Abstract
This article fits in the context of the approach to topological problems in terms of the underlying convergence space structures, and serves as yet another illustration of the power of the method. More specifically, we spell out convergence-theoretic characterizations of the notions of weak base, weakly first-countable space, semi-metrizable space, and symmetrizable spaces. With the help of the already established similar characterizations of the notions of Frchet-Ursyohn, sequential, and accessibility spaces, we give a simple algebraic proof of a classical result regarding when a symmetrizable (respectively, weakly first-countable, respectively sequential) space is semi-metrizable (respectively first-countable, respectively Fréchet) that clarifies the situation for non-Hausdorff spaces. Using additionally known results on the commutation of the topologizer with product, we obtain simple algebraic proofs of various results of Y. Tanaka on the stability under product of symmetrizability and weak first-countability, and we obtain the same way a new characterization of spaces whose product with every metrizable topology is weakly first-countable, respectively symmetrizable.
This work was supported by U.S. Education Department Hispanic-Serving Institutions – Science, Technology, Engineering & Math (HSI-STEM) program grant # P031C160155, and by the Separately Budgeted Research program of NJCU for the academic year 2017–2018.
(Communicated by Ľubica Holá)
Appendix A. Convergence spaces, notation and conventions
A.1 Set-theoretic conventions
If X is a set, we denote by 2X its powerset, by [X]<∞ the set of finite subsets of X and by [X]ω the set of countable subsets of X. If 𝓐 ⊂ 2X, we write
A family 𝓕 of non-empty subsets of X is called a filter if 𝓕 = 𝓕∩ = 𝓕↑. We denote by 𝔽X the set of filters on X. Note that 2X is the only family 𝓐 satisfying 𝓐 = 𝓐↑ = 𝓐∩ that has an empty element. Thus we sometimes call {∅}↑ = 2X the degenerate filter onX. The set 𝔽X is ordered by 𝓕 ≤ 𝓖 if for every F ∈ 𝓕 there is G ∈ 𝓖 with G ⊂ F. Maximal elements of 𝔽X are called ultrafilters and 𝓕 ∈ 𝔽X is an ultrafilter if and only if 𝓕 = 𝓕#. We denote by 𝕌X the set of ultrafilters on X.
A sequence
A filter that is induced by some sequence is called sequential filter. We denote by 𝔼X the set of sequential filters on X. A filter 𝓕 ∈ 𝔽X is called Fréchet if
Of course, every countably based filter, that is, a filter with a countable filter-base, is in particular a Fréchet filter.
A.2 Convergence
A convergenceξ on a set X is a relation between 𝔽X and X, denoted
whenever (x, 𝓕) ∈ ξ (and we then say that 𝓕 converges toxforξ), satisfying the following two conditions:
for every filters 𝓕 and 𝓖 on X, and every x ∈ X.
The pair (X, ξ) is then called a convergence space. We denote by
Every topology can be seen as a convergence. Indeed, if τ is a topology and 𝓝τ(x) denotes the neighborhood filter of x for τ, then
defines a convergence that completely characterizes τ. Hence, we do not distinguish between a topology τ and the convergence it induces. A convergence is called Hausdorff if the cardinality of lim 𝓕 is at most one, for every filter 𝓕. Of course, a topology is Hausdorff in the usual topological sense if and only if it is in the convergence sense. A point x of a convergence space (X, ξ) is isolated if
Given two convergences ξ and θ on the same set X, we say that ξ is finer thanθ or that θ is coarser thanξ, in symbols ξ ≥ θ, if limξ𝓕 ⊂ limθ𝓕 for every 𝓕 ∈ 𝔽X. With this order, the set 𝓒(X) of convergences on X is a complete lattice whose greatest element is the discrete topology and least element is the antidiscrete topology, and for which, given Ξ ⊂ 𝓒(X),
A base for a convergence space (X, ξ) is a family 𝓑 of subsets of X such that for every x ∈ X and every 𝓕 with x ∈ limξ𝓕, there is a filter 𝓖 with a filter-base composed of elements of 𝓑 with x ∈ limξ𝓖 and 𝓕 ≤ 𝓖. If this property is satisfied only for a specific x, then 𝓑 is a local base atx.
A.3 Continuity, initial and final constructions
A map f between two convergence spaces (X, ξ) and (Y, σ) is continuous if for every 𝓕 ∈ 𝔽X and x ∈ X,
where
Consistently with [7], we denote by |ξ| the underlying set of a convergence ξ, and, if (X, ξ) and (Y, σ) are two convergence spaces, we often write f : |ξ| → |σ| instead of f : X → Y even though one may see it as improper since many different convergences have the same underlying set. This allows to talk about the continuity of f : |ξ| → |σ| without having to repeat for what structures.
