arXiv:1703.06862v4 [math.GN] 16 May 2017 Quasi-components in widely-connected spaces David Lipham Department of Mathematics and Statistics Auburn University Auburn, AL 36849, United States E-mail: [email protected] Abstract An unanswered question in general topology asks whether the Stone-Čech compactification of a widely-connected space is necessarily an indecomposable continuum. We describe a property of X that is necessary and sufficient in order for βX to be indecomposable, and examine the case when X is a separable metric space. We also construct two widely-connected sets in R3 – one is contained in a composant of each of its compactifications, and the other fails to be indecomposable upon the addition of a single limit point. 1 Introduction A connected topological space W is widely-connected if every non-degenerate connected subset of W is dense in W . This paper addresses a collection of open problems concerning widely-connected spaces. They are contained in the following multi-part question from the 2004 Spring Topology and Dynamics Conference report, due to Jerzy Mioduszewski. Part (a) of the question is also posed in [1] by David Bellamy, who conjectured a positive answer to part (b). Question 1 (Mioduszewski; 23 in [9]). Let W be a widely-connected space. (a) Is βW an indecomposable continuum? (b) If W is metrizable and separable, does W have a metric compactification which is an indecomposable continuum? (c) If W is metrizable and separable, does W have a metric compactification γW such that for every composant P of γW , W ∩ P is (i) totally disconnected? (ii) finite? (iii) a singleton? 2010 Mathematics Subject Classification: Primary 54D05; Secondary 54F15, 54E50, 54G15. Key words and phrases: widely-connected, indecomposable, quasi-component 1 2 D. Lipham γW is a compactification of W if γW is a compact Hausdorff space in which W is densely embedded; βW denotes the Stone-Čech compactification of W ; P is a composant of γW if P is the union of all proper subcontinua of γW that contain a given point. A connected compact Hausdorff space (a continuum) is indecomposable if it is not the union of any two of its proper subcontinua. If X is a continuum, then X is indecomposable if and only if every proper subcontinuum of X is nowhere dense in X ([7] §48 V Theorem 2). Thus, the term indecomposable can be consistently defined for arbitrary topological spaces – say that X is indecomposable if every connected subset of X is either dense or nowhere dense in X. The process of compactifying a space can destroy indecomposability. In fact, by a result of Philip Bowers [2] every widely-connected separable metric space has a compactification equal to the Hilbert cube. There is also a widely-connected subset of Euclidean 3-space that fails to be indecomposable when a single limit point is added to it – Section 3.1. The existence of such an example was suggested by Mary Ellen Rudin [10], who proved that this scenario is impossible in the plane but never published her example. The special limit point glues together many disjoint closed subsets of the original space, and is therefore not representative of a point in the Stone-Čech compactification. Using properties of quasi-components, we prove that Questions 1(a) and 1(b) are equivalent when W is metrizable and separable. In the language of continuum theory they are equivalent to: If W is a connected separable metric space that is irreducible between every two of its points, then does W have an irreducible compactification? Perhaps surprisingly, W may not have a compactification that is irreducible between two points in W . There is a widely-connected W ⊆ R3 such that if γW is any compactification of W then there is a composant P of γW such that W ⊆ P – Section 3.2. This provides a complete (negative) answer to Question 1(c), improving one of our previous results in Section 4 of [8]. 