@ Applied General Topology c© Universidad Politécnica de Valencia Volume 4, No. 1, 2003 pp. 1–13 A note on separation and compactness in categories of convergence spaces Mehmet Baran and Muammer Kula Abstract. In previous papers, various notions of compact, T3, T4, and Tychonoff objects in a topological category were introduced and compared. The main objective of this paper is to characterize each of these classes of objects in the categories of filter and local filter conver- gence spaces as well as to examine how these various generalizations are related. 2000 AMS Classification: 54B30, 54D10, 54A05, 54A20, 18B99, 18D15 Keywords: Topological category, filter convergence spaces, Tychonoff ob- jects, compact objects, normal objects. 1. Introduction The following facts are well known: (1) A topological space X is compact if and only if the projection π2 : X× Y → Y is closed for each topological space Y , (2) A topological space X is Hausdorff if and only if the diagonal, ∆, is closed in X ×X, (3) For a topological space X the following are equivalent: (i) X is Tychonoff (completely regular T1); (ii) X is homeomorphic to a subspace of a compact Hausdorff space; (iii) X is homeomorphic to a subspace of some T4 space. The facts (1) and (2) are used by several authors (see, [7, 14, 22] and [25]) to motivate a closer look at analogous situations in a more general categorical setting. Categorical notions of compactness and Hausdorffness with respect to a factorization structure were defined in the case of a general category by Manes [25] and Herrlich, Salicrup and Strecker [22]. A categorical study of these notions with respect to an appropriate notion of “closedness” based on closure operators (in the sense of [17]) was done in [18] (for the categories of various types of filter convergence spaces) and [14] (for abstract categories). Baran in [2] and [4] introduced the notion of “closedness” and “strong closedness” 2 M. Baran and M. Kula in set-based topological categories and used these notions in [7] to generalize each of the notions of compactness and Hausdorffness to arbitrary set-based topological categories. By using (i) and (ii) of (3), in [7] and [14], there are various ways of general- izing the usual Tychonoff separation axiom to arbitrary set based topological categories. We further recall from [2] and [8] that for a T1 topological space X, the following are equivalent: (a) X is T3; (b) For every non-void subset F of X, the quotient space X/F (defined in 2.1 below) is T2 if it is T1; (c) For every non-void closed subset F of X, the quotient space X/F is a PreT2 space, where a topological space is called PreT2 [2](or R1 in [13]) if for any two distinct points, if there is a neighbourhood of one missing the other, then the two points have disjoint neighbourhoods. The equivalence of (b) and (c) follows from the facts that for T1 topological spaces, T2 is equivalent to PreT2, and F is closed iff X/F is T1. We note also: (d) A topological space X is T4 iff X is T1 and for every non-void subset F of X, the space X/F is T3 if it is T1. In view of (b) - (d), in [2] and [8], there are various ways of generalizing each of the usual T3 and T4 separation axioms to arbitrary set based topological categories. The aim of this paper is to introduce, by using (3), various generalizations of Tychonoff objects for an arbitrary set based topological category and compare them with the ones that were given in [7, 9], and [14]. Furthermore, each of the classes of T3 and T4-objects, compact and strongly compact objects, and Tychonoff objects in the categories of filter and local filter convergence spaces are characterized and relationships among various forms of these Tychonoff objects are investigated in these categories. 