BuzyakovaAGT.dvi @ Applied General Topology c© Universidad Politécnica de Valencia Volume 10, No. 2, 2009 pp. 221-225 On an algebraic version of Tamano’s theorem Raushan Z. Buzyakova Abstract. Let X be a non-paracompact subspace of a linearly or- dered topological space. We prove, in particular, that if a Hausdorff topological group G contains closed copies of X and a Hausdorff com- pactification bX of X then G is not normal. The theorem also holds in the class of monotonically normal spaces. 2000 AMS Classification: 54H11,22A05, 54F05 Keywords: topological group, Hausdorff compactification, normal space, stationary set 1. Introduction. This note is devoted to analysis of Tamano’s characterization [5] of para- compactness in the context of Hausdorff topological groups. The Tamano’s argument implies that if a Tychonov space X is not paracompact then X × bX is not normal for every Hausdorff compactification bX of X. A natural alge- braic analysis of this statement leads to the following conjecture: Conjecture. Let X be a non-paracompact topological space and bX a Haus- dorff compactification of X. If a topological group G contains closed copies of X and bX then G is not normal. We believe that the conjecture has a good chance for a positive resolution. In this note we give a proof of this conjecture in the class of generalized ordered spaces (=”subspaces of linearly ordered spaces”), or more generally, in the class of monotonically normal spaces. Since Tamano’s theorem is a criterion it would be natural to ask if given a paracompact space X one can find a Hausdorff compactification bX and a normal group G such that G contains closed copies of X and bX. The author does not know if such G and bX exist without additional requirements on X besides paracompactness. It is worth to mention, however, that if X n is Lindelöf for every n ∈ ω then the free group F (X ⊕bX) over X ⊕bX is normal (even Lindelöf) and contains closed copies of 222 R. Z. Buzyakova X and bX. This fact is well-known and mentioned, in particular, in the recent survey [4]. Also, we would like to mention that in [2] the author proved that if a group G contains closed copies of an uncountable regular cardinal τ and τ + 1, then G contains a closed copy of τ × (τ + 1), which makes G not normal. While some of the ideas of this result could be used to prove the main theorem of this note we use a different approach, which may be helpful in proving the general conjecture. All spaces in this note are assumed to be Tychonov. By βX we denote the Čech-Stone compactification of X. The symbol ⋆ is reserved for group operation. A subspace of a linearly ordered topological space will be called a GO-space. A point x ∈ X is a complete accumulation point for an infinite set A ⊂ X if every open neighborhood of x meets A by a subset of cardinality |A|. To prove our main result, we start with four folklore statements, two of which are left without proof. Fact 1. Let S be a stationary subset of an uncountable regular cardinal τ and bS a Hausdorff compactification of S. Then there exists a unique point p ∈ bS \ S that is a complete accumulation point for S. Moreover, S ∪ {p} is naturally homeomorphic to S ∪ {τ}. The point p in the above fact will be always identified with τ . Fact 2. Let S be a stationary subset of an uncountable regular cardinal τ , bS a Hausdorff compactification of S, and c(S × bS) a Hausdorff compactification of S × bS. Then there exists a unique point p ∈ c(S × bS) that is the only common and only complete accumulation point for S × {τ} and for {(α, α) : α ∈ S}. The point p in Fact 2 will be always identified with (τ, τ ). Lemma 1.1. Let S be a stationary subset of an uncountable regular cardinal τ . Let τ be a limit point for A ⊂ βS \ (S ∪ {τ}) in βS. Then ClβS (A) ∩ S is closed and unbounded in S. Proof. Let f be the continuous map from βS to τ + 1 that is the identity on S. Since τ is the only complete accumulation point for S in βS, f (A) ⊂ τ . Since τ is a limit point for A, f (A) is unbounded in τ . Assume the conclusion of Lemma is false. Then we may also assume that ClβS(A) ∩ S = ∅. Since f maps the remainder of S in βS to the remainder of S in τ + 1, we have f (ClβS(A) \ {τ}) is a closed unbounded subset of τ that does not meet S. This contradicts stationarity of S in τ . � Lemma 1.2. Let S be a stationary subset of an uncountable regular cardinal τ . If f : S → τ is continuous and unbounded then there exists λ ∈ S such that f (λ) = λ. On an algebraic version of Tamano’s theorem 223 Proof. Since f is