2 A property of quasi-components If X is a topological space, then X is strongly indecomposable means that for every two non-empty disjoint open sets U and V there are two closed sets A and B such that X = A ∪ B, A ∩ U 6= ∅, B ∩ U 6= ∅, and A ∩ B ⊆ V . Remark. X is strongly indecomposable if and only if the quasi-components of proper closed subsets of X are nowhere dense in X. Perfect totally disconnected ⇒ strongly indecomposable ⇒ indecomposable. Quasi-components 3 Lemma 1. If Y is strongly indecomposable and X ⊆ Y such that X = Y , then X is strongly indecomposable. Proof. Let U and V be non-empty disjoint open subsets of X. Let U 0 and V 0 be open subsets of Y such that U = U 0 ∩ X and V = V 0 ∩ X; X = Y implies U 0 ∩ V 0 = ∅. There are closed subsets A0 and B 0 of Y such that Y = A0 ∪ B 0 , U 0 ∩ A0 6= ∅, U 0 ∩ B 0 6= ∅, and A0 ∩ B 0 ⊆ V 0 . Let A = A0 ∩ X and B = B 0 ∩ X. Clearly A and B are closed in X, X = A ∪ B, and A ∩ B ⊆ V . Note that U 0 ∩ A0 = U 0 \ B 0 and U 0 ∩ B 0 = U 0 \ A0 are non-empty open subsets of Y . So X = Y implies that U ∩ A 6= ∅ and U ∩ B 6= ∅. The following is implicit in the proof of Lemma 9.8 in [12]. Lemma 2. If U and V are disjoint open subsets of X and W is open in βX such that W ∩ X = U ∪ V , then W0 := W ∩ clβX U and W1 := W ∩ clβX V are disjoint open sets in βX and W0 ∪ W1 = W . Proof. By density of X in βX we have W ⊆ clβX U ∪ clβX V . Thus W0 ∪ W1 = W . We need to show that W ∩ clβX U ∩ clβX V = ∅. Suppose for a contradiction that there exists p ∈ W ∩ clβX U ∩ clβX V . By Urysohn’s Lemma there is a mapping F : βX → [0, 1] such that F (p) = 0 and F [βX \ W ] = 1. Define f : X → [0, 1] by ( 1 if x ∈ U f (x) = F (x) if x ∈ / U. If T is open in [0, 1] then ( f −1 [T ] = F −1 [T ] ∩ X ∪ U F −1 [T ] ∩ V if 1 ∈ T if 1 ∈ / T, so f is continuous. Since p ∈ clβX V and f V = F V , we have βf (p) = F (p) = 0. On the other hand, p ∈ clβX U implies that βf (p) = 1, a contradiction. Theorem 3. βX is indecomposable if and only if X is strongly indecomposable. Proof. Suppose that βX is indecomposable. To show that X is strongly indecomposable, by Lemma 1 it suffices to show that βX is strongly indecomposable. Let U and V be non-empty disjoint open subsets of βX. Let T be a non-empty open subset of βX such that T ⊆ V . Since U is not contained in a component of βX \ T , by Theorem 6.1.23 in [5] there are disjoint closed sets A0 and B 0 such that A0 ∪B 0 = βX \T , A0 ∩ U 6= ∅, and B 0 ∩ U 6= ∅. Then A := A0 ∪ T and B := B 0 ∪ T are as desired. Suppose that X is strongly indecomposable. Let K be a proper closed subset of βX with non-empty interior; we show that K is not connected. Let S and T be 4 D. Lipham non-empty open subsets of βX such that S ⊆ K and K ∩ clβX T = ∅. Let A and B be disjoint closed subsets of X such that A ∪ B = X \ T , A ∩ S 6= ∅, and B ∩ S 6= ∅. Note that U := A \ clβX T and V := B \ clβX T are open in X, U ∩ S 6= ∅, and V ∩ S 6= ∅. By