2. Preliminaries Let E be a category and Set be the category of sets. The functor U : E → Set is said to be topological, and E is said to be a topological category over Set, if U is concrete (i.e., faithful and amnestic, (i.e., if U(f) = id and f is an isomorphism, then f = id )), has small (i.e., set) fibers, and for which every U-source has an initial lift or, equivalently, for which each U-sink has a final lift [19, 21, 26] or [29]. Note that a topological functor U : E → Set is said to be normalized if there is only one structure on the empty set and on a point [2] or [26]. Let E be a topological category and X ∈ E. Then F is called a subspace of X if the inclusion map i : F → X is an initial lift (i.e, an embedding) and we denote this by F ⊂ X. A Note on Separation and Compactness 3 The categorical terminology is that of [20]. Let B be a set and p ∈ B. Let B ∨ p B be the wedge at p ([2] p. 334), i.e., two disjoint copies of B identified at p, or in other words, the pushout of p : 1 → B along itself (where 1 is a terminal object in Set). More precisely, if i1 and i2 : B → B ∨ p B denote the inclusions of B as the first and second factor, respectively, then i1p = i2p is a pushout diagram. A point x in B ∨ p B will be denoted by x1 (x2) if x is in the first (resp. the second) component of B ∨ p B. Note that p1 = p2. The skewed p-axis map Sp : B ∨ p B → B 2 is given by Sp(x1) = (x,x) and Sp(x2) = (p,x). The fold map at p, ∇p : B ∨ p B → B is given by ∇p(xi) = x for i = 1, 2 ([2] p. 334 or [4] p. 386). Note that the maps Sp and ∇p are the unique maps arising from the above pushout diagram for which Spi1 = (id,id) : B → B2, Spi2 = (p,id) : B → B2, and ∇pij = id,j = 1, 2, respectively, where, id : B → B is the identity map and p : B → B is the constant map at p. The infinite wedge product ∨∞ p B is formed by taking countably many dis- joint copies of B and identifying them at the point p. Let B∞ = B ×B × . . . be the countable cartesian product of B. Define A∞p : ∨∞ p B → B ∞ by A∞p (xi) = (p,p, . . . ,x,p,p, . . .), where xi is in the i-th component of the infinite wedge and x is in the i-th place in (p,p, . . . ,x,p,p, . . .) and 5∞p : ∨∞ p B → B by 5∞p (xi) = x for all i, [2] p. 335 or [4] p. 386. Note, also, that the map A∞p is the unique map arising from the multiple pushout of p : 1 → B for which A∞p ij = (p,p,p, . . . ,p,id,p, . . .) : B → B∞, where the identity map, id, is in the j-th place. Definition 2.1. (cf. [2] p. 335 or [4] p. 386). Let U : E → Set be topological and X an object in E with UX = B. Let F be a non-empty subset of B. We denote by X/F the final lift of the epi U-sink q : U(X) = B → B/F = (B\F)∪{∗}, where q is the epi map that is the identity on B\F and identifies F with a point ∗ ([2] p. 336). Let p be a point in B. (1) X is T1 at p iff the initial lift of the U-source {Sp : B ∨ p B → U(X 2) = B2 and ∇p : B ∨ p B → UD(B) = B} is discrete, where D is the discrete functor which is a left adjoint to U. (2) p is closed iff the initial lift of the U-source {A∞p : ∨∞ p B → B ∞ = U(X∞) and 5∞p : ∨∞ p B → UD(B) = B} is discrete. (3) F ⊂ X is strongly closed iff X/F is T1 at ∗ or F = ∅. (4) F ⊂ X is closed iff ∗, the image of F, is closed in X/F or F = ∅. (5) If B = F = ∅, then we define F to be both closed and strongly closed. Remark 2.2. (1). In Top, the category of topological spaces, the notion of closedness coincides with the usual closedness [2], and F is strongly closed iff F is closed and for each x 6∈ F there exists a neighbourhood of F missing x [2]. If a topological space is T1, then the notions of closedness and strong closedness coincide [2]. 4 M. Baran and M. Kula (2). In general, for an arbitrary topological category, the notions of closed- ness and strong closedness are independent of each other [4]. Even if X ∈ E is T1, where E is a topological category, then these notions are still independent of each other ([8] p. 64). Let A be a set and L a function on A that assigns to each point x of A a set of filters (proper or not, where a filter δ is proper iff δ does not contain the empty set, ∅, i.e., δ 6= [∅]), called the “filters converging to x”. L is called a convergence structure on A (and (A,L) a filter convergence space) iff it satisfies the following two conditions: 1. [x] = [{x}] ∈ L(x) for each x ∈ A (where [F] = {B ⊂ A | F ⊂ B}). 