The Cichon’s Diagram

The Cichon’s Diagram is a diagram that shows the relationships among ten small cardinals – four cardinals associated with the \sigma-ideal of sets of Lebesgue measure zero, four cardinals associated with the \sigma-ideal of sets of meager sets, the bounding number \mathfrak{b}, and the dominating number \mathfrak{d}. What makes this interesting is that elements of analysis, topology and set theory flow into the same spot. Here’s the diagram.

Figure 1 – The Cichon’s Diagram

In this diagram, \alpha \rightarrow \beta means \alpha \le \beta. The preceding three posts (the first post, the second post and the third post) give the necessary definitions and background to understand the diagram. In addition to the above diagram, the following relationships also hold.

Figure 2 – The Cichon’s Diagram – Additional Relationships

The Cardinal Characteristics of a \sigma -Ideal

For any \sigma-ideal \mathcal{I} on a set X, there are four associated cardinals – \text{add}(\mathcal{I}), \text{non}(\mathcal{I}), \text{cov}(\mathcal{I}) and \text{cof}(\mathcal{I}). The first one is the additivity number, which is the least number of elements of \mathcal{I} whose union is not an element of \mathcal{I}. The second cardinal is called the uniformity number, which is the least cardinality of a subset of X that is not an element of \mathcal{I}. The third cardinal is called the covering number, which is the least cardinality of a subfamily of \mathcal{I} that is also a covering of X. The fourth cardinal is called the cofinality number, which is the least cardinality of a subfamily of \mathcal{I} that is cofinal in \mathcal{I}. For more information, see the first post. The four cardinals are related in a way that is depicted in the following diagram. Again, \alpha \Rightarrow \beta means \alpha \le \beta.

Figure 3 – Cardinal Characteristics of a \sigma -Ideal

…Cichon…\displaystyle \begin{array}{ccccccccccccc} \text{ } & \text{ } & \text{ } &\text{ } & \bold n \bold o \bold n ( \mathcal{I} ) &\text{ } &\Longrightarrow &\text{ } &\bold c \bold o \bold f (\mathcal{I} )&\text{ } &\Longrightarrow & \text{ } & \lvert \mathcal{I} \lvert\\  \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ } & \text{ } \\     \text{ } & \text{ } & \text{ } &\text{ } & \Uparrow &\text{ } &\text{ } &\text{ } &\Uparrow&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ }& \text{ }  & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \aleph_1 & \text{ } & \Longrightarrow & \text{ } &\bold a \bold d \bold d ( \mathcal{I} ) & \text{ } &\Longrightarrow & \text{ } &\bold c \bold o \bold v ( \mathcal{I} ) &\text{ } &\text{ } &\text{ }  & \text{ }    \end{array}

Figure 3 explains the basic orientation of the Cichon’s Diagram. Filling it with three \sigma-ideals produces the Cichon’s Diagram.

The Three \sigma -Ideals in the Cichon’s Diagram

Let \mathcal{K} be the \sigma-ideal of bounded subsets of \omega^\omega. It is known that \mathfrak{b}=\text{add}(\mathcal{K})=\text{non}(\mathcal{K}) (this is called the bounding number) and \mathfrak{d}=\text{cov}(\mathcal{K})=\text{cof}(\mathcal{K}) (this is called the dominating number). The ideal \mathcal{K} is discussed in this previous post. Let \mathcal{M} be the \sigma-ideal of meager subsets of the real line \mathbb{R} (this is discussed in this previous post). Let \mathcal{L} be the \sigma-ideal of Lebesgue measure zero subsets of the real line.

Thus the Cichon’s Diagram (Figure 1 above) houses information about three \sigma-ideals. The two numbers for the \sigma-ideal \mathcal{K} are situated in the middle of the diagram (\mathfrak{b} and \mathfrak{d}). The four numbers for the \sigma-ideal \mathcal{M} are situated in the center portion of the diagram. The four numbers for the \sigma-ideal \mathcal{L} are located on the left side and the right side. The Cichon’s Diagram (Figure 1) is flanked by \aleph_1 on the lower left and by continnum 2^{\aleph_0} on the upper right.

More on the Cichon’s Diagram

One interesting aspect of the Cichon’s Diagram: it is a small diagram with small cardinals where elements of analysis (measure) and topology (category) come together. The following diagram shows the path that includes both the bounding number and the dominating number.

Figure 4 – The Cichon’s Diagram – The Main Path

The path circled in the above diagram involves all three \sigma-ideals. It is also one of the longest increasing paths in the diagram.

    \aleph_1 \le \text{add}(\mathcal{L}) \le \text{add}(\mathcal{M}) \le \mathfrak{b} \le \mathfrak{d} \le \text{cof}(\mathcal{M}) \le \text{cof}(\mathcal{L}) \le 2^{\aleph_0}

There are fifteen arrows in Figure 1. The proofs of these arrows (or inequalities) require varying degrees of effort. Three are basic information – \aleph_1 \le \text{add}(\mathcal{L}), \mathfrak{b} \le \mathfrak{d} and \text{cof}(\mathcal{L}) \le 2^{\aleph_0}. Because \mathcal{L} is a \sigma-ideal, its additivity number must be uncountable. By definition, \mathfrak{b} \le \mathfrak{d}. The \sigma-ideal \mathcal{L} has a cofinal subfamily consisting of Borel sets. Thus \text{cof}(\mathcal{L}) \le 2^{\aleph_0}.

Four of the arrows follow from the relative magnitude of the four cardinals of a \sigma-ideal as shown in Figure 3 – \text{add}(\mathcal{L}) \le \text{cov}(\mathcal{L}), \text{non}(\mathcal{L}) \le \text{cof}(\mathcal{L}), \text{add}(\mathcal{M}) \le \text{cov}(\mathcal{M}) and \text{non}(\mathcal{M}) \le \text{cof}(\mathcal{M}).

Three of the arrows are proved in this previous post\mathfrak{b} \le \text{non}(\mathcal{M}), \mathfrak{d} \le \text{cov}(\mathcal{M}) and \text{add}(\mathcal{M}) \le \mathfrak{b}. The last inequality follows from this fact: if F \subset \omega^\omega is an unbounded set, then there exist \lvert F \lvert many meager subsets of the real line whose union is a non-meager set, essentially a result in Miller [8].

The proofs of the remaining five arrows can be found in [3] – \mathfrak{d} \le \text{cof}(\mathcal{M}), \text{add}(\mathcal{L}) \le \text{add}(\mathcal{M}), \text{cov}(\mathcal{L}) \le \text{non}(\mathcal{M}), \text{cof}(\mathcal{M}) \le \text{cof}(\mathcal{L}) and \text{cov}(\mathcal{M}) \le \text{non}(\mathcal{L}). The proofs of two additional relationships displayed in Figure 2 can also be found in [3].

The fifteen arrows in the Cichon’s Diagram represent the only inequalities among the ten cardinals (not counting \aleph_1 and 2^{\aleph_0}) that are provable in ZFC [1] and [5]. As illustration, we give an example of non-ZFC provable relation in the next section.

An Example of an Inequality Not Provable in ZFC

In the following diagram, the cardinals \mathfrak{b} and \text{cov}(\mathcal{M}) are encircled. These two numbers are not connected by arrows.

Figure 5 – The Cichon’s Diagram – An Example of Non-ZFC Provable

We sketch out a proof that no inequalities can be established between \mathfrak{b} and \text{cov}(\mathcal{M}). First Martin’s Axiom (MA) implies that \mathfrak{b} \le \text{cov}(\mathcal{M}). Topologically, the statement MA (\kappa) means that any compact Hausdorff space X that satisfies the countable chain condition cannot be the union of \kappa or fewer many nowhere dense sets. The Martin’s Axiom (MA) is the statement that MA (\kappa) holds for all \kappa less than 2^{\aleph_0}. It follows that MA implies that \text{cov}(\mathcal{M}) cannot be less than 2^{\aleph_0} and thus \text{cov}(\mathcal{M})=2^{\aleph_0}. It is always the case that the bounding number \mathfrak{b} is \le 2^{\aleph_0}.

On the other hand, in Laver’s model [6] for the Borel conjecture, \mathfrak{b} > \text{cov}(\mathcal{M}). In Laver’s model, every subset of the real line that is of strong measure zero is countable. Since any set with the Rothberger property is of strong measure zero, every subset of the real line that has the Rothberger property is countable in Laver’s model. Let \text{non}(\text{Rothberger}) be the least cardinality of a subset of the real line that does not have the Rothberger property. Thus in Laver’s model, \text{non}(\text{Rothberger})=\aleph_1. It is well known that \text{non}(\text{Rothberger})=\text{cov}(\mathcal{M}); see Theorem 5 in [10]. Thus in Laver’s model, \text{cov}(\mathcal{M})=\aleph_1.

In Laver’s model, \mathfrak{b} > \aleph_1. Note that \mathfrak{b}= \aleph_1 implies that there is an uncountable subset of the real line that is concentrated about \mathbb{Q}, the set of all rational numbers; see Theorem 10.2 in [12]. Any concentrated set is of strong measure zero; see Theorem 3.1 in [9]. Thus it must be the case that \mathfrak{b} > \aleph_1=\text{cov}(\mathcal{M}) in Laver’s model.

