@ Appl. Gen. Topol. 15, no. 1 (2014), 25-32doi:10.4995/agt.2014.2049 c© AGT, UPV, 2014 Near metrizability via a new approach D. Mandal ∗,a and M. N. Mukherjee a a Department of Pure Mathematics, University of Calcutta, 35, Ballygunge Circular Road, Kolkata–700019, India (dmandal.cu@gmail.com, mukherjeemn@yahoo.co.in) Abstract The present article deals with near metrizability, initiated in an ear- lier paper [7], with a new orientation and approach. The notions of nearly regular and uniform pseudo-bases are introduced and analogues of some results concerning metrizability and paracompactness are ob- tained for near metrizability and near paracompactness respectively via the proposed approach, suitably formulated. 2010 MSC: 54D20; 54E99. Keywords: nearly paracompact space; regular open set; nearly regular and uniform pseudo-bases; nearly metrizable. 1. Introduction The idea of near paracompactness, a well known weaker form of paracom- pactness, was initiated by Singal and Arya [9], followed by an extensive study of the concept by many topologists from different perspectives and with different applications (for instance see [3], [4], [5], [6], [8]). Now, in [7] we introduced a neighbouring form of metrizability, termed near metrizability, which plays the same role with regard to near paracompactness as is done by metrizability vis- a-vis paracompactness. It was shown in [7] that there exist nearly metrizable, non-metrizable spaces that are not paracompact, moreover some other facts were established in [7]. ∗The author is thankful to the University Grants Commission, New Delhi- 110002, In- dia for sponsoring this work under Minor Research Project vide letter no. F. No. 41- 1388/2012(SR). Received November 2012 – Accepted July 2013 http://dx.doi.org/10.4995/agt.2014.2049 D. Mandal and M. N. Mukherjee The intent of the present article is to do a further study of nearly mertizable spaces from an altogether new approach. The notion of pseudo-base was intro- duced and studied in [7], and here, we define regular and uniform pseudo-bases, and ultimately achieve analogues of two well known results on metrizability in our setting. At the outset we recall a few definitions which may be found in [1, 2]. A base B for a topological space X is called regular if for each x ∈ X and any neighbourhood U of x, there exists a neighbourhood O of x such that the set of all members of B that meet both O and X \ U, is finite; and a base B is called a uniform base if for each x ∈ X and every neighbourhood U of x, the set of all members of B that contain x and meet X \ U, is finite. It is clear that every regular base is a uniform base. The next two metrization theorems are known (see [1, 2]), which have been formulated in terms of the above special base. Theorem 1.1. (a). A T3-paracompact space X with a uniform base B is metrizable. (b). Every T1-space X with a regular base B is metrizable. As already proposed, our principal aim in this paper is to achieve analogous versions of the results in Theorem 1.1 for near metrizability with accessories formulated suitably. In what follows, by a space X we shall mean a topological space X endowed with a topology τ(say). The notations ‘clA’, ‘intA’ and ‘|A|’ will respectively stand for the closure, interior and cardinality of a set A of a space X. A set A(⊆ X) is called regular open if A = intclA, and the complement of a regular open set is called regular closed. The set of all regular open (resp. closed) sets of a space X will be denoted by RO(X)(resp. RC(X)). We shall sometimes write A∗ for intclA for a subset A of X and C# = {A∗ : A ∈ C}, for any open cover C of a space X. Singal and Arya formulated the following definitions which are quite well known by now. Definition 1.2 ([10]). A topological space X is called nearly paracompact if every regular open cover of X has a locally finite open refinement. Definition 1.3 ([9]). A topological space X is said to be almost regular, if for any regular closed set A and any x ∈ X \ A, there exist disjoint open sets U and V in X such that x ∈ U and A ⊆ V . 