2. β ⊃ α ∈ L(x) implies β ∈ L(x) for any filter β on A. A map f : (A,L) → (B,S) between filter convergence spaces is called con- tinuous iff α ∈ L(x) implies f(α) ∈ S(f(x)) (where f(α) denotes the filter generated by {f(D) | D ∈ α}. The category of filter convergence spaces and continuous maps is denoted by FCO (see [15] p.45 or [30] p.354). A filter con- vergence space (A,L) is said to be a local filter convergence space (in [29], it is called a convergence space) if α ∩ [x] ∈ L(x) whenever α ∈ L(x) ([28] p.1374 or [29] p.142). These spaces are the objects of the full subcategory LFCO (in [29] Conv) of FCO. Note that both of these categories are (normalized) topological categories [28], or [29]. More on these categories can be found in [1, 16, 24, 28, 29], and [30]. For filters α and β we denote by α∪β the smallest filter containing both α and β. Remark 2.3. An epimorphism f : (A,S) → (B,L) in FCO (resp., LFCO) is final iff for each b ∈ B, α ∈ L(b) implies that f(β) ⊂ α for some point a ∈ A and filter β ∈ S(a) with f(a) = b ([28] p.1374 or [29] p.143). Remark 2.4. A source {fi : (B,L) → (Bi,Li), i ∈ I} in FCO (resp., LFCO) is initial iff α ∈ L(a), for a ∈ B, precisely when fi(α) ∈ Li(fi(a)) for all i ∈ I ([15] p.46, [28] p.1374 or [29] p.20). We give the following useful lemmas which will be needed later. Lemma 2.5. (cf. [3], Lemma 3.16). Let ∅ 6= F ⊂ B and let q : B → B/F be the epi map that is the identity on B\F and identifies F to the point ∗. (1) For a ∈ B with a 6∈ F, q(α) ⊂ [a] iff α ⊂ [a], (2) q(α) ⊂ [∗] iff α∪ [F] is proper. Lemma 2.6. (cf. [10], Lemma 3.2). Let f : A → B be a map. (1) If α and β are proper filters on A, then f(α) ∪f(β) ⊂ f(α∪β). (2) If δ is a proper filter on B, then δ ⊂ ff−1(δ), where f−1(δ) is the proper filter generated by {f−1(D) | D ∈ δ}. Lemma 2.7. (cf. [8], Lemma 1.4) Let α and β be proper filters on B. Then q(α)∪q(β) is proper iff either α∪β is proper or α∪ [F] and β∪ [F] are proper. A Note on Separation and Compactness 5 3. T2-objects Recall, in [2] and [6], that there are various ways of generalizing the usual T2 separation axiom to topological categories. Moreover, the relationships among various forms of T2-objects are established in [6]. Let B be a set and B2 ∨ ∆ B 2 the wedge product of B2, i.e. two disjoint copies of B2 identified along the diagonal, ∆. A point (x,y) in B2 ∨ ∆ B 2 will be denoted by (x,y)1 (resp. (x,y)2)) if (x,y) is in the first (resp., second) compo- nent of B2 ∨ ∆ B 2 [10]. Recall that the principal axis map A : B2 ∨ ∆ B 2 → B3 is given by A(x,y)1 = (x,y,x) and A(x,y)2 = (x,x,y). The skewed axis map S : B2 ∨ ∆ B 2 → B3 is given by S(x,y)1 = (x,y,y), S(x,y)2 = (x,x,y), and the fold map, 5 : B2 ∨ ∆ B 2 → B2 is given by 5(x,y)i = (x,y) for i = 1,2 [2]. Definition 3.1. Let U : E → Set be topological and X an object in E with UX = B. 1. X is T ′0 iff the initial lift of the U-source {id : B2 ∨ ∆ B2 → U(B2 ∨ ∆ B2)′ = B2 ∨ ∆ B2 and 5 : B2 ∨ ∆ B2 → UD(B2) = B2} is discrete, where (B2 ∨ ∆ B 2)′ is the final lift of the U-sink {i1, i2 : U(X2) = B2 → B2 ∨ ∆ B 2}. Here, i1 and i2 are the canonical injec- tions. 2. X is T1 iff the initial lift of the U-source {S : B2 ∨ ∆ B 2 → U(X3) = B3 and 5 : B2 ∨ ∆ B 2 → UD(B2) = B2} is discrete. 3. X is PreT ′2 iff the initial lift of the U-source {S : B2 ∨ ∆ B 2 → U(X3) = B3} and the final lift of the U-sink {i1, i2 : U(X2) = B2 → B2 ∨ ∆ B 2} coincide. 4. X is ∆T2 iff the diagonal, ∆, is closed in X2. 5. X is ST2 iff ∆ is strongly closed in X2. 6. X is T ′2 iff X is T ′ 0 and PreT ′ 2. Remark 3.2. (1). Note that for the category Top of topological spaces, T ′0, T1, PreT ′2, and all of the T2’s reduce to the usual T0, T1, PreT2 and T2 separation axioms, respectively [2]. (2) If U : E → B, where B is a topos [23], then parts (1) - (3), (5), and (6) of Definition 3.1 still make sense since each of these notions requires only finite products and finite colimits in their definitions. Furthermore, if B has infinite products and infinite wedge products, then Definition 3.1 (4) also makes sense. Lemma 3.3. Let (B,L) be in FCO (resp., LFCO) and ∅ 6= F ⊂ B. (1) (B,L) is T1 iff for each distinct pair of points x and y in B, [x] 6∈ L(y). (2) All objects (B,L) in FCO (resp., LFCO) are T ′0. (3) ∅ 6= F ⊂ B is closed iff for any a 6∈ F , if there exists α ∈ L(a) such that α∪ [F] is proper, then [a] 6∈ L(c) for all c ∈ F . 