Lemma 2, W := βX \ clβX T is the union of the two disjoint open sets W0 := W ∩ clβX U and W1 := W ∩ clβX V . Since K ⊆ W and Wi ∩ K 6= ∅ for each i < 2, K is not connected. Corollary 4. If X is compact Hausdorff, then X is indecomposable if and only if X is strongly indecomposable. Proof. βX ' X when X is compact Hausdorff. Corollary 5. βX is an indecomposable continuum if and only if X is strongly indecomposable and connected. Proof. βX is a continuum if and only if X is connected. Widely-connected spaces are usually constructed as dense subsets of indecomposable continua. Gary Gruenhage [6] constructed completely regular and perfectly normal examples by a different method, assuming Martin’s Axiom and the Continuum Hypothesis, respectively. Their co-infinite subsets are totally disconnected (it’s worth noting that there is no connected metric space with this property), thus both examples are strongly indecomposable, and so their Stone-Čech compactifications are indecomposable. If W is not strongly indecomposable, then it has a perfect hereditarily disconnected subset which also fails to be strongly indecomposable. In point of fact, let U and V be open subsets of W for which the definition fails. Let T be a non-empty open set with T ⊆ V . T is the union of two relatively clopen subsets A and B. Then Y := W \ A is an open hereditarily disconnected subset of W . Since W is connected, Y is perfect. And no clopen partition of Y \ B divides U . Example 1. A perfect hereditarily disconnected set which is not strongly indecomposable. Let C be the middle-thirds Cantor set in [0, 1]. If Q and P denote the rationals and irrationals, respectively, then let X = (C 0 × Q) ∪ (C 00 × P), where C 0 is the set of endpoints of intervals removed during the construction of C, and C 00 = C \ C 0 . The significant property of X is that for each c ∈ C, X ∩ ({c} × R) is a quasi-component of X. Clearly X ∩ ({c} × R) is also dense in {c} × R for each c ∈ C. Thus X is connectible, as defined in [8]. Let X1 = X ∩ (R × [0, 1]), and let X10 = {h−y, xi : hx, yi ∈ X1 } be the copy of X1 rotated 90◦ about the origin. Then Y := X1 ∪ X10 is hereditarily disconnected, Quasi-components 5 and X10 is contained in a quasi-component of Y \ (C ∩ (1/2, 1]) × [0, 1]. To prove the latter, suppose that A is a clopen subset of Y \ (C ∩ (1/2, 1]) × [0, 1] and A ∩ X10 6= ∅. Since A is open, there exists c ∈ C 0 \ {0, 1} such that A ∩ ((−1, 0) × {c}) 6= ∅. Since c ∈ Q and X10 is connectible, h0, ci ∈ A. Then X1 ∩ ({0} × (0, 1)) ⊆ A because X1 ∩ ([0, 1/2) × (0, 1)) is connectible. So X10 ∩ ((−1, 0) × C 0 \ {0, 1}) ⊆ A. Since A is closed, we have X10 ⊆ A. Figure 1: Y ⊆ [−1, 0] × C ∪ C × [0, 1] From now on, assume that X is a separable metric space. The standard metric on P the Hilbert cube [0, 1]ω is given by ρ(y, y 0 ) = n∈ω y(n) − y 0 (n) · 2−n . ([0, 1]ω )X is the set of functions from X into [0, 1]ω , endowed with the complete metric %(f, g) = sup{ρ(f (x), g(x)) : x ∈ X}. Lemmas 6 through 8 will be used to show Questions 1(a) and 1(b) are equivalent for separable metric spaces – Theorem 9. Lemma 6 (§46 V Theorem 3 in [7]). There is a continuous ϕ : X → 2ω such that the quasi-components of X coincide with the non-empty point inverses of ϕ. Lemma 7 (§44 V Corollary 4a and §44 VI Lemma in [7]). If A and B are disjoint closed subsets of X, then n o X g ∈ [0, 1]ω : g[A] ∩ g[B] = ∅ is a dense open subset of ([0, 1]ω )X . Lemma 8 (§44 VI Theorem 2 in [7]). The set of homeomorphic embeddings from X into [0, 1]ω is a dense Gδ -set in ([0, 1]ω )X . Theorem 9. βX is indecomposable if and only if X has an indecomposable metric compactification. Proof. Suppose that βX is indecomposable. Let {Un : n ∈ ω} be a countable basis for X. For each n ∈ ω let ϕn : X \ Un → 2ω be given according to Lemma 6. 