Remarks

The Cichon’s Diagram is a remarkable diagram. It blends elements of analysis and topology into a small diagram. The fifteen arrows shown in the diagram are obviously far from the end of the story. The Cichon’s Diagram had been around for a long time. Much had been written about it. The article [13] posted some questions about the diagram. See [1], [2], [4] and [11] for further information on the cardinals in the diagram.

Reference

  1. Bartoszynski, T., Judah H., Shelah S.,The Cichon Diagram, J. Symbolic Logic, 58(2), 401-423, 1993.
  2. Bartoszynski, T., Judah H., Shelah S.,Set theory: On the structure of the real line, A K
    Peters, Ltd.. Wellesley, MA, 1995.
  3. Blass, A., Combinatorial Cardinal Characteristics of the Continuum, Handbook of Set Theory (M. Foreman, A. Kanamori, eds), Springer Science+Business Media B. V., Netherlands, 395-489, 2010.
  4. Fremlin, D. H., Cichon’s diagram. In Seminaire d’Initiation ´a l’Analyse, 23, Universite Pierre et Marie Curie, Paris, 1984.
  5. Garcia, H., da Silva S. G., Identifying Small with Bounded: Unboundedness, Domination, Ideals and Their Cardinal Invariants, South American Journal of Logic, 2 (2), 425-436, 2016.
  6. Laver, R., On the consistency of Borel’s conjecture, Acta Math., 137, 151-169, 1976.
  7. Miller, A. W., Some Properties of Measure and Category, Trans. Amer. Math. Soc., 266 (1), 93-114, 1981.
  8. Miller, A. W., A Characterization of the Least Cardinal for Which the Baire Category Theorem Fails, Proc. Amer. Math. Soc., 86 (3), 498-502, 1982.
  9. Miller, A. W., Special Subsets of the Real Line, Handbook of Set-Theoretic Topology (K. Kunen and J. E. Vaughan, eds), Elsevier Science Publishers B. V., Amsterdam, 201-233, 1984.
  10. Miller A. W., Fremlin D. H., On some properties of Hurewicz, Menger, and Rothberger, Fund. Math., 129, 17-33, 1988.
  11. Pawlikowski, J., Reclaw I., Parametrized Cichon’s diagram and small sets, Fund. Math., 127, 225-239, 1987.
  12. Van Douwen, E. K., The Integers and Topology, Handbook of Set-Theoretic Topology (K. Kunen and J. E. Vaughan, eds), Elsevier Science Publishers B. V., Amsterdam, 111-167, 1984.
  13. Vaughn, J. E., Small uncountable cardinals and topology, Open Problems in Topology (J. van Mill and G.M. Reed, eds), Elsevier Science Publishers B.V. (North-Holland), 1990.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

The ideal of meager sets

This is the third in a series of four posts leading to a diagram called The Cichon’s Diagram. This post focuses on the \sigma-ideal of meager subsets of the real line. The links to the previous posts: the first post and the second post.

The next post is: the Cichon’s Diagram.

The notion of meager sets can be defined on any topological space. Let Y be a space. A subset A of Y is a nowhere dense set if \overline{A}, the closure of A in the space Y, contains no open sets. Equivalently, A is a nowhere dense set if for each non-empty open subset U of Y, there exists a non-empty open subset V of U such that V \cap A=\varnothing. We can always find a part of any open set that misses a nowhere dense set. Thus nowhere dense sets are considered “thin” sets. A subset of Y is said to be a meager set if it is the union of countably many nowhere dense sets. A meager set is also called a set of first category. A non-meager set is then called a set of second category.

Though the notion of meager sets can be considered in any space, we would like to focus on the real line \mathbb{R} or the space of all irrational numbers \mathbb{P}. Note that \mathbb{P} is homeomorphic to \omega^\omega (see here). Instead of working with \mathbb{P}, we work with \omega^\omega, which is the product space of countably many copies of the countable discrete space \omega.

\sigma -Ideal of Meager Sets

The notion of meager sets is a topological notion of small sets. The real line and the space of irrationals \omega^\omega are “big” sets. This means that they are not the union of countably many meager sets (this fact is a consequence of the Baire category theorem). Let \mathcal{M} be the set of all subsets of the real line that are meager sets. It is straightforward to verify that \mathcal{M} is a \sigma-ideal on the real line \mathbb{R}. Because of the Baire category theorem, \mathcal{M} is a proper ideal, i.e. \mathbb{R} \notin \mathcal{M}. Naturally, we would like to consider the four cardinals associated with this ideal – \text{add}(\mathcal{M}) (the additivity number), \text{cov}(\mathcal{M}) (the covering number), \text{non}(\mathcal{M}) (the uniformity number) and \text{cof} (\mathcal{M}) (the cofinality number). These four numbers are displayed in the following diagram.

Figure 1 – Cardinal Characteristics of the \sigma -Ideal of Meager Sets

…Cichon…\displaystyle \begin{array}{ccccccccccccc} \text{ } & \text{ } & \text{ } &\text{ } & \bold n \bold o \bold n ( \mathcal{M} ) &\text{ } &\Longrightarrow &\text{ } &\bold c \bold o \bold f (\mathcal{M} )&\text{ } &\Longrightarrow & \text{ } & 2^{\aleph_0}\\  \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ } & \text{ } \\     \text{ } & \text{ } & \text{ } &\text{ } & \Uparrow &\text{ } &\text{ } &\text{ } &\Uparrow&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ }& \text{ }  & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \aleph_1 & \text{ } & \Longrightarrow & \text{ } &\bold a \bold d \bold d ( \mathcal{M} ) & \text{ } &\Longrightarrow & \text{ } &\bold c \bold o \bold v ( \mathcal{M} ) &\text{ } &\text{ } &\text{ }  & \text{ }    \end{array}

In the above diagram, an arrow means \le. So \alpha \Rightarrow \beta means \alpha \le \beta. The inequalities displayed in this diagram always hold for any \sigma-ideal. The only inequality that requires explanation is \text{cof} (\mathcal{M}) \le 2^{\aleph_0}. Any meager set is a subset of an F_\sigma-set. To see this, let A=\bigcup_{n \in \omega} X_n where each X_n is a nowhere dense subset of the real line. Then A \subset \bigcup_{n \in \omega} \overline{X_n}. Each \overline{X_n} is also nowhere dense. Thus the set of all F_\sigma nowhere dense sets is cofinal in \mathcal{M}. This cofinal set has cardinality continuum. Of course, if continuum hypothesis holds (\aleph_1=2^{\aleph_0}), then all four cardinals are identical and are \aleph_1.

\sigma -Ideal of Bounded Sets

In some respects, it is more advantageous to consider the \sigma-ideal of meager subsets of \mathbb{P}, the set of all irrational numbers, or equivalently \omega^\omega. Thus we consider the \sigma-ideal of meager subsets of \omega^\omega. We also use \mathcal{M} denote this \sigma-ideal. Note that the calculation of the four cardinals \text{add}(\mathcal{M}), \text{cov}(\mathcal{M}), \text{non}(\mathcal{M}) and \text{cof} (\mathcal{M}) yields the same values regardless of whether \mathcal{M} is the \sigma-ideal of meager subsets of the real line or of \omega^\omega. In the remainder of this post, \mathcal{M} is the \sigma-ideal of meager subsets of \omega^\omega, the space of the irrational numbers.

Let \mathcal{S} be the collection of all \sigma-compact subsets of \omega^\omega. In this previous post, the following \sigma-ideal is discussed.

    \mathcal{K}=\{ A \subset \omega^\omega: \exists \ B \in \mathcal{S} \text{ such that } A \subset B \}

It is straightforward to verify that \mathcal{K} is indeed a \sigma-ideal on \omega^\omega. This is what we know about this \sigma-ideal from this previous post.

  • A \in \mathcal{K} if and only if A is a bounded subset of \omega^\omega.
  • \mathfrak{b}=\text{add}(\mathcal{K})=\text{non}(\mathcal{K}).
  • \mathfrak{d}=\text{cov}(\mathcal{K})=\text{cof}(\mathcal{K}).

So the sets in \mathcal{K} are simply the bounded sets. For this \sigma-ideal, the additivity number and the uniformity numbers are \mathfrak{b}, the bounding number. The covering number and the cofinality number are the dominating number \mathfrak{d}. As we will see below, these facts provide insight on the \sigma-ideal \mathcal{M}.

The following lemma connects the \sigma-ideal \mathcal{K} with the \sigma-ideal \mathcal{M}.

Lemma 1
Let A be a compact subset of \omega^\omega. Then A is a closed and nowhere dense subset of \omega^\omega. Hence any \sigma-compact subset of \omega^\omega is a meager subset of \omega^\omega.

Proof of Lemma 1
Since A is compact, for each n, the projection of A into the nth factor of \omega^\omega is compact and thus finite. Let [0, g(n)]=\{ j \in \omega: 0 \le j \le g(n) \} be a finite set that contains the nth projection of A. Thus A \subset \prod_{n \in \omega} [0, g(n)]. It is straightforward to verify that \prod_{n \in \omega} [0, g(n)] is nowhere dense in \omega^\omega. Thus A is a closed nowhere dense subset of \omega^\omega. It follows that any \sigma-compact subset of \omega^\omega is a meager subset of \omega^\omega. \square

Theorem 2
As a result of Lemma 1, we have \mathcal{K} \subset \mathcal{M}. However, \mathcal{M} \not \subset \mathcal{K}. Thus the two \sigma-ideals are not the same.