6 M. Baran and M. Kula (4) ∅ 6= F ⊂ B is strongly closed iff for any a ∈ B with a 6∈ F , [a] 6∈ L(c) for all c ∈ F and α∪ [F] is improper for all α ∈ L(a). (5) (B,L) is ∆T2 iff for all x 6= y in B, [x] 6∈ K(y) iff (B,L) is T1. (6) (B,L) is ST2 iff for all x 6= y in B, L(x) ∩L(y) = {[∅]}. (7) (B,L) is PreT ′2 (T ′ 2) iff (B,L) is discrete, i.e, for all x in B, L(x) = {[∅], [x]}. Proof. (1), (2), and (7) are proved in [5]. The proof of (3)-(6) are given in [4]. � Corollary 3.4. Let (B,L) be in FCO (resp. LFCO) and ∅ 6= F ⊂ B. (1) If (B,L) is T1, then B/F is T1 iff F is strongly closed. (2) If (B,L) is T1, then F is always closed. (3) If (B,L) is T1, then F is strongly closed iff ∀x ∈ B if x 6∈ F and α ∈ L(x), then α∪ [F] is improper. (4) If (B,L) is T ′2, then all the subsets of B are both closed and strongly closed. 4. T3-objects We now recall, ([2] and [8]), various generalizations of the usual T3 separation axiom to arbitrary set based topological categories and characterize each of them for the topological categories FCO and LFCO. Definition 4.1. Let U : E → Set be topological and X an object in E with UX = B. Let F be a non-empty subset of B. 1. X is ST ′3 iff X is T1 and X/F is PreT ′ 2 for all strongly closed F 6= ∅ in U(X). 2. X is T ′3 iff X is T1 and X/F is PreT ′ 2 for all closed F 6= ∅ in U(X). 3. X is ∆T3 iff X is T1 and X/F is ∆T2 if it is T1, for all F 6= ∅ in U(X). 4. X is ST3 iff X is T1 and X/F is ST2 if it is T1, for all F 6= ∅ in U(X). Remark 4.2. (1). For the category Top of topological spaces, all of the T3’s reduce to the usual T3 separation axiom ([2] and [8]). (2). If U : E → B, where B is a topos [23], then Parts (1), (3), and (4) of Definition 4.1 still make sense since each of these notions requires only finite products and finite colimits in their definitions. Furthermore, if B has infinite products and infinite wedge products, then Definition 4.1 (2), also, makes sense. Theorem 4.3. Let (B,L) be in FCO (resp. LFCO). (1) (B,L) is ∆T3 iff (B,L) is T1. (2) (B,L) is ST3 iff (B,L) is ST2. (3) (B,L) is ST ′3 iff for all x 6= y in F , [x] 6∈ L(y) and for any x ∈ B and for any proper filter α ∈ L(x), either α = [x] or F ∈ α for all non-empty strongly closed subsets F of B. (4) (B,L) is T ′3 iff for all x 6= y in F , [x] 6∈ L(y) for any x ∈ B and for any proper filter α ∈ L(x) either α = [x] or F ∈ α for any non-empty subset F of B. A Note on Separation and Compactness 7 Proof. (1). This follows from Definition 4.1 and Corollary 3.4. (2). Suppose (B,L) is ST3. Take F ={a}, a one point set. It now follows from Lemma 3.3 and Corollary 3.4 that (B,L) is ST2. Conversely, suppose (B,L) is ST2. By Corollary 3.4, (B,L) is T1. Suppose B/F is T1, then by Corollary 3.4, F is a strongly closed subset of B. We show that B/F is ST2. Let x 6= y in B and α ∈ L′(x) ∩ L′(y), where L′ is the quotient structure on B/F induced by the map q : B → B/F that identifies F with a point ∗ and is the identity on B\F. If α is improper, then, by Corollary 3.4, we are done. Suppose α is proper. Since q is the quotient map this implies (see Remark 2.3) that ∃β ∈ L(a) and ∃δ ∈ L(b) such that q(β) ⊂ α, q(δ) ⊂ α, and qa = x, qb = y. It follows that q(β) ∪ q(δ) is proper and, by Lemma 2.7, either β ∪ δ is proper or β ∪ [F] and δ ∪ [F] are proper. The first case cannot occur since (B,L) is ST2. Since x 6= y, we may assume a 6∈ F. Since F is strongly closed, by Corollary 3.4, β ∪ [F] is improper. This shows that the second case also cannot hold. Therefore, α must be improper and by Corollary 3.4, we have the result. (3). Suppose (B,L) is ST ′3. Since (B,L) is T1, by Corollary 3.4, for all x 6= y in B, [x] 6∈ L(y). If α ∈ L(x), where x ∈ B, then q(α) ∈ L′(qx). Since B/F is PreT ′2, (F is a non-empty strongly closed subset of B) by Corollary 3.4, q(α) = [qx] (since α is proper). If x 6∈ F, then, by Lemma 2.6, [x] = q−1(x) = q−1q(α) ⊂ α and consequently α = [x]. If x ∈ F, it follows easily that q(α) = [∗] iff F ∈ α. Conversely, suppose the conditions hold. By Corollary 