Given a map f : |ξ| → Y, there is the finest convergence fξ on Y making f continuous (from ξ), and given f : X → |σ|, there is the coarsest convergence f−σ on X making f continuous (to σ). The convergences fξ and f−σ are called final convergence forfandξ and initial convergence forf andσ respectively. Note that
If A ⊂ |ξ|, the induced convergence byξonA, or subspace convergence, is i−ξ, where i : A → |ξ| is the inclusion map. If ξ and τ are two convergences, the product convergenceξ × τ on |ξ| × |τ| is the coarsest convergence on |ξ| × |τ| making both projections continuous, that is,
where pξ : |ξ| × |τ| → |ξ| and pτ : |ξ| × |τ| → |τ| are the projections defined by pξ(x, y) = x and pτ(x, y) = y respectively.
A.4 Topologies and pretopologies
In fact, the category Top of topological spaces and continuous maps is a reflective subcategory of Conv and the corresponding reflector T, called topologizer, associates with each convergence ξ on X its topological modification Tξ, which is the finest topology on X among those coarser than ξ (in Conv). Concretely, Tξ is the topology whose closed sets are the subsets of |ξ| that are ξ-closed, that is, subsets C satisfying
A subset O of |ξ| is ξ-open if its complement is closed, equivalently if
The neighborhood filter 𝓝ξ(x) of x for ξ is the neighborhood filter of x for the topology Tξ, that is, the filter generated by the family of ξ-open sets containing x. Because Top is a (full) reflective subcategory of Conv, Top is closed under initial constructions so that a subspace of a topological convergence space is topological and a product of topological convergence spaces is topological. However, the convergence induced by Tξ on a subset A and T(i−ξ) do not need to coincide, and similarly, the topological modification of a product generally does not coincide with the product of the topological modifications.
Given a convergence ξ and x ∈ |ξ|, the filter
is called the vicinity filter ofx for ξ. In general, 𝓥ξ(x) does not need to converge to x for ξ. If x ∈ limξ𝓥ξ(x) for all x ∈ |ξ|, we say that ξ is a pretopology or is pretopological. Note that a local base at x of a pretopology is necessarily a filter-base of 𝓥ξ(x).
The category PrTop of pretopological spaces and continuous maps is a (full) reflective subcategory of Conv. The corresponding reflector S0, called pretopologizer, associates with each convergence ξ its pretopological modification S0ξ, which is the finest among the pretopologies coarser than ξ. Explicitly, x ∈ limS0ξ𝓕 if 𝓕 ≥ 𝓥ξ(x) so that 𝓥ξ(x) = 𝓥S0ξ(x). Topologies are in particular pretopological, so that S0 ≥ T. Moreover,
and ξ is topological if and only if 𝓥ξ(x) has a filter-base composed of open sets, in which case 𝓥ξ(x) = 𝓝ξ(x). We distinguish between the inherence
of A ⊂ |ξ|} and its interior
In contrast to T, the reflector S0 does commute with initial convergence (so that the pretopological modification of an induced convergence is the convergence induced by the pretopological modification), but not with suprema, hence not with products.
A.5 Sequential, Fréchet, and strongly Fréchet convergences
With a convergence ξ, we associate its sequentially based modification Seqξ defined by
A useful immediate consequence of [4: Corollary 10] (that can be traced all the way back to [14], but [4] gives a formulation more consistent with our terminology) is:
Proposition 24
Ifσ is a Hausdorff pretopology, then
In the same vein, letting 𝔽1X denote the set of filters with a countable filter-base, we can associate to each convergence ξ its modification of countable character I1ξ defined by
Both Seq and I1 are concrete coreflectors from Conv to the full categories of sequentially based and convergences of countable character respectively.
Recall that a topological space is sequential if every sequentially closed subset (that is, subset that contains the limit point of every sequence on it) is closed. It is easily seen that a subset of a topology ξ is sequentially closed if and only if it is Seq ξ-closed, so that, ξ is sequential if and only if ξ = T Seq ξ. As σ ≤ Seq σ for every convergence σ, the inequality ξ ≤ T Seq ξ is true for every topology ξ. Hence if ξ = Tξ then
and it turns out that
and we take one or the other of these inequalities as a definition of a sequential convergence.
A topological space X is Fréchet if for every x ∈ X and A ⊂ X, if x ∈ clA then there is a sequence on A converging to x. It is easily seen (e.g., [15][2]) that if ξ is a topology then
and we take either one of these inequalities as a definition of a Fréchet convergence. Note that a pretopology is Fréchet if and only if each vicinity filter is a Fréchet filter in the sense of (A.1).