Any example that proves Theorem 2 would be an unbounded meager set. One such example is constructed in this previous post.

More on \sigma -Ideal of Meager Sets

For \mathcal{K}, the \sigma-ideal generated by \sigma-compact subsets of \omega^\omega, and for \mathcal{M}, the \sigma-ideal of meager sets in \omega^\omega, we are interested in the four associated cardinals add, non, cov and cof in each \sigma-ideal. For \mathcal{K}, the four cardinals are just two, \mathfrak{b} and \mathfrak{d}. We would like to relate these six cardinals, plus \aleph_1 and 2^{\aleph_0}. They are represented in the following diagram.

Figure 5 – Partial Cichon’s Diagram

…Cichon…\displaystyle \begin{array}{ccccccccccccc} \text{ } & \text{ } & \text{ } &\text{ } & \bold n \bold o \bold n ( \mathcal{M} ) &\text{ } &\Longrightarrow &\text{ } &\bold c \bold o \bold f (\mathcal{M} )&\text{ } &\Longrightarrow & \text{ } & 2^{\aleph_0}\\  \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ } & \text{ }\\   \text{ }& \text{ }  & \text{ } &\text{ } & \Uparrow &\text{ } &\text{ } &\text{ } &\Uparrow &\text{ } &\text{ }  & \text{ } & \text{ }\\     \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\     \text{ } & \text{ } & \text{ } &\text{ } & \mathfrak{b} &\text{ } &\Longrightarrow &\text{ } &\mathfrak{d}&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ }& \text{ }  & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ } & \text{ } & \text{ } &\text{ } & \Uparrow &\text{ } &\text{ } &\text{ } &\Uparrow &\text{ } &\text{ }  & \text{ } & \text{ }\\   \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \aleph_1 & \text{ } & \Longrightarrow & \text{ } &\bold a \bold d \bold d ( \mathcal{M} ) & \text{ } &\Longrightarrow & \text{ } &\bold c \bold o \bold v ( \mathcal{M} ) &\text{ } &\text{ } &\text{ }  & \text{ }    \end{array}

As in the other diagrams, arrows mean \le. So \alpha \Longrightarrow \beta means \alpha \le \beta. Furthermore, we have two additional relations.

Additional Relationships

…Cichon…\text{add}(\mathcal{M})=\text{min}(\mathfrak{b},\ \text{cov}(\mathcal{M}))
…Cichon…\text{cof}(\mathcal{M})=\text{max}(\mathfrak{d},\ \text{non}(\mathcal{M}))

As shown, Figure 5 is not complete. It only has information on the \sigma-ideal \mathcal{M} on meager sets. The usual Cichon’s diagram would also include the four associated cardinals for \mathcal{L}, the \sigma-ideal of Lebesgue measure zero sets. In this post we focus on \mathcal{M}. The full Cichon’s diagram will be covered in a subsequent post. insert

In Figure 5, the cardinals go from smaller to the larger from left to right and bottom to top. It starts with \aleph_1 on the lower left and moves toward the continuum on the upper right. Because of Theorem 3, the cardinals associated with the \sigma-ideal \mathcal{K} are represented by \mathfrak{b} and \mathfrak{d} in the diagram. We next examine the inequalities between the cardinals associated with \mathcal{M} and \mathfrak{b} and \mathfrak{d}.

There are four inequalities to account for. First, \mathfrak{b} \le \text{non}(\mathcal{M}) and \text{cov}(\mathcal{M}) \le \mathfrak{d}. The first inequality follows from the fact that \mathfrak{b} = \text{non}(\mathcal{K}) and that \mathcal{K} \subset \mathcal{M}. The second inequality follows from the fact that \mathfrak{d} = \text{cov}(\mathcal{K}) and that \mathcal{K} \subset \mathcal{M}.

The inequality \text{add}(\mathcal{K}) \le \mathfrak{b} follows from the fact that if F \subset \omega^\omega is an unbounded set, then there exist \lvert F \lvert many meager sets whose union is a non-meager set. This fact is established in this previous post (see Theorem 1 in that post).

For the inequality \mathfrak{d} \le \text{cof}(\mathcal{M}), see Corollary 5.4 of [1]. For the additional inequalities, see Theorem 5.6 in [1].

The next post is on the full Cichon’s Diagram.

Reference

  1. Blass, A., Combinatorial Cardinal Characteristics of the Continuum, Handbook of Set Theory (M. Foreman, A. Kanamori, eds), Springer Science+Business Media B. V., Netherlands, 395-489, 2010.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

The ideal of bounded sets

This is the second in a series of posts leading to a diagram called The Cichon’s Diagram. In this post, we examine an ideal that will provide insight on the ideal of meager sets, which is part of the Cichon’s Diagram. For the definitions of ideal and \sigma-ideal, see the first post.

The next two posts are: the third post and the fourth post – the Cichon’s Diagram.

Let \omega be the set of all non-negative integers, i.e. \omega=\{ 0,1,2,\cdots \}. Let X=\omega^\omega, the set of all functions from \omega into \omega. We can also think of X as a topological space since it is a product space of countably many copies of the discrete space \omega. As a product space, X=\omega^\omega is homeomorphic to \mathbb{P}, the space of all irrational numbers with the usual real line topology (see here).

Recall that for f,g \in \omega^\omega, f \le^* g means that f(n) \le g(n) for all but finitely many n. This is a partial order that is called the eventual domination order. A subset F of \omega^\omega is a bounded set if there is a g \in \omega^\omega such that g is an upper bound of F with respect to the partial order \le^*, i.e. for each f \in F, we have f \le^* g. The set F is an unbounded set of it is not bounded. The set F is a dominating set if for each g \in \omega^\omega, there exists f \in F such that g \le^* f, i.e. the set F is cofinal in \omega^\omega with respect to the eventual domination order \le^*.

We are interested in the least cardinality of an unbounded set and the least cardinality of a dominating set. The former is denoted by \mathfrak{b} and is called the bounding number while the latter is denoted by \mathfrak{d} and is called the dominating number.

An Interim Ideal

We define two ideals on X=\omega^\omega. Let \mathcal{S} be the collection of all \sigma-compact subsets of \omega^\omega.

    \mathcal{K}_\sigma=\{ A \subset \omega^\omega: \exists \ B \in \mathcal{S} \text{ such that } A \subset B \}

    \mathcal{K}_b=\{ A \subset \omega^\omega: A \text{ is a bounded set} \}

The first one \mathcal{K}_\sigma is the set of all subsets of \omega^\omega, each of which is contained in a \sigma-compact set. The second one \mathcal{K}_b is simply the set of all bounded subsets. It is straightforward to verify that \mathcal{K}_\sigma is a \sigma-ideal on \omega^\omega. Note that any countable set \{ f_0, f_1,f_2,\cdots \} \subset \omega^\omega is a bounded set (via a diagonal argument). Thus the union of countably many bounded sets A_0,A_1,A_2,\cdots with A_n having an upper bound f_n must be a bounded set. The f_n have an upper bound f, which is an upper bound of the union of the sets A_n. Thus \mathcal{K}_b is a \sigma-ideal on \omega^\omega.

Furthermore, since \omega^\omega is not \sigma-compact, \mathcal{K}_\sigma is a proper ideal. Likewise \omega^\omega is an unbounded set, \mathcal{K}_b is a proper ideal. The ideal \mathcal{K}_\sigma is called the \sigma-ideal generated by \sigma-compact subsets of \omega^\omega. The ideal \mathcal{K}_b is the \sigma-ideal of bounded subset of \omega^\omega. However, these two ideals are one and the same.

Theorem 1
Let F \subset \omega^\omega. Then the following conditions are equivalent.

  1. The set F is bounded.
  2. There exists a \sigma-compact set X such that F \subset X \subset \omega^\omega.
  3. With F as a subset of the real line, the set F is an F_\sigma-subset of F \cup \mathbb{Q} where \mathbb{Q} is the set of all rational numbers.

Theorem 1 is the Theorem 1 found in
. The sets satisfying Condition 1 of this theorem are precisely the elements of the \sigma-ideal \mathcal{K}_b. The sets satisfying Condition 2 of this theorem are precisely the elements of the \sigma-ideal \mathcal{K}_\sigma. According to this theorem, the two \sigma-ideals are the same. Each is a different characterization of the same \sigma-ideal. As a result, we drop the subscript and call this \sigma-ideal \mathcal{K}.

Four Cardinals

With the \sigma-ideal \mathcal{K} from the preceding section, we would like to examine the four associated cardinals \text{add}(\mathcal{K}) (the additivity number), \text{non}(\mathcal{K}) (the uniformity number), \text{cov}(\mathcal{K}) (the covering number) and \text{cof}(\mathcal{K}) (the cofinality number). For the definitions of these numbers, see the first post.

