter and so is the filter G generated by the image of f (Fp) . Claim: µ(G) = µ(∗f (p)). Since ∗f (p) ∈ µ(G) , µ(G) ⊂ µ(∗f (p)). On the other hand, ∗f (p) ∈ ∗S ⇒ f −1(S) ∈ Fp thereby S ⊃ f (f −1(S)) ∈ G therefore S ∈ G. Let V ∈ U and A ∈ Fp be given such that ∗f (∗A) ⊂ ∗V (p) (τ̂U -continuity of ∗f in p). This tells us that µ(G) ⊂ o( ∗f (p)) , therefore µ(∗f (p)) ⊂ o(∗f (p)). � Corollary 5.13. ∗f (p) ∈ pns ∗X ⇔ ∗f is τ̂U -continuous in p. Proposition 5.14. pns ∗X is the biggest subset of ∗X containing X such that τU and τ̂U agree. Theorem 5.15. (X, t) is totally bounded ⇔ τU = τ̂(∗)U . 24 R. Vera-Mendoza Corollary 5.16. If ∗X = ∪pβ ′ pX and X is p-totally bounded for all p ∗N \ N , then X is totally bounded. Definition 5.17. (X, t) is p-complete if any function f : N → X , ∗f (p) ∈ pns ∗X ⇒ ∗f (p) ∈ X̂. Proposition 5.18. p-compact ⇔ p-complete and p-totally bounded. Definition 5.19. p ≤ta q if q-totally bounded ⇒ p-totally bounded. p ≤cc q if q-complete ⇒ p-complete. Questions 2.20 and 2.21 In [3] appeared Q 2.20 and Q 2.21 (open questions) whose nonstandard versions are: Q 2.20: Is there p ∈∗ N \ N such that TC (p) ∩ wP = ∅ ? TC (p) = {q ∈ ∗N | q ≤C p and p ≤C q} Q 2.21: Is β′N ∩ wP 6= for all p ∈∗ N \ N ? or Is there p ∈∗ N \ N such that β′N ∩ wP = ∅ ? I will prove that if Q 2.21 is true then Q 2.20 is true. Definition 5.20. For q ∈∗ N \ N , Nq = {p ∈ ∗ N | q 6∈ β′pN}. Remark 5.21. 1.- Np ⊂ Nq ⇒ p ∈ β ′ qN. 2.- For all f : N → N , ∗f (Nq) ⊂ Nq. Definition 5.22. We will say that Nq is p-compact if each function f : N → Nq →֒ ∗N , its extension f̂ : ∗N → ∗N satisfies f̂ (p) ∈ Nq. Remark 5.23. Nq p-compact ⇒ p ∈ Nq. Definition 5.24. p ∈∗ N \ N is a weak P-point if p 6∈C ⊂∗ N \ N ⇒ p 6∈ C. We denote the set of weak P-points by wP. Theorem 5.25. q ∈ wP ⇒ Nq p-compact for all p ∈ Nq. Theorem 5.26. p ≤C q and p ∈ wP ⇒ Np ⊂ Nq ⇒ p ∈ β ′ qN. Proof. If there is some z ∈ Np\ Nq then both Np is z-compact and q ∈ β ′ z N . Hence, there is f : N → N such that q ∈ ∗A ⇔ ∗f (z) ∈ ∗A for all A ⊂ N which implies that ĝ(q) ∈ Np since ĝ( ∗f (z)) ∈ Np for all g : N → Np Therefore Np is q-compact and, because p ≤C q , Np is p-compact then (Theorem 5.25) p ∈ Np . This contradiction tells us that there is not z ∈ Np\ Nq. � p-Compact, p-Bounded and p-Complete 25 Now we can restate Q 2.21: Is there p ∈∗ N \ N such that Nq is p-compact for all q ∈ wP ? or⋂ q∈wP Nq 6= ∅ ? If this last inequality is true, that is, if there is p ∈ ⋂ q∈wP Nq then q 6≤C p ∀ q ∈ wP thereby TC (p) ∩ wP = ∅ (Q 2.21) Acknowledgements. I want to thank to Dr. Fernando Hernández for his valuated suggestions. References [1] M. Davis, Applied Nonstandard Analysis, John Wiley NY, (1977). [2] S. Garcia-Ferreira, Comfort types of ultrafilters, Proc. Amer. Math. Soc. 120, no. 4 (1994), 1251–1260. [3] S. Garcia-Ferreira, Three orderings on β(ω) \ ω, Topology Appl. 50 (1993), 199–216. [4] A. Robinson, Non-Standard Analysis, Princeton Landmarks in Math, (1996). [5] K. D. Stroyan and W. A. J. Luxemburg, Introduction to the theory of infinitesimals, Academic Press, (1976). (Received September 2009 – Accepted May 2010) R. Vera-Mendoza (rigovera@gmail.com) Facultad de Ciencias F́ısico-Matemáticas, Universidad Michoacana de San Nicolás de Hidalgo, Morelia, Michoacán, México. p-Compact, p-Bounded and p-Complete. By R. Vera-Mendoza