The paratopologizer S1 is defined by
and defines a concrete reflector. A fixed point of S1 is called a paratopology.
A topological space X is strongly Fréchet if whenever x ∈ adh𝓗 for a countably based filter 𝓗 on X, there is a sequential filter (xn)n with (xn)n ≥ 𝓗 and x ∈ lim(xn)n. It is easily seen (e.g., [2]) that if ξ is a topology then
and we take either one of these inequalities as a definition of a strongly Fréchet convergence.
A.6 Properties of concrete functors
Several other topological properties can be characterized with the aid of a functorial inequality; see [2, 7]. All functors considered here (in particular T, S0, S1, Seq and I1 ) are concrete endofunctors of Conv, and as such each satisfy the following properties of a modifier F acting on convergence spaces (for all ξ,θ and f):
All five functors T, S0, S1, Seq and I1 are also idempotent, that is, satisfy F(Fξ) = Fξ for all ξ. The reflectors T, S0 and S1 are additionally contractive (Fξ ≤ ξ for all ξ) and the coreflectors Seq and I1 are additionally expansive (Fξ ≥ ξ for all ξ).
A.7 Compactness and local compactness
A subset K of a convergence space is ξ-compact if limξ𝓤 ∩ K ≠ ∅ for every ultrafilter 𝓤 on K, and countably compact if every countably based filter 𝓗 with K ∈ 𝓗#, there is an ultrafilter 𝓤 ≥ 𝓗 with limξ𝓤 ∩ K ≠ ∅. Given two convergences ξ and σ on the same set, we say that ξ is locally (countably)σ-compact if every ξ-convergent filter has a (countably) σ-compact element.
References
[1] Arhangel’skii, A. V.: Mappings and spaces, Russian Math. Surveys 21 (1966), 115–162.10.1070/RM1966v021n04ABEH004169Search in Google Scholar
[2] Dolecki, S.: Convergence-theoretic methods in quotient quest, Topology Appl. 73 (1996), 1–21.10.1016/0166-8641(96)00067-3Search in Google Scholar
[3] Dolecki, S.—Greco, G. H.: Topologically maximal pretopologies, Studia Math. 77 (1984), 265–281.10.4064/sm-77-3-265-281Search in Google Scholar
[4] Dolecki, S.—Jordan, F.—Mynard, F.: Reflective classes of sequentially based convergence spaces, sequential continuity and sequence-rich filters, Topology Proc. 31(2) (2007), 457–479.Search in Google Scholar
[5] Dolecki, S.—Mynard, F.: Productively sequential spaces, Math. Slovaca 68(3) (2018), 667–676.10.1515/ms-2017-0133Search in Google Scholar
[6] Dolecki, S.—Mynard, F.: Convergence-theoretic mechanisms behind product theorems, Topology Appl. 104 (2000), 67–99.10.1016/S0166-8641(99)00012-7Search in Google Scholar
[7] Dolecki, S.—Mynard, F.: Convergence Foundations of Topology, World Scientific, 2016.10.1142/9012Search in Google Scholar
[8] Dolecki, S.—Pillot, M.: Topologically maximal convergences, accessibility, and covering maps, Math. Bohem. 123(4) (1998), 371–384.10.21136/MB.1998.125968Search in Google Scholar
[9] Gruenhage, G.: Generalized metric spaces. Handbook of Set-Theoretic Topology (K. Kunen, J. E. Vaughan eds.), Elsevier, 1984, pp. 423–501.10.1016/B978-0-444-86580-9.50013-6Search in Google Scholar
[10] Harley III, P. W.—Stephenson Jr, R. M: Symmetrizable and related spaces, Trans. Amer. Math. Soc. 219 (1976), 89–111.10.1090/S0002-9947-1976-0418048-0Search in Google Scholar
[11] Heath, R.: Arc-wise connectedness in semi-metric spaces, Pacific J. Math. 12(4) (1962), 1301–1319.10.2140/pjm.1962.12.1301Search in Google Scholar
[12] Hong, W. C.: Notes on Fréchet spaces, Int. J. Math. Math. Sci. 22(3) (1999), 659–665.10.1155/S0161171299226592Search in Google Scholar
[13] Hong, W. C.: A note on weakly first countable spaces, Commun. Korean Math. Soc. 17(3) (2002), 531–534.10.4134/CKMS.2002.17.3.531Search in Google Scholar
[14] Kantorovich, L. V.—Vulich, B. Z.—Pinsker, A. G.: Functional Analysis in Vector Order Spaces, GITTL, 1950.Search in Google Scholar
[15] Kent, D. C.—Richardson, G. D.: The decomposition series of a convergence space, Czechoslovak Math. J. 23(98) (1973), 437–446.10.21136/CMJ.1973.101185Search in Google Scholar
[16] Michael, E.: A quintuple quotient quest, Gen. Topology Appl. 2 (1972), 91–138.10.1016/0016-660X(72)90040-2Search in Google Scholar
[17] Mynard, F.: Strongly sequential spaces, Comment. Math. Univ. Carolin. 41 (2000), 143–153.Search in Google Scholar