Figure 1 – Cardinal Characteristics of the \sigma -Ideal Generated by \sigma -Compact Sets

…Cichon…\displaystyle \begin{array}{ccccccccccccc} \text{ } & \text{ } & \text{ } &\text{ } & \bold n \bold o \bold n ( \mathcal{K} ) &\text{ } &\Longrightarrow &\text{ } &\bold c \bold o \bold f (\mathcal{K} )&\text{ } &\Longrightarrow & \text{ } & 2^{\aleph_0}\\  \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ } & \text{ } \\     \text{ } & \text{ } & \text{ } &\text{ } & \Uparrow &\text{ } &\text{ } &\text{ } &\Uparrow&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ }& \text{ }  & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \aleph_1 & \text{ } & \Longrightarrow & \text{ } &\bold a \bold d \bold d ( \mathcal{K} ) & \text{ } &\Longrightarrow & \text{ } &\bold c \bold o \bold v ( \mathcal{K} ) &\text{ } &\text{ } &\text{ }  & \text{ }    \end{array}

In the diagram, \alpha \Rightarrow \beta means that \alpha \le \beta. The additivity number \text{add}(\mathcal{K}) is lowered bounded by \aleph_1 on the lower right in the diagram since the ideal \mathcal{K} is a \sigma-ideal. The middle of the diagram shows the relationships that hold for any \sigma-ideal. To see that \text{cof}(\mathcal{K}) \le 2^{\aleph_0}, define B_f=\{ h \in \omega^\omega: h \le^* f \} for each f \in \omega^\omega. The set of all B_f is cofinal in \mathcal{K}. The inequality holds since there are 2^{\aleph_0} many sets B_f.

We can further refine Figure 1. The following theorem shows how.

Theorem 2
The values of the four cardinals associated with the \sigma-ideal \mathcal{K} are the bounding numbers \mathfrak{b} and the dominating number \mathfrak{d}. Specifically, we have the following equalities.

    \mathfrak{b}=\text{add}(\mathcal{K})=\text{non}(\mathcal{K})
    \mathfrak{d}=\text{cov}(\mathcal{K})=\text{cof}(\mathcal{K})

Proof of Theorem 2
Based on the discussion in the first post, \text{add}(\mathcal{K}) \le \text{non}(\mathcal{K}) and \text{cov}(\mathcal{K}) \le \text{cof}(\mathcal{K}) always hold. We establish the equalities by showing the following.

    \mathfrak{b} \le \text{add}(\mathcal{K}) \le \text{non}(\mathcal{K}) = \mathfrak{b}
    \mathfrak{d} \le \text{cov}(\mathcal{K}) \le \text{cof}(\mathcal{K}) \le \mathfrak{d}

Viewing \mathcal{K} as a \sigma-ideal of bounded sets, \text{non}(\mathcal{K}) is the least cardinality of an unbounded set. Thus \mathfrak{b}=\text{non}(\mathcal{K}).

To see \mathfrak{b} \le \text{add}(\mathcal{K}), let \mathcal{A} \subset \mathcal{K} such that \lvert \mathcal{A} \lvert=\text{add}(\mathcal{K}) and Y=\bigcup \mathcal{A} \notin \mathcal{K}. Note that each A \in \mathcal{A} is a bounded set with an upper bound f(A) \in \omega^\omega. We claim that F=\{ f(A): A \in \mathcal{A} \} is unbounded. This is because Y=\bigcup \mathcal{A} is unbounded. Since there exists an unbounded set F with cardinality \text{add}(\mathcal{K}), it follows that \mathfrak{b} \le \text{add}(\mathcal{K}).

To see \text{cof}(\mathcal{K}) \le \mathfrak{d}, let F \subset \omega^\omega be a dominating set such that \lvert F \lvert=\mathfrak{d}. Note that for each f \in \omega^\omega, the set B_f=\{ h \in \omega^\omega: h \le^* f \} is a bounded set and thus B_f \in \mathcal{K}. It can be verified that \mathcal{B}=\{ B_f: f \in F \} is cofinal in \mathcal{K}. Since there is a cofinal set \mathcal{B} with cardinality \mathfrak{d}, it follows that \text{cof}(\mathcal{K}) \le \mathfrak{d}.

To see \mathfrak{d} \le \text{cov}(\mathcal{K}), let \mathcal{W} \subset \mathcal{K} such that \lvert \mathcal{W} \lvert=\text{cov}(\mathcal{K}) and \bigcup \mathcal{W}=\omega^\omega. For each A \in \mathcal{W}, let f(A) be an upper bound of A. It can be verified that the set F=\{ f(A): A \in \mathcal{W} \} is a dominating set. Since we have a dominating set F with cardinality \text{cov}(\mathcal{K}), we have \mathfrak{d} \le \text{cov}(\mathcal{K}). This completes the proof of Theorem 2. \square

With additional information from Theorem 2, Figure 1 can be revised as follows:

Figure 2 – Revised Figure 1

…Cichon…\displaystyle \begin{array}{ccccccccccccc} \text{ } & \text{ } & \text{ } &\text{ } & \mathfrak{b}=\bold n \bold o \bold n ( \mathcal{K} ) &\text{ } &\Longrightarrow &\text{ } &\mathfrak{d}=\bold c \bold o \bold f (\mathcal{K} )&\text{ } &\Longrightarrow & \text{ } & 2^{\aleph_0}\\  \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ } & \text{ } \\     \text{ } & \text{ } & \text{ } &\text{ } & \parallel &\text{ } &\text{ } &\text{ } &\parallel&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ }& \text{ }  & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \aleph_1 & \text{ } & \Longrightarrow & \text{ } &\mathfrak{b}=\bold a \bold d \bold d ( \mathcal{K} ) & \text{ } &\Longrightarrow & \text{ } &\mathfrak{d}=\bold c \bold o \bold v ( \mathcal{K} ) &\text{ } &\text{ } &\text{ }  & \text{ }    \end{array}

Note that there are only four cardinals in this diagram – \aleph_1, \mathfrak{b}, \mathfrak{d} and 2^{\aleph_0}. Of course, if continuum hypothesis holds, there would only one number in the diagram, namely \aleph_1.

The next post is on the \sigma-ideal \mathcal{M} of meager sets.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

Cardinals associated with an ideal

This is the first in a series of posts leading to a diagram called the Cichon’s diagram. The diagram displays relationships among twelve small cardinals, eight of which are defined by using \sigma-ideals. The purpose of this post is to set up the scene.

The next three posts are: the second post, the third post and the fourth post – the Cichon’s Diagram.

Ideals and \sigma -Ideals

Let X be a set. Let \mathcal{I} be a collection of subsets of X. We say that \mathcal{I} is an ideal on X if the following three conditions hold.

  1. \varnothing \in \mathcal{I}.
  2. If A \in \mathcal{I} and B \subset A, then B \in \mathcal{I}.
  3. If A, B \in \mathcal{I}, then A \cup B \in \mathcal{I}.

If X \notin \mathcal{I}, then \mathcal{I} is said to be a proper ideal. Note that if X \in \mathcal{I}, then \mathcal{I} would simply be the power set of X. Thus we would only want to focus on proper ideals. Thus by ideal we mean proper ideal.

We say that \mathcal{I} is a \sigma-ideal on X if it is an ideal with the additional property that it is closed under taking countable unions, i.e. if for each n \in \omega, A_n \in \mathcal{I}, then \bigcup_{n \in \omega} A_n \in \mathcal{I}. For the discussion that follows, we even require that all singleton subsets of X are members of any \sigma-ideal \mathcal{I}.

Elements of an ideal or a \sigma-ideal are considered “small sets” or “negligible sets”. The definition of \sigma-ideal does indeed reflect how small sets should behave. The empty set is naturally a small set. Any subset of a small set should be a small set. The union of countably many small sets should also be a small set as is any countable set (the union of countably many singleton sets).

Let \mathcal{B} be a collection of subsets of the set X. We assume that \mathcal{B} is closed under taking countable unions. It is easy to verify that the set

    \mathcal{I}=\{ A \subset X: \exists \ B \in \mathcal{B} \text{ such that } A \subset B \}

is a \sigma-ideal on X. This \sigma-ideal \mathcal{I} is said to be generated by the set \mathcal{B}. The set \mathcal{B} is called a base for the \sigma-ideal \mathcal{I}. A subbase for a \sigma-ideal is simply a collection of subsets of X. Then a base would be generated by taking countable unions of sets in the subbase.

Four Cardinals

We now discuss the cardinal characteristics associated with a \sigma-ideal. As before, let X be a set and \mathcal{I} be a \sigma-ideal on X. As discussed above, we require that all singleton sets are in \mathcal{I}. We define the following four cardinals.

    Additivity Number
    \text{add}(\mathcal{I})=\text{min} \{ \lvert \mathcal{A} \lvert: \mathcal{A} \subset \mathcal{I} \text{ and } \bigcup \mathcal{A} \notin \mathcal{I} \}

    Covering Number
    \text{cov}(\mathcal{I})=\text{min} \{ \lvert \mathcal{A} \lvert: \mathcal{A} \subset \mathcal{I} \text{ and } \bigcup \mathcal{A} = X \}

    Uniformity Number
    \text{non}(\mathcal{I})=\text{min} \{ \lvert A \lvert: A \subset X \text{ and } A \notin \mathcal{I} \}

    Cofinality Number
    \text{cof} (\mathcal{I})=\text{min} \{ \lvert \mathcal{B} \lvert: \mathcal{B} \subset \mathcal{I} \text{ and } \mathcal{B} \text{ is cofinal in } \mathcal{I} \}

A subset \mathcal{B} of \mathcal{I} is said to be cofinal in \mathcal{I} if for each A \in \mathcal{I}, there exists B \in \mathcal{B} such that A \subset B, i.e. \mathcal{B} is cofinal in the partial order \subset. Such a \mathcal{B} is a base for \mathcal{I}.