unbounded and τ is regular we can select X = {xβ : β < τ} such that (1) xα > max{xβ , f (xβ )} if α > β; (2) f (xα) > max{xβ , f (xβ )} if α > β. Observe that property 1 and regularity of τ imply that X is unbounded in τ . Since S is stationary there exist λ ∈ S and limit α ∈ τ such that λ is limit for {xβ : β < α} and xβ < λ for all β < α. By 1 and 2 and continuity of f , we have f (λ) = λ. � For our main result we need the following fundamental theorem. Theorem (R. Engelking and D. Lutzer [3]). A GO-space X is paracompact iff no closed subspace of X is homeomorphic to a stationary subset of a regular uncountable cardinal. Theorem 1.3. Let L be a non-paracompact GO-space and bL a Hausdorff compactification of L. If a topological group G contains closed copies of L and bL, then G is not normal. Proof. We may assume that L is a closed subset of G and bL′ ⊂ G is a copy of bL, where L′ is a copy of L with a fixed homeomorphism x ↔ x′. Let S be a closed subset of L that is homeomorphic to a stationary subset of an uncountable regular cardinal τ . Such an S exists due to Theorem’s hypothesis and Engelking-Lutzer theorem. As agreed earlier, by τ we denote the only complete accumulation point for S in any Hausdorff compactification and by τ ′ the only complete accumulation point for S′ in bL′. Let H = S × {τ ′} and D = {(α, α′) : α ∈ S}. The sets H and D are closed in S × bS′ and not functionally separated. Let HG = ⋆(H) = {α ⋆ τ ′ : α ∈ S} and DG = ⋆(D) = {α ⋆ α ′ : α ∈ S}. Claim 1: ⋆̃(τ, τ ′) 6∈ G, where ⋆̃ is the continuous extension of ⋆ over the Čech- Stone compactification. To prove the claim, observe that HG is a closed subset of G homeomor- phic to S. This is because multiplication by a constant is a continuous automorphism. By Fact 2 and Fact 1, (τ, τ ′) is the only complete ac- cumulation point for H in β(G × G)). The set HG does not have a complete accumulation point in G. Therefore ⋆̃(τ, τ ′) 6∈ G. The claim is proved. Put Hα = {(β, τ ′) : β ≥ α, β ∈ S}, HαG = {β ⋆ τ ′ : β ≥ α, β ∈ S}, Dα = {(β, β′) : β ≥ α, β ∈ S}, and DαG = {β ⋆ β ′ : β ≥ α, β ∈ S}. 224 R. Z. Buzyakova Claim 2: There exists λ < τ such that HλG ∩ ClG(D λ G) = ∅. To prove the claim assume the contrary. Then for any α < τ there exists pα ∈ Clβ(G×G)(D α) such that ⋆̃(pα) ∈ H α G. By Lemma 1.1, Clβ(G×G){pα : α < τ} meets D by a closed subset T of cardinality τ . Since HG is closed and ⋆ is continuous we have ⋆(T ) ⊂ HG. Since |T | = τ we have ⋆̃(τ, τ ′) is a complete accumulation point for ⋆(T ) in βG. By Lemma 1.2, there exists (γ, γ′) ∈ T such that ⋆(γ, γ′) = γ ⋆ γ′ = γ ⋆ τ ′. Therefore, γ′ = τ ′, contradicting to the fact that (τ, τ ′) 6∈ T . The claim is proved. By Claim 2, HλG and ClG(D λ G) are closed and disjoint in G. If G were normal, then HλG and ClG(D λ G) would have been functionally separated and so would Hλ and Dλ in G × G. But Hλ and Dλ are not functionally separated for every λ < τ . � Observe that the proof of the theorem uses only one property of L, namely, the fact that L contains a closed copy of a stationary subset of an uncountable regular cardinal τ . Since the theorem of Engelking and Lutzer holds for mono- tonically normal spaces as well (proved by Balogh and Rudin [1]) we have the following. Theorem 1.4. Let X be non-paracompact and monotonically normal and bX a Hausdorff compactification of X. If a topological group G contains closed copies of X and bX, then G is not normal. We would like to finish the paper with two questions (which may have been asked before by other authors) related to the discussion in the beginning of this work. Question 1. Is there a paracompact space that cannot be embedded in a normal group as a closed subspace? Question 2. Let X n be paracompact for every n ∈ ω. Is F (X) normal? References [1] Z. Balogh and M. E. Rudin, Monotone normality, Topology Appl. 47 (1992), no. 2, 115–127. [2] R. Z. Buzyakova, Ordinals in topological groups, Fund. Math. 196 (2007), no. 2, 127–138. [3] D. J. Lutzer, Ordered topological spaces, Surveys in general topology, pp. 247–295, Aca- demic Press, New York-London-Toronto, Ont., 1980. [4] O. V. Sipachëva, The topology of free topological group, Fundam. Prikl. Mat. 9 (2003), no. 2, 99–204; English translation in J. Math. Sci. (N. Y.) 131 (2005), no. 4, 5765–5838. [5] H. Tamano, On paracompactness, Pacific J. Math. 10 (1960) 1043–1047. On an algebraic version of Tamano’s theorem 225 Received April 2008 Accepted February 2009 Raushan Z. Buzyakova (Raushan Buzyakova@yahoo.com) Mathematics Department, UNCG, Greensboro, NC 27402 USA