2. Main results We start by recalling a few definitions from [7] as follows: Definition 2.1. If X and Y are two topological spaces, then a continuous, injective map f : X → Y is called a pseudo-embedding of X into Y , if for any A ∈ RO(X), f(A) is open. If there is a pseudo-embedding f of X into Y , then we say that X is pseudo- embeddable in Y . If a pseudo-embedding f : X → Y is surjective, we say that f is a pseudo-embedding of X onto Y . c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 26 Near metrizability It is known [7] that every embedding is a pseudo-embedding; but the converse is false. Definition 2.2 ([7]). A space X is called nearly metrizable if it is pseudo- embeddable in a metric space Y . Definition 2.3 ([7]). Suppose B is a family of open subsets of X. We say that B is a pseudo-base in X if for any A ∈ RO(X), there is a subfamily B0 of B such that A = ⋃ {B : B ∈ B0}. We now define a family B of open subsets of X to be a pseudo-base at a point x ∈ X if for each U ∈ RO(X) containing x, there exists a B ∈ B such that x ∈ B ⊆ U. Clearly, a family B of open subsets of a space X is pseudo-base for X if and only if it is so at each x ∈ X. We shall call a pseudo-base B σ-locally finite if B can be expressed as B = ∞⋃ n=1 Bn, where Bn is locally finite, for each n ∈ N. We now define another type of bases as follows: Definition 2.4. Let (X, τ) be a topological space. (a) A family B of subsets of X is called nearly regular if for each U ∈ B and any point x ∈ U, there exists a regular open set Ox containing x such that the set {V ∈ B : V ⋂ Ox 6= φ and V ⋂ (X \ U) 6= φ} is finite. (b) A pseudo-base B for X is called nearly regular if for each x ∈ X and any regular open set Ox containing x, there exists a regular open set Gx containing x such that the set {U ∈ B : U ⋂ Gx 6= φ and U ⋂ (X \ Ox) 6= φ} is finite. Remark 2.5. It is clear from the above definition that a subfamily of a nearly regular family is a nearly regular family. Proposition 2.6. If B is a nearly regular pseudo-base for a space X, then so is B# = {B∗ : B ∈ B}. Proof. First let x ∈ X and U a regular open set in X such that x ∈ U. As B is a pseudo-base for X, there exists B ∈ B such that x ∈ B ⊆ U. Then x ∈ B∗ ⊆ U∗ = U, and hence B# is a pseudo-base for X. Next, let x ∈ X and Ox be any regular open set in X containing x. As B is a nearly regular pseudo-base, there exists a regular open set Gx containing x such that the set {B ∈ B : B ⋂ Gx 6= φ 6= B ⋂ (X \ Ox)} is finite. It suffices to show that {B∗ ∈ B# : B∗ ⋂ Gx 6= φ 6= B ∗ ⋂ (X \ Ox)} is finite, for which we need only to show that {B∗ ∈ B# : B∗ ⋂ Gx 6= φ 6= B ∗ ⋂ (X \ Ox)} ⊆ {B ∈ B : B ⋂ Gx 6= φ 6= B ⋂ (X \ Ox)}. In fact, B ⋂ Gx = φ ⇔ intclB ⋂ intclGx = φ ⇔ B ∗ ⋂ Gx = φ, and B ⋂ (X \ Ox) = φ ⇒ B ⊆ Ox ⇒ B∗ ⊆ intclOx = Ox ⇒ B ∗ ⋂ (X \ Ox) = φ. � We shall call a space X to be an almost T3-space if it is almost regular and Hausdorff. c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 27 D. Mandal and M. N. Mukherjee Theorem 2.7. A T2-space X, possessing a nearly regular pseudo-base B is an almost T3-space. Proof. Let F be a regular closed set and x ∈ X \F . Then there exists a regular open set Ox containing x such that Ox ⋂ F = φ, i.e., F ⊆ X \ Ox. By hypothesis, there exists a regular open set Gx containing x such that the family U = {U ∈ B : U ⋂ Gx 6= φ and U ⋂ (X \ Ox) 6= φ} is finite. Put O = Ox ⋂ Gx. Then O is a regular open set containing x such that O ⋂ F = φ. Consider the family C = {U ∈ B : U ⋂ O 6= φ and U ⋂ F 6= φ}. Since F ⊆ X \ Ox, C is finite. Now for each U ∈ C, |U| ≥ 2 as O ⋂ F = φ. Let B′ = B \ C. We show that B′ is a pseudo-base for X. In fact, let p ∈ X and W a regular open set containing p. Let us enumerate C as {W1, W2, ..., Wn} and let