6 D. Lipham Let {Ci : i ∈ ω} be the canonical clopen basis for 2ω . For each hn, ii ∈ ω 2 , if −1 ω Ahn,ii = ϕ−1 n [Ci ] and Bhn,ii = ϕn [2 \ Ci ] then n o X g ∈ [0, 1]ω : g[Ahn,ii ] ∩ g[Bhn,ii ] = ∅ is a dense open subset of ([0, 1]ω )X by Lemma 7. By Lemma 8 and the fact that ([0, 1]ω )X is complete, there is a homeomorphic embedding e : X ,→ [0, 1]ω such that e[Ahn,ii ] ∩ e[Bhn,ii ] = ∅ for each hn, ii ∈ ω 2 . The metric compactification e[X] is indecomposable. Let K be a proper closed subset of e[X] with non-empty interior. There exists n ∈ ω such that K ∩ e[Un ] = ∅. Since X is strongly indecomposable, there exists i ∈ ω such that K∩e[Ahn,ii ] 6= ∅ and K∩e[Bhn,ii ] 6= ∅. As e[X] = e[Un ]∪e[Ahn,ii ]∪e[Bhn,ii ], we have K ⊆ e[Ahn,ii ]∪e[Bhn,ii ]. Finally, e[Ahn,ii ] ∩ e[Bhn,ii ] = ∅, so K is not connected. Now suppose that γX is a (metric) indecomposable compactification of X. By Corollary 4, γX is strongly indecomposable. X also is strongly indecomposable by Lemma 1, thus βX is indecomposable by Theorem 3. If X is a connected space and p, q ∈ X, then X is irreducible between p and q if no proper closed connected subset of X contains both p and q. A continuum is said to be irreducible if it is irreducible between some pair of its points, i.e., if it has more than one composant. Corollary 10. If X is an indecomposable connected space, then βX is indecomposable if and only if X has an irreducible compactification (which is necessarily indecomposable). Proof. If βX is indecomposable, then by Theorem 9 X has a metric indecomposable compactification which necessarily has an uncountable number of composants. Now suppose that βX is decomposable. Let γX be a compactification of X. If ι : X → X is the identity map, then the Stone-Čech extension βι : βX → γX maps proper subcontinua to proper subcontinua and maps onto γX. Thus to prove γX has only one composant, it suffices to show that βX has only one composant. To that end, let p, q ∈ βX. Let H and K be proper subcontinua of βX such that βX = H ∪K. Assume that p ∈ K. If q ∈ K, then p and q are in the same composant. Otherwise, q ∈ U := βX \ K. Since X is indecomposable and U ∩ X 6= X, U ∩ X is not connected, whence U is not connected by Lemma 2. Let T be a proper clopen subset of U with q ∈ T . Then K ∪ T is a proper subcontinuum of βX containing p and q. Again p and q are in the same composant of βX. Suppose that γX is irreducible (between two points p and q); we show that γX is indecomposable. Well, otherwise there is a proper subcontinuum K ⊆ γX with Quasi-components 7 non-empty interior such that p ∈ K. Then βι−1 [K] is a proper closed subset of βX with non-empty interior. Thus, if q 0 ∈ βι−1 {q} then there is a proper subcontinuum L ⊆ βX such that q 0 ∈ L and L ∩ βι−1 [K] 6= ∅ (composants are dense). We know that βX is indecomposable, so L is nowhere dense. Thus L ∪ βι−1 [K] is a proper closed subset of βX, implying that βι[L] ∪ K is a proper subcontinuum of γX containing p and q, a contradiction. Remark. The main