The numbers \text{add}(\mathcal{I}), \text{cov}(\mathcal{I}) and \text{cof} (\mathcal{I}) are the minimum cardinalities of certain subfamilies of the \sigma-ideal \mathcal{I} which fail to be small, i.e. not in \mathcal{I}. The additivity number \text{add}(\mathcal{I}) is the least cardinality of a subfamily of \mathcal{I} whose union is not in \mathcal{I}. The covering number \text{cov}(\mathcal{I}) is the minimum cardinality of a subfamily of \mathcal{I} whose union is the entire set X. The covering number is the minimum cardinality of a covering of X with elements of \mathcal{I}. The cofinality number \text{cof} (\mathcal{I}) is the least cardinality of a subfamily of \mathcal{I} that is a cofinal in \mathcal{I}. Equivalently, the cofinality number is the least cardinality of a base that generates the \sigma-ideal. The uniformity number \text{non}(\mathcal{I}) is the least cardinality of a subset of X that is not an element of \mathcal{I}.

The elements of the \sigma-ideal \mathcal{I} are “small” sets. The additivity number \text{add}(\mathcal{I}) is the smallest number of small sets whose union is not small. The covering number \text{cov}(\mathcal{I}) is then the smallest number of small sets that fill up the entire set X. The uniform number is the least cardinality of a non-small set.

Of these four cardinals, the smallest one is \text{add}(\mathcal{I}) and the largest one is \text{cof} (\mathcal{I}). Because \mathcal{I} is a \sigma-ideal, all four cardinals must be uncountable, hence \ge \aleph_1. Obviously \mathcal{I} is confinal in \mathcal{I}. Thus \text{cof} (\mathcal{I}) \le \lvert \mathcal{I} \lvert. The following inequalities also hold.

    \aleph_1 \le \text{add}(\mathcal{I}) \le \text{cov}(\mathcal{I}) \le \text{cof} (\mathcal{I}) \le \lvert \mathcal{I} \lvert

    \aleph_1 \le \text{add}(\mathcal{I}) \le \text{non}(\mathcal{I}) \le \text{cof} (\mathcal{I}) \le \lvert \mathcal{I} \lvert

    \displaystyle \aleph_1 \le \text{add}(\mathcal{I}) \le \text{min} \{ \text{cov}(\mathcal{I}), \text{non}(\mathcal{I}) \} \le \text{max} \{ \text{cov}(\mathcal{I}), \text{non}(\mathcal{I}) \} \le \text{cof} (\mathcal{I}) \le \lvert \mathcal{I} \lvert

Displaying the Four Cardinals in a Diagram

The inequalities shown in the preceding section can be displayed in a diagram such as the following.

Figure 1 – Cardinal Characteristics of a \sigma -Ideal

…Cichon…\displaystyle \begin{array}{ccccccccccccc} \text{ } & \text{ } & \text{ } &\text{ } & \bold n \bold o \bold n ( \mathcal{I} ) &\text{ } &\Longrightarrow &\text{ } &\bold c \bold o \bold f (\mathcal{I} )&\text{ } &\Longrightarrow & \text{ } & \lvert \mathcal{I} \lvert\\  \text{ } & \text{ } & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ } & \text{ } \\     \text{ } & \text{ } & \text{ } &\text{ } & \Uparrow &\text{ } &\text{ } &\text{ } &\Uparrow&\text{ } &\text{ }  & \text{ }& \text{ } \\    \text{ }& \text{ }  & \text{ } &\text{ } & \text{ } &\text{ } &\text{ } &\text{ } &\text{ }&\text{ } &\text{ }  & \text{ }& \text{ } \\    \aleph_1 & \text{ } & \Longrightarrow & \text{ } &\bold a \bold d \bold d ( \mathcal{I} ) & \text{ } &\Longrightarrow & \text{ } &\bold c \bold o \bold v ( \mathcal{I} ) &\text{ } &\text{ } &\text{ }  & \text{ }    \end{array}

In the above diagram, a \Rightarrow b means a \le b. The smallest cardinal \text{add}(\mathcal{I}) is lower bounded by \aleph_1 on the lower left since \mathcal{I} is a \sigma-ideal. The largest cardinal \text{cof}(\mathcal{I}) is upper bounded by the cardinality of \mathcal{I} on the upper right since \mathcal{I} is cofinal in \mathcal{I}. The diagram tells us that the additivity number is less than or equal to the minimum of the uniformity number and the covering number. On the other hand, the cofinality number is greater than or equal to the maximum of the uniformity number and the covering number.

Examples

In the subsequent posts, we would like to focus on two \sigma-ideals, hence eight associated cardinals. To define these two ideals, let X=\mathbb{R}, the real line. Let \mathcal{M} be set of all meager subsets of the real line and let \mathcal{L} be the set of all subsets of the real line that are of Lebesgue measure zero.

    \mathcal{M}=\{ A \subset \mathbb{R}: A \text{ is a Meager set} \}

    \mathcal{L}=\{ A \subset \mathbb{R}: A \text{ is of Lebesgue measure zero} \}

In the real line, a set is nowhere dense if its closure contains no open set. A meager set is the union of countably many nowhere dense sets. It is straightforward to verify that \mathcal{M} is a \sigma-ideal on the real line \mathbb{R}. Because of the Baire category theorem, \mathbb{R} \notin \mathcal{M}. Thus it is a proper ideal. Similarly, it is straightforward to verify that \mathcal{L} is a \sigma-ideal as well as a proper ideal.

Before we examine the ideal \mathcal{M}, we consider the \sigma-ideal of bounded subsets of \omega^\omega in the next post.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

Adding up to a non-meager set

The preceding post gives a topological characterization of bounded subsets of \omega^\omega. From it, we know what it means topologically for a set to be unbounded. In this post we prove a theorem that ties unbounded sets to Baire category.

A set is nowhere dense if its closure has empty interior. A set is a meager set if it is the union of countably many nowhere dense sets. By definition, the union of countably many meager sets is always a meager set. In order for meager sets to add up to a non-meager set (though taking union), the number of meager sets must be uncountable. What is this uncountable cardinal number? We give an indication of how big this number is. In this post we give a constructive proof to the following fact:

Theorem 1 …. Given an unbounded set F \subset \omega^\omega, there exist \kappa=\lvert F \lvert many meager subsets of the real line whose union is not meager.

We will discuss the implications of this theorem after giving background information.

We use \omega to denote the set of all non-negative integers \{ 0,1,2,\cdots \}. The set \omega^\omega is the set of all functions from \omega into \omega. It is called the Baire space when it is topologized with the product space topology. It is well known that the Baire space is homeomorphic to the space of irrational numbers \mathbb{P} (see here).

The notion of boundedness or unboundedness used in Theorem 1 refers to the eventual domination order (\le^*) for functions in the product space. For f,g \in \omega^\omega, by f \le^* g, we mean f(n) \le g(n) for all but finitely many n. A set F \subset \omega^\omega is bounded if it has an upper bound with respect to the partial order \le^*, i.e. there is some f \in \omega^\omega such that g \le^* f for all g \in F. The set F is unbounded if it is not bounded. To spell it out, F is unbounded if for each f \in \omega^\omega, there exists g \in F such that g \not \le^* f, i.e. f(n)<g(n) for infinitely many n.

All countable subsets of the Baire space are bounded (using a diagonal argument). Thus unbounded sets must be uncountable. It does not take extra set theory to obtain an unbounded set. The Baire space \omega^\omega is unbounded. More interesting unbounded sets are those of a certain cardinality, say unbounded sets of cardinality \omega_1 or unbounded sets with cardinality less than continuum. Another interesting unbounded set is one that is of the least cardinality. In the literature, the least cardinality of an unbounded subset of \omega^\omega is called \mathfrak{b}, the bounding number.

Another notion that is part of Theorem 1 is the topological notion of small sets – meager sets. This is a topological notion and is defined in topological spaces. For the purpose at hand, we consider this notion in the context of the real line. As mentioned at the beginning of the post, a set is nowhere dense set if its closure has empty interior (i.e. the closure contains no open subset). Let A \subset \mathbb{R}. The set A is nowhere dense if no open set is a subset of the closure \overline{A}. An equivalent definition: the set A is nowhere dense if for every nonempty open subset U of the real line, there is a nonempty subset V of U such that V contains no points of A. Such a set is “thin” since it is dense no where. In any open set, we can also find an open subset that has no points of the nowhere dense set in question. A subset A of the real line is a meager set if it is the union of countably many nowhere dense sets. Another name of meager set is a set of first category. Any set that is not of first category is called a set of second category, or simply a non-meager set.

Corollaries

Subsets of the real line are either of first category (small sets) or of second category (large sets). Countably many meager sets cannot fill up the real line. This is a consequence of the Baire category theorem (see here). By definition, caountably many meager sets cannot fill up any non-meager subset of the real line. How many meager sets does it take to add up to a non-meager set?

Theorem 1 gives an answer to the above question. It can take as many meager sets as the size of an unbounded subset of the Baire space. If \kappa is a cardinal number for which there exists an unbounded subset of \omega^\omega whose cardinality is \kappa, then there exists a non-meager subset of the real line that is the union of \kappa many meager sets. The bounding number \mathfrak{b} is the least cardinality of an unbounded set. Thus there is always a non-meager subset of the real line that is the union of \mathfrak{b} many meager sets.