3.4, clearly, (B,L) is T1. We now show that B/F is PreT ′2 for all nonempty strongly closed subsets F of X. If x ∈ B/F and α ∈ L′(x), it follows that there exists β ∈ L(a) such that q(β) ⊂ α and qa = x. If β is improper, then so is α. If β is proper, then by assumption β = [a] or F ∈ β. If the first case holds, then [qa] = q(β) ⊂ α and thus α = [qa]. If the second case holds, then {∗} = q(F) ∈ q(β) ⊂ α and consequently α = [∗]. Hence, by Lemma 3.3, B/F is PreT ′2 and by Definition 3.1, (B,L) is ST ′ 3. The proof of (4) is similar to the proof of (3), on using Definition 3.1, Lemma 3.3 and Corollary 3.4. � Remark 4.4. For the category FCO (resp., LFCO), we have : (1) By Theorem 4.3, ST ′3 ⇒ T ′3 ⇒ ST3 ⇒ ∆T3, but the converse of each implication is not true in general. (2) By Lemma 3.3 and Theorem 4.3, ST ′3 ⇒ T ′2 ⇒ ST3 ≡ ST2 ⇒ ∆T3 = ∆T2, but the converse of each implication is not true in general. (3) By Corollary 3.4 and Theorem 4.3, if (B,L) is ST ′3 or T ′ 3, then all subsets of X are both closed and strongly closed. (4) By Corollary 3.4 and Theorem 4.3, if (B,L) is ∆T3, then F is always closed and F is strongly closed iff ∀x ∈ B if x 6∈ F and α ∈ K(x), then α∪ [F] is improper. 8 M. Baran and M. Kula 5. T4-objects We now recall various generalizations of the usual T4 separation axiom to arbitrary set based topological categories that are defined in [2] and [8], and characterize each of them for the topological categories FCO and LFCO. Definition 5.1. Let U : E → Set be topological and X an object in E with UX = B. Let F be a non-empty subset of B. 1. X is ST ′4 iff X is T1 and X/F is ST ′ 3 if it is T1, where F is any non- empty subset of U(X). 2. X is T ′4 iff X is T1 and X/F is T ′ 3 if it is T1, where F is any non-empty subset of U(X). 3. X is ∆T4 iff X is T1 and X/F is ∆T3 if it is T1, for all F 6= ∅ in U(X). 4. X is ST4 iff X is T1 and X/F is ST3 if it is T1, for all F 6= ∅ in U(X). Remark 5.2. (1). For the category Top of topological spaces, all of the T4’s reduce to the usual T4 separation axiom by the Introduction, [2], and [8]. (2). If U : E → B, where B is a topos [23], then Definition 5.1 still makes sense since each of these notions requires only finite products and finite colimits in their definitions. Theorem 5.3. Let (B,L) be in FCO (resp., LFCO). (1) (B,L) is ∆T4 iff (B,L) is T1. (2) (B,L) is ST4 iff (B,L) is ST2. (3) (B,L) is ST ′4 (T ′ 4) iff the following two conditions hold: (i) For all x 6= y in B, we have [x] 6∈ L(y). (ii) For any x ∈ B and for any proper filter α ∈ L(x), and for any non-empty disjoint strongly closed (resp., closed ) subsets F and F ′ of B, we have either condition (I ) or (II ) below: (I) α = [x]; (II) F ∈ α or F ′ ∈ α. Proof. (1). This follows from Definition 5.1 and Theorem 4.3. (2). The proof has the same form as that of Theorem 4.3 (2). One has only to replace the term ST3 by ST4 and the numbers 3.3, 3.4, 3.4, 3.4, 3.4, 2.3, 3.4, 3.4 respectively by 3.1, 3.3, 4.3, 3.4, 4.3, 2.3, 3.3, 4.3. (3). Suppose (B,L) is ST ′4. Since (B,L) is T1, by Corollary 3.4, for all x 6= y in B, [x] 6∈ L(y). If α ∈ L(x), where x ∈ B, then q(α) ∈ L′(qx), where L′ is the quotient structure on B/F induced by the map q of Definition 2.1. Since B/F is ST ′3, (F is a non-empty strongly closed subset of B, i.e., B/F is T1) by Corollary 3.4, we have either q(α) = [qx] (since α is proper) or F ′ ∈ q(α), for any non-empty strongly closed subset F ′ of B/F not containing the point ∗ (Note that q−1(F ′) = F ′ and F ′ is disjoint from F). Suppose that q(α) = [qx]. If x 6∈ F, then, by Lemma 2.6, [x] = q−1(x) = q−1q(α) ⊂ α, and consequently α = [x]. If x ∈ F, it follows easily that q(α) = [∗] iff F ∈ α. If F ′ ∈ q(α) for any non-empty strongly closed subset F ′ of B/F not con- taining the point ∗, then it follows easily that F ′ ∈ α. A Note on Separation and Compactness 9 Conversely, suppose the conditions hold. By Lemma 3.3, clearly, (B,L) is T1. We now show that B/F is ST ′3 for all non-empty strongly closed subsets F of B. If x ∈ B/F and α ∈ L′(x), it follows that there exists β ∈ L(a) such that q(β) ⊂ α and qa = x. If β is improper, then so is