result of [2] is that every separable metric nowhere locally compact space densely embeds into the Hilbert space `2 ' (0, 1)ω . Widely-connected Hausdorff spaces have no compact neighborhoods by Theorem 6.2.9 in [5]. Thus, every widely-connected separable metric space has a compactification equal to the Hilbert cube [0, 1]ω , a decomposable continuum with exactly one composant. Observe that a widely-connected space is a connected space that is irreducible between every two of its points. By Theorem 9 and Corollary 10, Question 1(b) is equivalent to the following. Question 1(b)0 . If W is a connected separable metric space that is irreducible between every two of its points, then does W have an irreducible compactification? 3 3.1 Examples in R3 Destroying indecomposability with one limit point Let K and K0 be as defined in Section 3.2. Wojciech Dębski [3] constructed a closed set A ( K such that K \ A is connected and A intersects every composant of K. In [8], we constructed a completely metrizable widely-connected set W by removing ω-many copies of Dębski’s set from K. Assume that A is one of the copies in K0 \W , c := W ∪ A to be on the surface of the unit sphere S2 ⊆ (R3 , d). and consider W c → R3 be the mapping x 7→ inf{d(x, y) : y ∈ A} · x; q shrinks A Let q : W to the single point at the origin h0, 0, 0i. By the construction of W , A intersects c ∩ K0 (see Section 6 of [8]), so q[W c ∩ K0 ] is connected. every quasi-component of W c ] is not indecomposable. Meanwhile, compactness of A implies that Therefore q[W q W is a homeomorphism. In summary q[W ] is a widely-connected subset of R3 that fails to be indecomposable when a single limit point is added to it. 3.2 A composant-locked widely-connected set In this section we show that W may not have a compactification that is irreducible between two points in W . This provides a complete negative answer to Question 1(c) (see the statement of Theorem 14). By Corollary 10, a positive answer to any 8 D. Lipham part of Question 1(c) would have implied a proof of Bellamy’s conjecture (a positive answer to Question 1(b)). Our counterexample demonstrates that a different line of attack is needed. Let e ∈ 2Z be the point defined by concatenating all finite binary sequences in S the positive and negative directions of Z. That is, if {bi : i ∈ ω} enumerates n∈ω 2n , then let e Z+ = b0_ b1_ b2_ ..., and define e(n) = e(−n) for n ∈ Z− . Let η : 2Z → 2Z be the shift map η(x)(n) = x(n + 1); clearly η is a homeomorphism. The backward and forward orbits of e, E0 = {η n (e) : n ∈ Z− } and E1 = {η n (e) : n ∈ Z+ \ {0}}, are dense in 2Z by construction. Finally, E0 ∩ E1 = ∅. For if n, m ∈ ω such that η −n (e) = η m (e), then η n+m (e) = e. Density of E1 implies that e is not a periodic point. So n + m = 0, whence n = m = 0. We have η −n (e) = η m (e) = e ∈ E0 \ E1 . Figure 2 shows a rectilinear version of the bucket-handle continuum K ⊆ [0, 1]2 minus the diagonal Cantor set ∆ = {hx, xi ∈ K : x ∈ [0, 1]}. K \ ∆ is the union of two open sets K0 = {hx, yi ∈ K : x < y} and K1 = {hx, yi ∈ K : x > y}. K0 K0 K1 K3 Figure 2: K \ ∆ = K0 ∪ K1 Let X be any widely-connected subset of K (see [11] and [8] for examples), and for each i < 2 put Wi = (X ∩Ki )∪∆. Note that each Wi is hereditarily disconnected, but W0 ∪ W1 = X ∪ ∆ is connected since X is connected and (necessarily) dense in K. Define W ⊆ 2Z × K by [ W = (Ei × Wi ); i<2 W is