Let \kappa_A be the least cardinal number \kappa such that there exist \kappa many meager subsets of the real line whose union is not meager. Based on Theorem 1, the bounding number \mathfrak{b} is an upper bound of \kappa_A. These two corollaries just discussed are:

  • There always exists a non-meager subset of the real line that is the union of \mathfrak{b} many meager sets.
  • \kappa_A \le \mathfrak{b}.

The bounding number \mathfrak{b} points to a non-meager set that is the union of \mathfrak{b} many meager sets. However, the cardinal \kappa_A is the least number of meager sets whose union is a non-meager set and this number is no more than the bounding number. The cardinal \kappa_A is called the additivity number.

There are other corollaries to Theorem 1. Let A(c) be the statement that the union of fewer than continuum many meager subsets of the real line is a meager set. For any cardinal number \kappa, let A(\kappa) be the statement that the union of fewer than \kappa many meager subsets of the real line is a meager set. We have the following corollaries.

  • The statement A(c) implies that there are no unbounded subsets of \omega^\omega that have cardinalities less than continuum. In other words, A(c) implies that the bounding number \mathfrak{b} is continuum.
  • Let \kappa \le continuum. The statement A(\kappa) implies that there are no unbounded subsets of \omega^\omega that have cardinalities less than \kappa. In other words, A(\kappa) implies that the bounding number \mathfrak{b} is at least \kappa, i.e. \mathfrak{b} \ge \kappa.

Let B(c) be the statement that the real line is not the union of less than continuum many meager sets. Clearly, the statement A(c) implies the statement B(c). Thus, it follows from Theorem 1 that A(c) \Longrightarrow B(c) + \mathfrak{b}=2^{\aleph_0}. This is a result proven in Miller [1]. Theorem 1.2 in [1] essentially states that A(c) is equivalent to B(c) + \mathfrak{b}=2^{\aleph_0}. The proof of Theorem 1 given here is essentially the proof of one direction of Theorem 1.2 in [1]. Our proof has various omitted details added. As a result it should be easier to follow. We also realize that the proof of Theorem 1.2 in [1] proves more than that theorem. Therefore we put the main part of the constructive in a separate theorem. For example, Theorem 1 also proves that the additivity number \kappa_A is no more than \mathfrak{b}. This is one implication in the Cichon’s diagram.

Proof of Theorem 1

Let 2=\{ 0,1 \}. The set 2^\omega is the set of all functions from \omega into \{0, 1 \}. When 2^\omega is endowed with the product space topology, it is called the Cantor space and is homemorphic to the middle-third Cantor set in the unit interval [0,1]. We use \{ [s]: \exists \ n \in \omega \text{ such that } s \in 2^n \} as a base for the product topology where [s]=\{ t \in 2^\omega:  s \subset t \}.

Let F \subset \omega^\omega be an unbounded set. We assume that the unbounded set F satisfies two properties.

  • Each g \in F is an increasing function, i.e. g(i)<g(j) for any i<j.
  • For each g \in F, if j>g(n), then g(j)>g(n+1).

One may wonder if the two properties are satisfied by any given unbounded set. Since F is unbounded, we can increase the values of each function g \in F, the resulting set will still be an unbounded set. More specifically, for each g \in F, define g^*\in \omega^\omega as follows:

  • g^*(0)=g(0)+1,
  • for each n \ge 1, g^*(n)=g(n)+\text{max}\{ g^*(i): i<n \} + n+1.

The set F^*=\{ g^*: g \in F \} is also an unbounded set. Therefore we use F^* and rename it as F.

Fix g \in F. Define an increasing sequence of non-negative integers n_0,n_1,n_2,\cdots as follows. Let n_0 be any integer greater than 1. For each integer j \ge 1, let n_j=g(n_{j-1}). Since n_0>1, we have n_1=g(n_0)>g(1). It follows that for all integer k \ge 1, n_k>g(k).

For each g \in F, we have an associated sequence n_0,n_1,n_2,\cdots as described in the preceding paragraph. Now define C(g)=\{ q \in 2^\omega: \forall \ k, q(n_k)=1 \}. It is straightforward to verify that each C(g) is a closed and nowhere dense subset of the Cantor space 2^\omega. Let X=\bigcup \{C(g): g \in F \}. The set X is a union of meager sets. We show that it is a non-meager subset of 2^\omega. We prove the following claim.

Claim 1
For any countable family \{C_n: n \in \omega \} where each C_n is a nowhere dense subset of 2^\omega, we have X \not \subset \bigcup \{C_n: n \in \omega \}.

According to Claim 1, the set X cannot be contained in any arbitrary meager subset of 2^\omega. Thus X must be non-meager. To establish the claim, we define an increasing sequence of non-negative integers m_0,m_1,m_2,\cdots with the property that for any k \ge 1, for any i<k, and for any s \in 2^{m_k}, there exists t \in 2^{m_{k+1}} such that s \subset t and [t] \cap C_i=\varnothing.

The desired sequence is derived from the fact that the sets C_n are nowhere dense. Choose any m_0<m_1 to start. With m_1 determined, the only nowhere dense set to consider is C_0. For each s \in 2^{m_1}, choose some integer y>m_1 such that there exists t \in 2^{y+1} such that s \subset t and [t] \cap C_0=\varnothing. Let m_2 be an integer greater than all the possible y‘s that have been chosen. The integer m_2 can be chosen since there are only finitely many s \in 2^{m_1}.

Suppose m_0<\cdots<m_{k-1}<m_k have been chosen. Then the only nowhere dense sets to consider are C_0,\cdots,C_{k-1}. Then for each i \le k-1, for each s \in 2^{m_k}, choose some integer y>m_k such that there exists t \in 2^{y+1} such that s \subset t and [t] \cap C_i=\varnothing. As before let m_{k+1} be an integer greater than all the possible y‘s that have been chosen. Again m_{k+1} is possible since there are only finitely many i \le k-1 and only finitely many s \in 2^{m_k}.

Let Z=\{ m_k: k \in \omega \}. We make the following claim.

Claim 2
There exists h \in F such that the associated sequence n_0, n_1,n_2,\cdots satisfies the condition: \lvert [n_k,n_{k+1}) \cap Z \lvert \ge 2 for infinitely many k where [n_k,n_{k+1}) is the set \{ m \in \omega: n_k \le m < m_{k+1} \}.

Suppose Claim 2 is not true. For each g \in F and its associated sequence n_0, n_1,n_2,\cdots,

    (*) there exists some integer b such that for all k>b, \lvert [n_k,n_{k+1}) \cap Z \lvert \le 1.

Let f \in \omega^\omega be defined by f(k)=m_k for all k. Choose \overline{f} \in \omega^\omega in the following manner. For each k \in \omega, define d_k \in \omega^\omega by d_k(n)=f(n+k) for all n. Then choose \overline{f} \in \omega^\omega such that d_k \le^* \overline{f} for all k.

Fix g \in F. Let m_j be the least element of [n_b, \infty) \cap Z. Then for each k>b, we have g(k) \le n_k \le m_{j+k}=f(j+k)=d_j(k). Note that the inequality n_k \le m_{j+k} holds because of the assumption (*). It follows that g \le^* d_j \le^* \overline{f}. This says that \overline{f} is an upper bound of F contradicting that F is an unbounded set. Thus Claim 2 must be true.

Let h \in F be as described in Claim 2. We now prove another claim.

Claim 3
For each n, C_n is a nowhere dense subset of C(h).

Fix C_n. Let p be an integer such that [n_p,n_{p+1}) \cap Z has at least two points, say m_k and m_{k+1}. We can choose p large enough such that n<k. Choose s \in 2^{m_k}. Since n_p is arbitrary, [s] is an arbitrary open set in 2^\omega. Since m_k is in between n_p and n_{p+1}, [s] contains a point of C(h). Thus [s] \cap C(h) is an arbitrary open set in C(h). By the way m_k and m_{k+1} are chosen originally, there exists t \in 2^{m_{k+1}} such that s \subset t and [t] \cap C_n=\varnothing. Because m_k and m_{k+1} are in between n_p and n_{p+1}, [t] \cap C(h) \ne \varnothing. This establishes the claim that C_n is nowhere dense subset of C(h).

Note that C(h) is a closed subset of the Cantor space 2^\omega and hence is also compact. Thus C(h) is a Baire space and cannot be the union of countably many nowhere dense sets. Thus C(h) \not \subset \cup \{C_n: n \in \omega \}. Otherwise, C(h) would be the union of countably many nowhere dense sets. This means that X=\bigcup \{C(g): g \in F \} \not \subset \cup \{C_n: n \in \omega \}. This establishes Claim 1.

Considering the Cantor space 2^\omega as a subspace of the real line, each C(g) is also a closed nowhere dense subset of the real line. The set X=\bigcup \{C(g): g \in F \} is also not a meager subset of the real line. This establishes Theorem 1. \square

Reference

  1. Miller A. W., Some properties of measure and category, Trans. Amer. Math. Soc., 266, 93-114, 1981.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

Topological meaning of bounded sets

In this post, we discuss a topological characterization of bounded sets in \omega^\omega. We also give an example of an unbounded meager set. We also briefly discuss the \sigma-ideal generated by \sigma-compact subsets of \omega^\omega and \sigma-ideal of meager subsets of \omega^\omega.