x1, x2,..., xn be points from W1, W2,..., Wn respectively different from p. Since X is T2, each {xi} is regular closed and so X \{x1, x2, ..., xn} is a regular open set containing p and hence there exists a B1 ∈ B such that p ∈ B1 ⊆ X \ {x1, x2, ..., xn}. Again there exists B2 ∈ B such that p ∈ B2 ⊆ W . Thus there exists B3 ∈ B such that p ∈ B3 ⊆ B1 ⋂ B2 ⊆ W i.e., p ∈ B3 ⊆ W where B3 6∈ C. This shows that B ′ is a pseudo-base for X. Put G = {U ∈ B′ : U ⋂ F 6= φ} and G = ⋃ {U : U ∈ G}. Then F ⊆ G and G ⋂ O = φ with x ∈ O (since for U ∈ G, if U ⋂ O 6= φ then U ∈ C, a contradiction). This shows that F and x are strongly separated. Thus X is almost regular and consequently X is an almost T3-space. � Definition 2.8 ([2]). Let X be a topological space and B a family of subsets of X. An element U of B is called a maximal element of B if it is not contained in any element of B other than U. We denote by m(B), the set of all maximal elements of B and call m(B) the surface of B. Theorem 2.9. Let B be a nearly regular family which is a cover of X. Then the surface m(B) of B is a cover of X and is locally finite. Proof. Let x ∈ X be taken arbitrarily and kept fixed, and let U ∈ B such that x ∈ U. If U 6∈ m(B), then the family λU = {V ∈ B : V % U} is finite. In fact, by definition of B, there exists a regular open set Ox containing x such that the collection D = {V ∈ B : V ⋂ Ox 6= φ and V ⋂ (X \ U) 6= φ} is finite. Clearly, λU ⊆ D and therefore λU is finite (note that x ∈ V ⋂ Ox). Consequently λU has a maximal element V ′(say). Again x ∈ V ′ and V ′ ∈ m(B). Hence m(B) is a cover of X. We now show that m(B) is locally finite. As m(B) ⊆ B and B is nearly regular, m(B) is nearly regular. Again every element of m(B) is maximal in m(B) (because it is maximal in B and m(B) ⊆ B). Let x ∈ X. Then there exists a U ∈ m(B) such that x ∈ U. Since m(B) is nearly regular, there exists a regular open set Ox containing x such that the family B ′ = {V ∈ m(B) : V ⋂ Ox 6= φ and V ⋂ (X \ U) 6= φ} is finite. But V \ U 6= φ for all V ∈ m(B) with V 6= U c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 28 Near metrizability (because every element V in m(B) is maximal, there is no set L ∈ m(B) which properly contains V ). Thus {V ∈ m(B) : V ⋂ Ox 6= φ} = B ′ ⋃ {U} is a finite set and hence m(B) is locally finite. � Theorem 2.10. A space possessing a nearly regular pseudo-base B is nearly paracompact. Proof. Let G be any regular open cover of X and let GB = {U ∈ B : ∃G ∈ G with U ⊆ G}. We check that GB is a pseudo-base for X. In fact, let x ∈ X and G be any regular open set containing x. Now G being a cover, there exists G1 ∈ G such that x ∈ G1. Thus G ⋂ G1 is a regular open set containing x. Since B is a pseudo-base for X, there exists U ∈ B such that x ∈ U ⊆ G ⋂ G1 ⊆ G1 ∈ G ⇒ U ∈ GB with x ∈ U ⊆ G ⇒ GB is a pseudo-base for X. Since B is nearly regular and GB ⊆ B, GB is nearly regular. Thus by Theorem 2.9, m(GB) is an open cover of X and locally finite. Also clearly m(GB) is an open refinement of G. Hence X is nearly paracompact. � Analogous to the concept of uniform base, we now define a special type of base as follows: Definition 2.11. A pseudo-base B for a space X is called a uniform pseudo- base if for each x ∈ X and each regular open set Ox containing x, UOx = {U ∈ B : x ∈ U and U ⋂ (X \ Ox) 6= φ} is finite. Lemma 2.12. Let B be a family of open sets of a space X such that B# is a uniform pseudo-base for X. Then the surface m(B#) is a point finite regular open cover of X. Proof. Let x ∈ X. Then there exists U∗ ∈ B# (where U ∈ B) such that x ∈ U∗. If U∗ 6∈ m(B#) then the set λU∗ = {V ∈ B # : V ⊇ U∗} is finite. In fact, U∗ is a regular open set containing x and hence the family V = {V ∈ B# : x ∈ V and V ⋂ (X \U#) 6= φ} is finite and λU∗ ⊆ V ⋃ {U∗}. Then λU∗ has a maximal element m(λU# ) which is also a maximal element of B # and which also contains x. Hence m(B#) is a regular