hereditarily disconnected, but it will become connected when certain pairs of its points are identified. Let p, q ∈ ∆ be such that K is irreducible between p and q. Those familiar with the Cantor set and the composants of K may choose p = h1, 1i and q = h 41 , 14 i, for instance. Define a relation on 2Z × K by ∼ = hx, pi, hη(x), qi : f = W/ ∼ . x ∈ 2Z , and finally, put W K2 Quasi-components 9 Remark. Recall that the quotient of a compact metric space is metrizable whenever Hausdorff. The entire quotient (2Z × K)/ ∼ is easily seen to be Hausdorff, and is therefore metrizable (and compact). Together with dim((2Z × K)/ ∼) = 1, this implies via the Menger-Nöbeling Theorem (1.11.4 in [4]) that (2Z × K)/ ∼ embeds f also embeds into R3 . into Euclidean 3-space R3 , and so of course W Lemma 11. Let X be a space. If U ⊆ X is open, A0 is clopen in U , A1 is clopen in X \ U , and A0 ∩ ∂U = A1 ∩ ∂U , then A0 ∪ A1 is clopen in X. Proof. Clearly A0 ∪ A1 is closed in X and (A0 ∪ A1 ) \ ∂U is open in X. We show that A0 ∪ A1 is open in X by showing that it is an X-neighborhood of each of its points in ∂U . Suppose that a ∈ (A0 ∪ A1 ) ∩ ∂U . Then a ∈ A0 ∩ A1 . There are two open subsets of X, V0 and V1 , such that a ∈ V0 ∩ U ⊆ A0 and a ∈ V1 ∩ (X \ U ) ⊆ A1 . Then a ∈ V0 ∩ V1 ⊆ A0 ∪ A1 ; to prove the inclusion, suppose x ∈ V0 ∩ V1 and consider the two cases x ∈ U and x ∈ X \ U . Lemma 12. For each i < 2, {0} × Wi is a quasi-component of [ Y := ({0} × Wi ) ∪ ({1/n} × W1−i ). n≥1 Proof. Fix i < 2. Clearly {0} × Wi contains a quasi-component of Y . Now let A be a clopen subset of Y such that A∩{0}×Wi 6= ∅; we show that {0}×Wi ⊆ A. For each a ∈ A ∩ ∆ there is an integer n(a) and an open U (a) ⊆ ∆ such that a ∈ [0, 1/n(a)] × U (a) ⊆ A (by [0, 1/n] we mean {0}∪{1/k : k ≥ n}). By compactness of A∩∆, there S is a finite F ⊆ A such that A ∩ ∆ ⊆ {[0, 1/n(a)] × U (a) : a ∈ F }. Similarly, there is a finite E ⊆ Y \ A and a collection of open sets {[0, 1/n(b)] × U (b) : b ∈ E} such S that (Y \ A) ∩ ∆ ⊆ {[0, 1/n(b)] × U (b) : b ∈ E}. Let m = max{n(y) : y ∈ F ∪ E}. Let A0 = A∩({0}×Wi ) and A1 = π[A∩({1/m}×W1−i )], where π : R3 → {0}×R2 is the projection onto the y-z-plane. A0 ∩ ({0} × ∆) = A1 ∩ ({0} × ∆) by the choice of m. Applying Lemma 11 with U = {0} × (Wi \ ∆), we have A0 ∪ A1 is clopen in {0} × (W0 ∪ W1 ). Since {0} × (W0 ∪ W1 ) = {0} × (X ∪ ∆) is connected, A0 ∪ A1 = {0} × (W0 ∪ W1 ). Thus {0} × Wi ⊆ A0 ⊆ A. f is widely-connected. Theorem 13. W Proof. For each i < 2 and e0 ∈ Ei , {e0 }×Wi is a quasi-component of W . This follows immediately from Lemma 12 since there is a sequence of points (en ) ∈ (E1−i )ω such that en → e0 as n → ∞. Thus, the relation ∼ links together the quasi-components f is connected. of W , so that W f is widely-connected, we let C be a non-empty connected subset To prove that W f with C 6= W f , and show that |C| = 1. of W 10 D. Lipham Let U × V be a non-empty open subset of 2Z × K such that {p, q} ∩ V = ∅ and C ∩ (U × V ) = ∅. Let x ∈ C. Let x0 ∈ W such that x = x0 / ∼, and let e0 be the first coordinate of x0 . Since E0 and E1 are dense in 2Z , there exist n, m > 0 such that {η −n (e0 ), η m (e0 )} ⊆ U . Since η n+m is a homeomorphism and 2Z has a basis of clopen sets, there is a clopen A ⊆ U such that η −n (e0 ) ∈ A and η n+m [A] ⊆ U . Let L and M be disjoint closed subsets of K such that K \ V = L ∪ M , p ∈ L, and q ∈ M . Then [ T := W \ (U × V ) ∩ (A × L) ∪ (η n+m [A] × M ) ∪ (η i [A] × K) ∼ 0<i<n+m f \ (U × V ). is a clopen subset of W Note that x ∈ T since e0 ∈ η n [A] and 0 < n < n + m. So C ⊆ T and x ∈ Q := T ∩ η i (e0 ) : −n ≤ i ≤ m × K / ∼ . If y ∈ T \ Q then by the construction of T there is a relatively clopen S ⊆ T such that Q ⊆ S ⊆ T \ {y}. Thus C ⊆ Q, which implies |C| = 1 (see Figure 3). p q Figure 3: Superset of Q if n = m = 2 and e0 = e f is a compactification of W f , then there is a composant P of Theorem 14. If γ W f such that W f ⊆ P. γW f there is a non-empty Proof. It suffices to show that for every two points x, y ∈ W f such that x and y are in the same quasi-component of W f \ T . This open T ⊆ W f is a compactification of W f and T 0 is open in γ W f such that would imply that if γ W f = T , then x and y are contained in a quasi-component of γ W f \ T 0 . By T0 ∩ W f \ T 0 , a proper Theorem 6.1.23 in [5], x and y are contained in a component of γ W f. subcontinuum of γ W Quasi-components 11 Let x and y be given. Pick two points x0 , y 0 ∈ W such that x = x0 / ∼ and y = y 0 / ∼, and let n, m ∈ Z such that η n (e) and η m (e) are the first coordinates of x0 and y 0 , respectively. Assume that n ≤ m. There is a non-empty clopen A ⊆ 2Z such that A ∩ {η i (e) : n − 1 ≤ i ≤ m + 1} = ∅. For i ∈ Z, set δ(i) = 0 if i ≤ 0 and δ(i) = 1 if i > 0. Then [ i R := η (e) × Wδ(i) ∼ n≤i≤m f \ T where T = (A × K)/ ∼. And {x, y} ⊆ R. Each fiber {η i (e)} × is a subset of W Wδ(i) , n ≤ i ≤ m, is a quasi-component of W \ (A × K) by density of E1−δ(i) in 2Z \ A (Lemma 12). These fibers are linked together by ∼, so that R is contained in f \ T. a quasi-component of W To make further progress on Question 1(a)-(b), one may try to construct a T1 or T2 widely-connected space that is not strongly indecomposable, or prove that βW indecomposable when W is a (widely-connected) subset of the plane. Now a question of our own: Question 2. Let P be an indecomposable metric space which is the union of an increasing collection of countably many continua. (a) Is P strongly indecomposable? (b) Is there a metric continuum with a composant equal to P ? References [1] D. Bellamy, Questions in and out of context, Open Problems in Topology II, Elsevier (2007) 259-262. [2] P. Bowers, Dense embeddings of nowhere locally compact separable metric spaces, Topology Appl. 26 (1987) 1-12. [3] W. Dębski, Note on a problem by H. Cook. Houston J. Math. 17 (1991) 439441. [4] R. Engelking, Dimension Theory, Volume 19 of Mathematical Studies, NorthHolland Publishing Company, 1978. [5] , General Topology, Revised and completed edition Sigma Series in Pure Mathematics 6, Heldermann Verlag, Berlin, 1989. 12 D. Lipham [6] G. Gruenhage, Spaces in Which the Nondegenerate Connected Sets Are the Cofinite Sets. Proceedings of the American Mathematical Society Vol. 122 No. 3 (1994) 911-924. [7] K. Kuratowski, Topology Vol. II, Academic Press, 1968. [8] D. Lipham, Widely-connected sets in the bucket-handle continuum, Fundamenta Math., to appear. [9] J. Mioduszewski, Spring 2004 Conference Report, Topology Proceedings, Volume 28 (2004) 301-326. [10] M.E. Rudin, A property of indecomposable connected sets, Proc. Amer. Math. Soc. 8 (1957) 1152-1157. [11] P.M. Swingle, Two types of connected sets, Bull. Amer. Math. Soc. 37 (1931) 254-258. [12] R.C. Walker, The Stone-Čech Compactification, Springer, New York-Berlin, 1974.
© Copyright 2026 Paperzz