α. If β is proper, then by assumption either β = [a] or F ∈ β, or F ′ ∈ β for any strongly closed subset F ′ of B disjoint from F. If the first case holds, then [qa] = q(β) ⊂ α and thus α = [qa]. If the second case holds, then {∗} = q(F) ∈ q(β) ⊂ α, and consequently α = [∗] or F ′ = q(F ′) ∈ q(β) ⊂ α and consequently F ′ ∈ α. Hence, by Theorem 4.3, B/F is ST ′3 and by Definition 5.1, (B,L) is ST ′ 4. The proof for T ′4 is similar to the proof for ST ′ 4. � Remark 5.4. For the category FCO (resp., LFCO), we have : (1). By Theorem 4.3, ST ′4 ⇒ T ′4 ⇒ ST4 ⇒ ∆T4, but the converse of each implication is not true in general. (2). By Lemma 3.3, Theorem 4.3, and Theorem 5.3, ST ′4(T ′ 4) ⇒ ST ′3(T ′3) ⇒ T ′2 ⇒ ST4 = ST3 = ST2 ⇒ ∆T4 = ∆T3 = ∆T2, but the converse of each implication is not true in general. (3). By Remark 4.4 and Theorem 5.3, if (B,L) is ST ′4 or T ′ 4, then all subsets of X are both closed and strongly closed. (4). By Remark 4.4 and Theorem 5.3, if (B,L) is ∆T4, then all subsets F of X are closed and F is strongly closed iff ∀x ∈ B, if x 6∈ F and α ∈ L(x), then α∪ [F] is improper. Corollary 5.5. Let (B,L) be in FCO (resp., LFCO). If (B,L) is ∆T4,ST4, ST ′4 or T ′ 4, then any subspace of (B,L) is ∆T4,ST4,ST ′ 4 or T ′ 4, respectively. Proof. This follows from Remark 2.4, Theorem 5.3, and Remark 5.4 (3). � 6. Compact objects Recall that each of the notions of (strongly) closed morphism and (strongly) compact object in a topological category E over Set are introduced in [7]. Definition 6.1. Let U : E → Set be topological, X and Y objects in E, and f : X → Y a morphism in E. 1. f is said to be closed iff the image of each closed subobject of X is a closed subobject of Y . 2. f is said to be strongly closed iff the image of each strongly closed subobject of X is a strongly closed subobject of Y . 3. X is compact if and only if the projection π2 : X ×Y → Y is closed for each object Y in E. 4. X is strongly compact if and only if the projection π2 : X ×Y → Y is strongly closed for each object Y in E. Remark 6.2. (1). For the category Top of topological spaces, the notions of closed morphism and compactness reduce to the usual ones ([12] p. 97 and 10 M. Baran and M. Kula 103). Furthermore, by Remark 2.2 and Definition 6.1, one can show that the notions of compactness and strong compactness are equivalent. (2). If U : E → B is topological, where B is a topos with infinite products and infinite wedge products, then Definition 6.1 still makes sense. (3). Since the notions of closedness and strong closedness are, in general, different (see [4] p. 393), it follows that the notions of compactness and strong compactness are different, in general. (4). For an arbitrary topological category, it is not known in general whether the closure used in 2.1 is a closure operator in the sense of Dikranjan and Giuli [17] or not. However, it is shown, in [10], that the notions of closedness and strong closedness that are defined in 2.1 form appropriate closure operators in the sense of Dikranjan and Giuli [17] in case the category is one of the categories FCO and LFCO. The same two facts are proved in [11] for the categories Lim (limit spaces) and PrTop (pretopological spaces). Theorem 6.3. Let E be one of the categories FCO (resp. LFCO). (1) Every (B,L) ∈ E is compact. (2) (B,L) ∈ E is strongly compact iff every ultrafilter in B converges. Proof. (1). By Definition 5.1 (3) we need to show that, for all (A,S) ∈ E, π2 : (B,L)×(A,S) → (A,S) is closed. Suppose M ⊂ B×A is closed. Suppose that for any a ∈ A there exists c ∈ π2M such that [a] ∈ S(c). It follows that ∃x ∈ B such that (x,c) ∈ M. Note that [(x,a)] ∈ L2((x,c)), where L2 is the product structure on B × A, (since [x] ∈ L(x) and [a] ∈ S(c)). Since M is closed, (x,a) ∈ M and consequently a = π2(x,a) ∈ π2(M). Hence, by Lemma 3.3, π2(M) is closed and consequently, (B,L) is compact. (2). Suppose every ultrafilter in B converges. We show that (B,L) is strongly compact, i.e., by Definition 6.1 (4), we need to show that, for all (A,S) ∈ E, π2 : (B,L) × (A,S) → (A,S) is strongly closed. Suppose that M ⊂ B × A is strongly closed. To