Let \omega be the first infinite ordinal. We consider \omega to be the set of all non-negative integers \{0, 1, 2, \cdots \}. Let \omega^\omega be the set of all functions from \omega into \omega. When the set \omega is considered a discrete space, the set \omega^\omega is the product of countably many copies of \omega. As a product space, \omega^\omega is homeomorphic to the set \mathbb{P} of all irrational numbers (see here). The product space \omega^\omega is called the Baire space in the literature.

Even though the two are topologically the same, working in the product space has its advantage. With the Baire space, we can define a partial order. For f,g \in \omega^\omega, define f \le^* g if f(n) \le g(n) for all but finitely many n. Let F \subset \omega^\omega. The set F is a bounded set if there exists f \in \omega^\omega which is an upper bound of F according to the partial order \le^*. The set F is an unbounded set if it is not bounded. We prove the following theorem.

Theorem 1
Let F \subset \omega^\omega. Then the following conditions are equivalent.

  1. The set F is bounded.
  2. There exists a \sigma-compact set X such that F \subset X \subset \omega^\omega.
  3. With F as a subset of the real line, the set F is an F_\sigma-subset of F \cup \mathbb{Q} where \mathbb{Q} is the set of all rational numbers.

This theorem is Theorem 9.3 in p. 149 in [1], the chapter by Van Douwen in the Handbook of Set-Theoretic Topology. As mentioned above, the set \mathbb{P} of irrational numbers is homeomorphic to the Base space \omega^\omega. For \mathbb{P}, the irrational numbers are points on a straight line. For the Baire space, the irrational numbers are functions in a product space. The second condition of Theorem 1 tells us what it means topologically for a subset of the Baire space to be bounded. The third condition tells us what a bounded set means if the set is placed on a straight line.

A subset A of any topological space Y is said to be nowhere dense set if for any non-empty open subset U of Y, there exists a non-empty open subset V of U such that V \cap A=\varnothing. A subset M of the topological space Y is said to be a meager set if M is the union of countably many nowhere dense sets. Meager sets are “small” sets. The discussion that follows shows that bounded sets are meager sets.

For m \in \omega and s \in \omega^m, define [s]=\{ t \in \omega^\omega: \exists \ n \in \omega \text{ such that } s \subset t \}. Note that the set \mathcal{B} of all [s] over all s \in \omega^m over all m \in \omega is a base for the Baire space \omega^\omega. As discussed below in the proof of Theorem 1, any compact subset of \omega^\omega is a subset of A_g=\{ h \in \omega^\omega: \forall \ n, \ h(n) \le g(n) \} for some g \in \omega^\omega. Thus for any [s], there exists some [t] with s \subset t such that t is greater than some g(n). Thus sets of the form A_g and any compact subset of \omega^\omega are nowhere dense sets. Then any \sigma-compact subset of \omega^\omega is contained in the union of countably many sets of the form A_g and is thus a meager set. The following theorem follows from Theorem 1.

Theorem 2
If F is a bounded subset of \omega^\omega, then F is a meager set.

A natural question: is the converse of Theorem 2 true? If true, boundedness would be a characterization of meager subsets of \omega^\omega. We give an example of an unbounded nowhere dense set, thus showing that the converse is not true.

Example of an Unbounded Meager Set

Let \omega^{< \omega} be the union of all \omega^n where n \in \omega. Each \omega^n is the set of all functions t:n=\{0,1,\cdots, n-1 \} \rightarrow \omega. Recall that \mathcal{B} is a base of \omega^\omega that consists of sets of the form [s] where s \in \omega^{< \omega}. Note that [s] is the set of all t \in \omega^\omega such that s \subset t. To find a meager set, we remove [t] from each [s] \in \mathcal{B}. The points remaining in \omega^\omega form a nowhere dense set. We remove [t] in such a way that the resulting set is a dominating set, hence an unbounded set.

For [t] \in \mathcal{B}, we also notate [t] by [t]=[t(0),t(1),\cdots,t(n-1)] if t \in \omega^n. This would be the set of all h \in \omega^\omega such that h(i)=t(i) for all i \le n-1.

For each t \in \omega^1=\omega^{ \{ 0 \} }, let A_t=[t(0),t(0)+1]. Note that A_t \subset \omega^2 and A_t \subset [t]. We remove all such A_t from \omega^\omega.

For each t \in \omega^2=\omega^{ \{ 0,1 \} }, we define A_t where A_t=[t(0),t(1),j] where j= \text{max} \{ t(0),t(1) \}+1. Note that A_t \subset \omega^3 and A_t \subset [t]. We remove all such A_t.

For each t \in \omega^n=\omega^{ \{ 0,1,\cdots,n-1 \} }, we define A_t where A_t=[t(0),t(1),\cdots, t(n-1),j] where j= \text{max} \{ t(0),t(1),\cdots,t(n-1) \}+1. Note that A_t \subset \omega^{n+1} and A_t \subset [t]. We remove all such A_t.

Let X=\omega^\omega \backslash \bigcup_{t \in \omega^{< \omega}} A_t. The set X is clearly a nowhere dense subset of \omega^\omega since we remove an element of the base from each element of the base. We now show that X is a dominating set. To this end, let f \in \omega^\omega. We define g \in X such that f \le^* g. If f \in X, then we define g=f. Assume f \notin X. Choose the least n such that f \in A_t where t \in \omega^n. According to our notation t=[t(0),t(1),\cdots,t(n-1)]. Define g \in \omega^\omega as follows.

    g(i) = \begin{cases} t(i) & \ \ \ \mbox{if } i \le n-2 \\ t(n-1)+2 & \ \ \ \mbox{if } i=n-1 \\ \text{max} \{g(0),g(1), \cdots, g(n-1) \}+99 & \ \ \ \mbox{if } i = n  \\ \text{max} \{g(0),g(1), \cdots, g(n) \}+99 & \ \ \ \mbox{if } i = n+1  \\ \vdots & \ \ \ \ \ \ \ \ \ \ \vdots  \\ \text{max} \{g(0),g(1), \cdots, g(k-1) \} +99& \ \ \ \mbox{if } i = k \text{ and } k \ge n  \\ \vdots & \ \ \ \ \ \ \ \ \ \ \vdots    \end{cases}

Because g(n-1) > t(n-1), the basic open set [g(0),g(1),\cdots,g(n-1)] is not marked for removal. For i \ge n, because of the way g(i) is defined, the basic open set [g(0),g(1),\cdots,g(i)] is also not marked for removal. Thus g \in X.

Comment about the Example

The meager set that is a dominating set given above is not a Menger set (see here).

Theorem 2 and the example showing that the converse of Theorem 2 is not true speak to a situation involving two \sigma-ideals. One of the ideals is \mathcal{K}, which is the set of all subsets of \omega^\omega, each of which is contained in a \sigma-compact subset of \omega^\omega. The set \mathcal{K} is a \sigma-ideal. Theorem 1 says that elements of \mathcal{K} are precisely the bounded sets. Theorem 2 says that elements of \mathcal{K} are meager sets.

The other ideal is \mathcal{M}, which is the set of all meager subsets of \omega^\omega. This is also a \sigma-ideal. Then \mathcal{K} \subset \mathcal{M}. The example shows that the two \sigma-ideals are not the same, in particular \mathcal{M} \not \subset \mathcal{K}. The ideal \mathcal{K} is the \sigma-ideal generated by the \sigma-compact subsets of \omega^\omega. This \sigma-ideal is much smaller than the \sigma-ideal \mathcal{M} of meager subsets of \omega^\omega.

Proof of Theorem 1

It is helpful to set up notations and have a background discussion before proving the theorem. For g \in \omega^\omega, define the following sets:

    A_g=\{ h \in \omega^\omega: \forall \ n, \ h(n) \le g(n) \}

    B_g=\{ h \in \omega^\omega: h \le^* g \}

The set A_g is a compact set since it is the product of finite sets, i.e. A_g=\prod_{n \in \omega} [0,g(n)]. The set B_g is a \sigma-compact set. To see this, B_g=\bigcup_{n \in \omega} A_{h_n} for a sequence h_n \in  \omega^\omega. The sequence \{ h_n \} is obtained by considering, for each k \in \omega, all functions t \in \omega^\omega where t(i)=g(i) for all i \ge k while t(i) ranges over all non-negative integers for i<k. There are only countably many functions t for each k. Then enumerate all these functions in a sequence h_0, h_1,h_2,\cdots.

On the other hand, any compact subset of \omega^\omega is a subset of A_g for some g \in \omega^\omega. To see this, let \pi_n be the projection from \omega^\omega to the nth factor. Let K \subset \omega^\omega be compact. Then for each n, \pi_n(K) is compact in the discrete space \omega, hence finite. Since it is finite, for each n, \pi_n(K) \subset [0,g(n)] for some g(n) \in \omega. Then K \subset A_g.

It follows that any \sigma-compact subset of \omega^\omega is a subset of the union of countably many A_g, i.e. if K \subset \omega^\omega is \sigma-compact, then K \subset \bigcup_{n \in \omega} A_{g_n} for g_0, g_1, g_2, \cdots \in \omega^\omega.

1 \rightarrow 2
Suppose F is bounded. Let f \in \omega^\omega be an upper bound of F. It is clear that F \subset B_f, which is \sigma-compact.