[18] Mynard, F.: Coreflectively modified continuous duality appliedf to classical product theorems, Appl. Gen. Topol. 2(2) (2002), 119–154.10.4995/agt.2001.2145Search in Google Scholar
[19] Mynard, F.: More on strongly sequential spaces, Comment. Math. Univ. Carolin. 43(3) (2002), 525–530.Search in Google Scholar
[20] Mynard, F.: Coreflectively modified duality, Rocky Mountain J. Math. 34(2) (2004), 733–758.10.1216/rmjm/1181069877Search in Google Scholar
[21] Nedev, S. Ĭ.: o-metrizable spaces, Trudy Moskov. Mat. Obšč. 24 (1971), 201–236.Search in Google Scholar
[22] Siwiec, F.: On defining a space by a weak base, Pacific J. Math. 52 (1974), 233–245.10.2140/pjm.1974.52.233Search in Google Scholar
[23] Tanaka, Y.: On symmetric spaces, Proc. Japan Acad. 49 (1973), 106–111.10.3792/pja/1195519432Search in Google Scholar
[24] Tanaka, Y.: Note on products of symmetric spaces, Proc. Japan Acad. 50 (1974), 152–154.10.3792/pja/1195519073Search in Google Scholar
[25] Whyburn, G. T.: Accessibility spaces, Proc. Amer. Math. Soc. 24 (1970), 181–185.10.1090/S0002-9939-1970-0248722-0Search in Google Scholar
© 2019 Mathematical Institute Slovak Academy of Sciences
Articles in the same Issue
- Prof. RNDr. pavel brunovský, DrSc. passed away ∗dec. 5, 1934 – † dec. 14, 2018
- Ideals and congruences in pseudo-BCH algebras
- Regular double p-algebras
- Distributive nearlattices with a necessity modal operator
- States in generalized probabilistic models: An approach based in algebraic geometry
- Free power-associative n-ary groupoids
- On the 2-class field tower of subfields of some cyclotomic ℤ2-extensions
- On the space of generalized theta-series for certain quadratic forms in any number of variables
- Uniqueness and periodicity for meromorphic functions with partial sharing values
- Mild solution of stochastic partial differential equation with nonlocal conditions and noncompact semigroups
- An inverse boundary problem for fourth-order Schrödinger equations with partial data
- Entropy as an integral operator
- Global behavior of two third order rational difference equations with quadratic terms
- Measures on effect algebras
- Some topological and combinatorial properties preserved by inverse limits
- The convergence-theoretic approach to weakly first countable spaces and symmetrizable spaces
- Compositions of porouscontinuous functions
- A note on prime divisors of polynomials P(Tk); k ≥ 1
- On complete convergence for weighted sums of arrays of rowwise END random variables and its statistical applications
- Common fixed point theorems for a class of (s, q)-contractive mappings in b-metric-like spaces and applications to integral equations
Articles in the same Issue
- Prof. RNDr. pavel brunovský, DrSc. passed away ∗dec. 5, 1934 – † dec. 14, 2018
- Ideals and congruences in pseudo-BCH algebras
- Regular double p-algebras
- Distributive nearlattices with a necessity modal operator
- States in generalized probabilistic models: An approach based in algebraic geometry
- Free power-associative n-ary groupoids
- On the 2-class field tower of subfields of some cyclotomic ℤ2-extensions
- On the space of generalized theta-series for certain quadratic forms in any number of variables
- Uniqueness and periodicity for meromorphic functions with partial sharing values
- Mild solution of stochastic partial differential equation with nonlocal conditions and noncompact semigroups
- An inverse boundary problem for fourth-order Schrödinger equations with partial data
- Entropy as an integral operator
- Global behavior of two third order rational difference equations with quadratic terms
- Measures on effect algebras
- Some topological and combinatorial properties preserved by inverse limits
- The convergence-theoretic approach to weakly first countable spaces and symmetrizable spaces
- Compositions of porouscontinuous functions
- A note on prime divisors of polynomials P(Tk); k ≥ 1
- On complete convergence for weighted sums of arrays of rowwise END random variables and its statistical applications
- Common fixed point theorems for a class of (s, q)-contractive mappings in b-metric-like spaces and applications to integral equations