show that π2M is strongly closed, we as- sume the contrary and apply Lemma 3.3 (4). Thus for some point a ∈ A with a 6∈ π2M, we have either [a] ∈ S(c) for some c ∈ π2M or [π2M] ∪α is proper for some α ∈ S(a). If the first case holds, that is for some a ∈ A we have a 6∈ π2M and [a] ∈ S(c) for some c ∈ π2M, then it follows that ∃x ∈ B such that (x,a) 6∈ M. Note that [(x,a)] ∈ L2((x,c)), a contradiction, since M is strongly closed. In the second case, suppose that for some a ∈ A with a 6∈ π2M and α ∈ S(a), [π2M] ∪α is proper. Let σ = [M] ∪π−12 α. Note that σ is proper and π1(σ) is a filter on B. It follows that there exists an ultrafilter β on B with β ⊃ π1(σ). In view of the assumption on (B,L), there exists x ∈ B such that β ∈ L(x). Let γ = π−11 β ∪ π −1 2 α . Note that γ ∈ L 2(x,a) since π1(γ) = β ∈ L(x) and π2(γ) = α ∈ S(a). Since a 6∈ π2M, we have (x,a) 6∈ M. It follows from β ⊃ π1(σ) that [M] ∪γ is proper, a contradiction since M is strongly closed, by Lemma 3.3 (4). Hence, by Lemma 3.3 (4), π2(M) must be strongly closed and consequently, by Definition 6.1, (B,L) is strongly compact. A Note on Separation and Compactness 11 Conversely, assume that (B,L) is strongly compact and α is a non convergent ultrafilter of B, i.e., for all x ∈ B, α 6∈ L(x). Let A be the set obtained by adjoining a new element, say ∞, to B, i.e., A = B∪{∞}. Let (A,S), where S is defined by S(x) = {[∅], [x]} for each x 6= ∞ of A, and β ∈ S(∞) iff α = β∪[B], i.e., the trace of β on B coincides with α. Note that (A,S) ∈ FCO (resp., LFCO). Let ∆ = {(x,y) ∈ B ×A | x = y}⊂ B ×A. Let σ = π−11 [x] ∪π −1 2 α. Since π1σ = [x] ∈ L(x) and π2σ = α ∈ S(∞), σ ∈ L2((x,∞)), where L2 is the product structure on B×A. Note that σ∪[∆] is improper (let V = A\{x}∈ α and V ∩ ∆ = ∅). Since [∞] 6∈ S(c) for all c ∈ B, it follows that [(x,∞)] 6∈ L2(c,c). Hence, by Lemma 3.3, ∆ is strongly closed in B × A. Note that α ∪ [π2(∆)] is proper for α ∈ S(∞), a contradiction since (B,L) is strongly compact. � Remark 6.4. Results akin to Theorem 6.3 have been proved for the categories Lim (limit spaces) and PrTop (pretopological spaces) in ([11], Lemma 4.3). 7. Tychonoff objects We now define various forms of Tychonoff objects for an arbitrary set-based topological category. Furthermore, we characterize each of them for the cate- gories that are mentioned in Section 2 and investigate the relationships among them. Definition 7.1. Let U : E → Set be topological and X an object in E. 1. X is ∆T3 12 iff X is a subspace of ∆T4. 2. X is ST3 12 iff X is a subspace of ST4. 3. X is T ′ 3 12 iff X is a subspace of T ′4. 4. X is ST ′ 3 12 iff X is a subspace of ST ′4. 5. X is C∆T3 12 iff X is a subspace of a compact ∆T2. 6. X is CST3 12 iff X is a subspace of a compact ST2. 7. X is LT3 12 iff X is a subspace of a compact T ′ 2. 8. X is S∆T3 12 iff X is a subspace of a strongly compact ∆T2. 9. X is SST3 12 iff X is a subspace of a strongly compact ST2. 10. X is SLT3 12 iff X is a subspace of a strongly compact T ′ 2. Remark 7.2. (1). For the category Top of topological spaces, all ten of the properties defined in Definition 7.1 are equivalent and reduce to the usual T3 12 = Tychonoff, i.e, completely regular T1, spaces [27], Remark 5.2, and Remark 6.2. (2). For an arbitrary set-based topological category, properties (3–4) and (5–7) are defined in [8] and [7], respectively. (3). For the categories FCO and LFCO, it is shown in [10] that the notions of closedness and strong closedness form appropriate closure operators in the sense of [16]. As a consequence, properties (5) and (9) of Definition 6.1 reduce to Definition 8.1 of [13]. 12 M. Baran and M. Kula Theorem 7.3. Let X be in FCO (resp., LFCO). 1. X is ∆T3 12 (C∆T3 12 ) iff X is T1. 2. X is ST3 12 (CST3 12 ) iff X is ST2. 3. X is T ′ 3 12 iff X is T ′4. 4. X is ST ′ 3 12 iff X is ST ′4. 5. X is S∆T3 12 iff X is a subspace of a strongly compact T1. 6. X is SST3 12 iff X is a subspace of a strongly compact ST2. 