open cover of X. We now show that m(B#) is point finite. If possible let x ∈ X be such that x belongs to an infinite collection D of members of m(B#). Then we claim that D is a pseudo-base for X at x. If D is not a pseudo-base for X at x, there exists a regular open set W containing x such that x ∈ D ⊆ W holds for no D ∈ D, i.e., for all D ∈ D, D ⋂ (X\W) 6= φ. But {B ∈ D : B ⋂ (X \W) 6= φ} is finite as B# is a uniform pseudo-base. Hence D is a pseudo-base for X at x. Next let, U and V be two distinct (and hence non comparable) elements of D. Since x ∈ U ⋂ V and U ⋂ V is a regular open set, there exists a W ∈ D such that x ∈ W $ U ⋂ V (note that U ⋂ V 6∈ D, since otherwise U ⋂ V $ U would contradict the maximality of U ⋂ V ), i.e., x ∈ W $ U and hence W is not a c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 29 D. Mandal and M. N. Mukherjee maximal element of D although D ⊆ m(B#), a contradiction. Hence m(B#) is a point finite regular open cover of X. � Lemma 2.13. Let B be a family of open sets of a T2-space X such that B # is a uniform pseudo-base. Then there exists a countable family of point finite regular open covers which taken together is a pseudo-base for X. Proof. Let B # 1 = B # and B # 2 = B # 1 \ m ∗(B # 1 ), where m ∗(B # 1 ) is the collection of all maximal elements of B # 1 each of which contains at least two points. We first show that B # 2 is a pseudo-base for X. In fact, let x ∈ X. Then by Lemma 2.12, x belongs to only finitely many members U1, U2,..., Un (say) of m ∗(B # 1 ). Let xi ∈ Ui with x 6= xi for i = 1, 2, ..., n. Since X is T2, X \ {x1, x2, ..., xn} is a regular open set containing x and so there exists B in B# such that x ∈ B ⊆ X \ {x1, x2, ..., xn}. Let W be any regular open set containing x. Then there exists a B′ ∈ B# such that x ∈ B′ ⊆ W . Again there exists B1 ∈ B # such that x ∈ B1 ⊆ B ⋂ B′ ⇒ x ∈ B1 ⊆ W and B1 6∈ m ∗(B # 1 ) [B1 ∈ m ∗(B # 1 ) ⇒ B1 = Ui for some i = 1, 2, ..., n ⇒ xi ∈ B1 but (xi 6∈ B) ⇒ B1 6⊆ B, a contradiction]. Therefore, x ∈ B1 ⊆ W and B1 ∈ B # 2 . Again B # 2 ⊆ B # 1 and B # 1 is a uniform pseudo-base ⇒ B # 2 is a uniform pseudo-base. Now proceed by induction, if B # k is already defined then put B # k+1 = B # k \ m∗(B # k ) and as above, B # k+1 is a uniform pseudo-base for X. Then for each n ∈ N, B#n is a uniform pseudo-base for X and so m(B # n ) is a point finite regular open cover of X (by Lemma 2.12). Consider an arbitrary x ∈ X. For each n ∈ N, choose Un ∈ m(B#n ) such that x ∈ Un. If there is n ∈ N satisfying |Un| = 1 then {Un : n ∈ N} is a pseudo-base at x. If |Un| ≥ 2 for all n ∈ N then by definition of B#n , Un 6= Um for n 6= m. Hence L = {Un : n ∈ N} is an infinite set of elements of the uniform pseudo-base B#n , each containing x. We claim that L is a pseudo-base for X at x. If not, then for some regular open set D containing x, there does not exist any C ∈ L such that x ∈ C ⊆ D holds, i.e., for all C ∈ L, C ⋂ (X \ D) 6= φ. But since L ⊆ B#, {V ∈ B# : x ∈ U and U ⋂ (X \ D) 6= φ} is finite, a contradiction. Consequently, L is a pseudo-base for X at x. Hence {m(B#n ) : n ∈ N} is the required family. � Definition 2.14 ([11]). Let A be a family of subsets of a space X. The star of a point x ∈ X in A, denoted by St(x, A), is defined by the union of all members of A which contain x. A family A of subsets of a space X is said to be a star refinement of another family B of subsets of X if the family of all stars of points of X in A forms a covering of X which refines B. Theorem 2.15 ([10]). An almost regular space X is nearly paracompact if and only if every regular open covering of X has a regular open star refinement. Definition 2.16. Let X be a topological space and Γ a family of covers of X. We call Γ refined if for any point x ∈ X and any regular open set Ox containing c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 30 Near metrizability x, there exists B ∈ Γ such that