The example 106 from Counterexamples in Topology by Steen and Seebach [3] is the space where the first factor is the space of countable ordinals with the usual order topology and the second factor is the product of continuum many copies of the unit interval .
This space was previously discussed in this site. One of the key results from that discussion is that is not normal, a result not shown in Steen and Seebach. The proof that was given in this site (see here) is based on an article published in 1976 [1], long before the publication date of the first edition of Steen and Seebach in 1970. It turns out that the non-normality of was given as an exercise in Steen and Seebach in the problem section at the end of the book (problem 127 in page 211, Dover edition). Problem 127: Show that is not normal. This indicates that the result not shown in Steen and Seebach was because it was given as a problem and not because the tool for solving it was not yet available. The fact that it is given as an exercise also means that there is a more basic proof of the non-normality of . So, once this is realized, I set out to find a simpler proof or at least one that does not rely on the result from [1]. Interestingly, this proof brings out a broader discussion that is worthwhile and goes beyond the example at hand. The goal here is to examine the more basic proof and the broader discussion.
A Classic Example
Before talking about the promised proof, we consider the product of and its immediate successor.
As noted at the beginning, the space is the set of all countable ordinals with the order topology. The ordinal is the immediate successor of . It can be regarded as the result of adding one more point to . The extra point is , i.e., with greater than all points . The ordinal with the order topology is a compact space. Using interval notation, and . As ordinals, is the first uncountable ordinal and is the first uncountable successor ordinal. For more information, see here.
The product is a classic example of a product of a normal space (the first factor) and a compact space (the second factor) that is not normal. This example and others like it show that normality is easily broken upon taking product even if one of the factors is as nice as a compact space. The non-normality of is discussed here. In that proof, two disjoint closed sets and are given such that they cannot be separated by disjoint open sets. The and are:
The Basic Proof
To show that is not normal, we show that one of its closed subspaces is not normal. That closed subspace is . To this end, we show that can be embedded in the product space . With a non-normal closed subspace, it follows that is not normal. The remainder of the proof is to give the embedding.
We show that can be embedded as a closed subspace of , the product of many copies of . This means that is also a closed subspace of .
For each , define as follows:
Furthermore, define by letting for all . Consider the correspondence with and . The mapping is clearly one-to-one from onto . Upon closer inspection, the mapping in each direction is continuous (this is a good exercise to walk through). Thus, the mapping is a homeomorphism. It follows that can be considered a subspace of . Since is compact, it must be a closed subspace. With the cardinality of being less than or equal to continuum, it follows that can be embedded as a closed subspace of .
Stone-Cech Compactification
The first broader discussion is that of Stone-Cech compactification. More specifically, , i.e., the Stone-Cech compactification of the first uncountable ordinal is its immediate successor.
To see that , note that every continuous function defined on is bounded and is eventually constant (see result B here). As a result, every continuous function defined on can be extended to a continuous function defined on . For any continuous function , we can simply define to be the eventual constant value. A subspace of a space is -embedded in if every bounded continuous real-valued function on can be extended to . According to theorem 19.12 in [4], if is a compactification of and if is -embedded in , then is the Stone-Cech compactification of . Thus is -embedded in and is the Stone-Cech compactification of . In this instance, the Stone-Cech compactification agrees with the one-point compactification. Consider the following class theorem about normality in product space. The theorem is Corollary 3.4 in the chapter on products of normal spaces in the handbook of set-theoretic topology [2].
Theorem 1
Let be a space. The following conditions are equivalent.
- The space is paracompact.
- The product space is normal.
Based on the discussion presented above, the non-normality of is due to the non-normality of . Based on this theorem, the non-normality of is due to the non-paracompactness of . See result G here for a proof that is not paracompact.
The discussion up to this point points to two ways to prove that is not normal. One way is the basic proof indicated above. The other way is to use Theorem 1, along with the homeomorphic embedding from into , the fact that and the fact that is not paracompact. Both are valuable. The first way is basic and is a constructive proof. Because it is more hands-on, it is a better proof to learn from. The second way provides a broader perspective that is informative but requires quoting a couple of fairly deep results. Perhaps it is best used as a second proof for perspective.
Countable Tightness
The essence of the basic proof above goes like this: if the space contains a copy of , then the product space is not normal. The contrapositive statement would be the following:
Corollary
Let be a space. If the product space is normal, then cannot contain a copy of .
In the space of , note the following about the last point: but for any countable , i.e., the last point is the limit point of the set of all the points preceding it but is not in the closure of any countable set. This means that the space does not have countable tightness (or is not countably tight). See here for definition. The property of countable tightness is hereditary. If contains a copy of , then is not countably tight (or is uncountably tight). This brings us to the following theorem.
Theorem 2
Let be an infinite compact space. Then is normal if and only if has countable tightness.
Whenever we consider the normality of a product with the first factor being and the second factor being a compact space, the real story is the tightness of that compact space. If the tightness is countable, the product is normal. Otherwise, the product is not normal. The theorem is another reason that is not normal. Instead of embedding into , we can actually show that does not have countable tightness. This is the approach that was taken in this previous post.
Theorem 2 is the result from 1976 alluded to earlier [1]. A proof of Theorem 2 is found in this previous post. For results concerning normality in a product space with a compact factor (the other factor does not have to be ), see the chapter on products of normal spaces in the handbook of set-theoretic topology [2].
Reference
- Nogura, T., Tightness of compact Hausdorff space and normality of product spaces, J. Math. Soc. Japan, 28, 360-362, 1976.
- Przymusinski T. C., Products of Normal Spaces, Handbook of Set-Theoretic Topology (K. Kunen and J. E. Vaughan, eds), Elsevier Science Publishers B. V., Amsterdam, 781-826, 1984.
- Steen, L. A., Seebach, J. A., Counterexamples in Topology, Dover Publications, Inc., New York, 1995.
- Willard, S., General Topology, Addison-Wesley Publishing Company, 1970.
Dan Ma topology
Daniel Ma topology
Dan Ma math
Daniel Ma mathematics
2021 – Dan Ma