7. X is SLT3 12 iff X is a finite discrete space. Proof. (1)-(6) follow from Lemma 3.3, Theorem 5.3, Corollary 5.5, Theorem 6.3 and Definition 7.1. Note that every strongly compact discrete space is finite. Thus, (7) follows from Remark 2.4, Lemma 3.3, Theorem 6.3, Definition 7.1, and this fact. � Remark 7.4. For the Category FCO (resp. LFCO), one has: By Theo- rem 5.3, Theorem 6.3 and Theorem 7.3, SLT3 12 ⇒ ST ′ 3 12 ⇒ T ′ 3 12 ⇒ ST3 12 = CST3 12 ⇒ ∆T3 12 = C∆T3 12 and SLT3 12 ⇒ SST3 12 ⇒ S∆T3 12 but the converse of each implication is not true, in general. References [1] J. Adamek, H. Herrlich, G. E. Strecker, Abstract and Concrete Categories (John Wiley and Sons, New York, 1990). [2] M. Baran, Separation properties, Indian J. Pure Appl. Math. 23 (1992), 333–341. [3] M. Baran, Stacks and filters, Doğa Mat. (Turkish J. Math.) 16 (1992), 95–108. [4] M. Baran, The notion of closedness in topological categories, Comment. Math. Univ. Carolinae 34 (1993), 383–395. [5] M. Baran, Separation properties in category of filter convergence spaces, Bull. Calcutta Math. Soc. 85 (1993), 249–254. [6] M. Baran and H. Altındis, T2-objects in topological categories, Acta Math. Hungar. 71 (1996), 41–48. [7] M. Baran, A notion of compactness in topological categories, Publ. Math. Debrecen 50 (1997), 221–234. [8] M. Baran, T3 and T4-objects in topological categories, Indian J. Pure Appl. Math. 29 (1998), 59–69. [9] M. Baran, Completely regular objects and normal objects in topological categories, Acta Math. Hungar. 80 (1998), 211–224. [10] M. Baran, Closure operators in convergence spaces, Acta Math. Hungar., 87 (2000), 33–45. [11] M. Baran, Compactness, perfectness, separation, minimality and closedness with respect to closure operators, Applied Categorical Structures, 10 (2002), 403–415. [12] N. Bourbaki, General Topology (Addison-Wesley Publ. Co., 1966). [13] A. S. Davis, Indexed systems of neighbourhoods for general topological spaces, Amer. Math. Monthly, 68 (1961), 886–893. [14] M. M. Clementino, E. Giuli, and W. Tholen, Topology in a category: Compactness, Port. Math. 53 (1996), 397–433. [15] D. Dikranjan and W. Tholen, Categorical structure of closure operators (Kluwer Aca- demic Publishers, Dordrecht, 1995). [16] D. Dikranjan and E. Giuli, Epis in categories of convergence spaces, Acta Math. Hungar. 61 (1993), 195–201. A Note on Separation and Compactness 13 [17] D. Dikranjan and E. Giuli, Closure operators I, Topology Appl. 27 (1987), 129–143. [18] D. Dikranjan and E. Giuli, Compactness, minimality and closedness with respect to a closure operator, In: J. Adamek and S. MacLane (eds), Categorical Topology (Prague 1988), (World Scientific Publ., Singapore, 1990), 284–295. [19] D. Dikranjan, E. Giuli and A. Tozzi, Topological categories and closure operators, Quaes- tiones Math. 11 (1988), 323–337. [20] H. Herrlich and G. E. Strecker, Category Theory, 2nd ed. (Heldermann Verlag, Berlin, 1979). [21] H. Herrlich, Topological functors, Gen. Topology Appl. 4 (1974), 125–142. [22] H. Herrlich, G. Salicrup, and G. E. Strecker, Factorizations, denseness, separation, and relatively compact objects, Topology Appl. 27 (1987), 157–169. [23] P. T. Johnstone, Topos Theory, L. M. S Mathematics Monograph: No. 10 (Academic Press, New York, 1977). [24] E. Lowen-Colebunders, Function classes of Cauchy continuous maps (Marcel Dekker Inc., New York, 1989). [25] E. G. Manes, Compact hausdorff objects, Gen. Topology Appl. 4 (1974), 341–360. [26] M. V. Mielke , Geometric topological completions with universal final lifts, Topology Appl. 9 (1985), 277–293. [27] J. R. Munkres, Topology: A First Course (Prentice Hall, 1975). [28] L. D. Nel , Initially structured categories and Cartesian closedness, Canad. J. Math. 27 (1975), 1361–1377. [29] G. Preuss, Theory of Topological Structures, An Approach to Topological Categories (D. Reidel Publ. Co., Dordrecht, 1988). [30] F. Schwarz, Connections between convergence and nearness, Lecture Notes in Math. No. 719, (Springer-Verlag, 1978), 345–354. Received November 2001 Revised December 2002 Mehmet Baran and Muammer Kula Department of Mathematics, Erciyes University, Kayseri 38039, Turkey. E-mail address : baran@erciyes.edu.tr A note on separation and compactness in categories of convergence spaces. By M. Baran and M. Kula