St(x, B) ⊆ Ox. If all the members of Γ are regular open covers, then we say that Γ is a refined family of regular open covers. Theorem 2.17. Let B be a family of open sets of an almost T3 nearly para- compact space X such that B# is a uniform pseudo-base for X. Then X has a countable refined family of regular open covers. Proof. By Lemma 2.13, there exists a countable family of point finite regular open covers Bn, which taken together is a pseudo-base for X. Since X is almost regular and nearly paracompact, by Theorem 2.15, each Bn has a regular open star refinement Un. Now fix x ∈ X, and for each n ∈ N, choose Bn ∈ Bn so that St(x, Un) ⊆ Bn. Then {Bn : n = 1, 2, ...} is a pseudo-base for X at x. Let U be a regular open set containing x. Then there exists Bk(say) such that x ∈ Bk ⊆ U and then x ∈ St(x, Uk) ⊆ Bk ⊆ U. Thus {Un : n = 1, 2, ...} is a countable refined family of regular open covers. � Theorem 2.18 ([7]). A space X is nearly metrizable if and only if it is almost T3 and possesses a σ-locally finite pseudo-base. Theorem 2.19. Let X be an almost T3 nearly paracompact space such that X has a countable refined family {Ui} ∞ i=1 of regular open covers. Then X is nearly metrizable. Proof. Since X is nearly paracompact, each Ui has a locally finite open refine- ment Bi. Let B = ∞⋃ i=1 Bi. We show that B is a pseudo-base for X. In fact, let x ∈ X and U be any regular open set containing x. Then since {Ui} ∞ i=1 is a refined family of covers there exists k ∈ N such that x ∈ St(x, Uk) ⊆ U. But Bk being a cover of X, there exists Bk ∈ Bk such that x ∈ Bk and Bk is contained in some member of Uk containing x and hence is contained in St(x, Uk). Thus x ∈ Bk ⊆ U. Hence B is a σ-locally finite pseudo-base for X and hence by Theorem 2.18, X is nearly metrizable. � Theorem 2.20. Let B be a family of open sets of an almost T3 nearly para- compact space X such that B# is a uniform pseudo-base for X. Then X is nearly metrizable. Proof. Follows from Theorems 2.17 and 2.19. � Theorem 2.21. Every almost T3-space X with a nearly regular pseudo-base B is nearly mertizable. Proof. By Theorem 2.10, X is nearly paracompact. Again by Proposition 2.6, B# is a nearly regular pseudo-base. Since every nearly regular pseudo-base is a uniform pseudo-base, B# is a uniform pseudo-base for X, and then by Theorem 2.20, it follows that X is nearly metrizable. � c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 31 D. Mandal and M. N. Mukherjee Acknowledgements. The authors are grateful to the referee for some sug- gestions towards certain improvement of the paper. References [1] A. V. Arhangel’skii and V. I. Ponomarev, Fundamentals of general topology: Problems and exercises, Hindustan publishing corporation(India), 1984. [2] R. Engelking, General Topology, Sigma series in Pure Mathematics, Berlin, Heldermann, 1989. [3] N. Ergun, A note on nearly paracompactness, Yokahama Math. Jour. 31 (1983), 21–25. [4] I. Kovačević, Almost regularity as a relaxation of nearly paracompactness, Glasnik Mat. 13 (33)(1978), 339–341. [5] I. Kovačević, On nearly paracomapct spaces, Publications De L’institut Mathematique 25 (1979), 63–69. [6] M. N. Mukherjee and D. Mandal, On some new characterizations of near paracompact- ness and associated results, Mat. Vesnik 65, no. 3 (2013), 334–345. [7] M. N. Mukherjee and D. Mandal, Concerning nearly metrizable spaces, Applied General Topology 14, no. 2 (2013), 135–145. [8] T. Noiri, A note on nearly paracompact spaces, Mat. Vesnik 5 (18)(33)(1981), 103–108. [9] M. K. Singal and S. P. Arya, On almost regular spaces, Glasnik Mat. 4 (24)(1969), 89–99. [10] M. K. Singal and S. P. Arya, On nearly paracompact spaces, Mat. Vesnik 6 (21)(1969), 3–16. [11] J. W. Tukey, Convergence and uniformity in topology, Princeton University Press, Princeton, N. J. 1940. ix+90 pp. Transl. (2), 78 (1968), 103–118. c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 1 32