2 \rightarrow 3
Let F \subset \omega^\omega. Suppose that F \subset X \subset \omega^\omega where X=\bigcup_{n \in \omega} X_n with each X_n being a compact subset of \omega^\omega. For each n, let Y_n=F \cap X_n. Consider the sets F, X_n and Y_n as subsets of the real line. Since each X_n is compact and X_n \cap \mathbb{Q}=\varnothing, each Y_n is a closed subset of F \cup \mathbb{Q}. Thus F is an F_\sigma subset of F \cup \mathbb{Q}.

3 \rightarrow 1
Let F \subset \omega^\omega. Consider F as a subset of \mathbb{P}. Suppose that F=\bigcup_{n \in \omega} C_n where each C_n is a closed subset of F \cup \mathbb{Q}. For each n, let \overline{C_n} be the closure of C_n in the real line. Because it is a closed subset of the real line, \overline{C_n} is \sigma-compact. Since points of \mathbb{Q} are not in the closure of C_n in F \cup \mathbb{Q}, points of \mathbb{Q} are not in the closure of C_n in the real line. It follows that \overline{C_n} \subset \mathbb{P}. Now consider each \overline{C_n} as a subset of \omega^\omega. According to the above discussion, each \overline{C_n} is a subset of \bigcup_{j \in \omega} A_{g_{n,j}} where g_{n,0}, g_{n,1}, g_{n,2}, \cdots \in \omega^\omega. Choose f \in \omega^\omega such that g_{n,j} \le^* f for all n,j combinations. Then f is an upper bound of F. This completes the proof of Theorem 1. \square

Reference

  1. Van Douwen, E. K., The Integers and Topology, Handbook of Set-Theoretic Topology (K. Kunen and J. E. Vaughan, eds), Elsevier Science Publishers B. V., Amsterdam, 111-167, 1984.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

The space of irrational numbers is not Menger

This post could very well be titled Killing Two Birds in One Stone. We present one proof that can show two results – the set \mathbb{P} of all irrational numbers does not satisfy the Menger property and that \mathbb{P} is homeomorphic to the product space \omega^\omega.

In this previous post, we introduce the notion of Menger spaces. A space X is a Menger space (or has the Menger property) if for every sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \bigcup_{n \in \omega} \mathcal{V}_n is an open cover of X. In the definition if the open covers \mathcal{U}_n are made identical, then the definition becomes that of a Lindelof space. Thus Menger implies Lindelof. For anyone encountering the notion of Menger spaces for the first time, one natural question is: are there Lindelof spaces that are not Menger? Another natural question: are there subsets of the real line that are not Menger?

The most handy example of “Lindelof but not Menger” is probably \mathbb{P}, the set of all irrational numbers. The way we show this fact in this previous post is by working in the product space \omega^\omega. This approach requires an understanding that \mathbb{P} and the product space \omega^\omega are identical topologically as well as working with the eventual domination order \le^* in \omega^\omega. In this post, we give a direct proof that \mathbb{P} is not Menger with the irrational numbers lying on a line. This proof will open up an opportunity to see that \mathbb{P} is homeomorphic to \omega^\omega. The product space \omega^\omega is also called the Baire space in the literature.

The sets \mathbb{P} and \omega^\omega are two different (but topologically equivalent) ways of looking at the irrational numbers. The first way is to look at the numbers on a line. With the rational numbers removed, the line will have countably many holes (but the holes are dense). The open sets in the line are open intervals or unions of open intervals. The second way is to view the irrational numbers as a product space – the product of countably many copies of \omega with \omega being the set of all non-negative numbers with the discrete topology. In many ways the product space version is easier to work with. For example, the eventual domination order \le^* on \omega^\omega help describe the classical covering properties such as Menger property and Hurewicz property.

A Direct Proof of Non-Menger

As mentioned above, the set \mathbb{P} is the set of all irrational numbers. We let \mathbb{Q} denote the set of all rational numbers.

To see that \mathbb{P} does not satisfy the Menger property, we need to exhibit a sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of \mathbb{P} such that whenever we pick, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n, the set \{ \mathcal{V}_n: n \in \omega \} cannot be a cover of \mathbb{P}, i.e. no matter how we pick the finite \mathcal{V}_n, there is always an irrational number x that is missed by all \mathcal{V}_n. The open covers \mathcal{U}_n are derived inductively, starting with \mathcal{U}_0.

To prepare for the inductive steps, we fix a scheme of choosing convergent sequences of rational numbers. For any interval (a,b) where a, b \in \mathbb{Q}, we choose a decreasing sequence \{ a_n \in (a,b) \cap \mathbb{Q}: n \in \omega \} such that a_n \rightarrow a from the right with a_0=b and an increasing sequence \{ b_n \in (a,b) \cap \mathbb{Q}: n \in \omega \} such that b_n \rightarrow b from the left with b_0=a. Enumerate \mathbb{Q} in a countable sequence \{ r_0, r_1, r_2, \cdots \} with r_0=0. Several points to keep in mind when choosing such sequences.

  1. In each inductive step, either the sequences of the type a_n or the sequences of the type b_n are chosen but not both.
  2. in the j-th stage of the inductive process, in picking the sequence a_n or b_n for the interval (a,b), we make sure that the rational number r_j is used in the sequence a_n or b_n if r_j is in the interval (a,b) and if r_j is not previously chosen. So at the end, all rational numbers are used up as endpoints of intervals in all \mathcal{U}_n.
  3. If sequence of the type a_n is chosen out of the interval (a,b), we derive the subintervals \cdots (a_3,a_2),(a_2,a_1),(a_1,a_0) whose union is (a,b). We want the sequence a_n chosen in such a way that the length of each subinterval (a_n,a_{n-1}) is less than 1/2 of b-a.
  4. If sequence of the type b_n is chosen out of the interval (a,b), we derive the subintervals (b_0,b_1), (b_1,b_2),(b_2,b_3),\cdots whose union is (a,b). We want the sequence b_n chosen in such a way that the length of each subinterval (b_n,b_{n+1}) is less than 1/2 of b-a.

To start, let \mathcal{U}_0=\{ (0,1),(1,2),(2,3),\cdots,(-1,0),(-2,-1),(-3,-2),\cdots \}. In other words, \mathcal{U}_0 consists of all open intervals whose endpoints are the integers and whose lengths are 1. Label the elements of \mathcal{U}_0 as \mathcal{U}_0=\{ O_0,O_1,O_2,\cdots \}.

For each O_i=(a,b) \in \mathcal{U}_0, choose a sequence of rational numbers b_n converging to b from the left (according to the scheme described above). Make sure that the rational number r_1 is picked if r_1 \in (a,b). From this, we obtain the intervals (b_0,b_1), (b_1,b_2),(b_2,b_3),\cdots covering all the irrational numbers in (a,b). Label these intervals as O_{i,0},O_{i,1},O_{i,2},\cdots. Then \mathcal{U}_1 consists of all such intervals obtained from each O_i=(a,b) \in \mathcal{U}_0.

Here’s how to define \mathcal{U}_2. For each O_{i,j}=(a,b) \in \mathcal{U}_0, choose a sequence of rational numbers a_n converging to a from the right (according to the scheme described above). Make sure that the rational number r_2 is picked if r_2 \in (a,b) and if r_2 has not been chosen previously. From this, we obtain the intervals \cdots (a_3,a_2),(a_2,a_1),(a_1,a_0) covering all the irrational numbers in (a,b). Label these intervals as O_{i,j,0},O_{i,j,1},O_{i,j,2},\cdots. Then \mathcal{U}_2 consists of all such intervals obtained from each O_{i,j}=(a,b) \in \mathcal{U}_1.

From here on out, we continue the same induction steps to derive \mathcal{U}_n, making sure that when n is odd, we choose sequence b_n converging to the end point b from the left and when n is even, we choose sequence a_n converging to the end point a from the right for each O_{i_0,i_2,\cdots,i_{n-1}}=(a,b) \in \mathcal{U}_{n-1}, producing the intervals O_{i_0,i_2,\cdots,i_{n-1},0}, O_{i_0,i_2,\cdots,i_{n-1},1}, O_{i_0,i_2,\cdots,i_{n-1},2}, O_{i_0,i_2,\cdots,i_{n-1},3}\cdots.

Now the sequence \{ \mathcal{U}_n: n \in \omega \} has been defined. Note that every rational number is used as the endpoint of some open interval in some \mathcal{U}_n. Each \mathcal{U}_n is clearly an open cover of \mathbb{P}. Another useful observation is that

    \bigcap_{n \in \omega} U_n=\bigcap_{n \in \omega} \overline{U_n}=\{ x \}

whenever U_n \in \mathcal{U}_n for all n and U_{n+1} \subset U_n for all n. This is because each chosen interval has length less than 1/2 of the previous interval. Furthermore, the point x in the intersection must be an irrational number since all rational numbers are used up as endpoints.

Let’s pick, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n. We now find an irrational number x that is not in any interval of any \mathcal{V}_n.

Since \mathcal{V}_0 is finite, choose some O_{m_0} \in \mathcal{U}_0 such that O_{m_0} \notin \mathcal{V}_0. Since \mathcal{V}_1 is finite, choose some O_{m_0,m_1} \in \mathcal{U}_1 such that O_{m_0,m_1} \notin \mathcal{V}_1. In the same manner, choose O_{m_0,m_1,m_2} \in \mathcal{U}_2 such that O_{m_0,m_1,m_2} \notin \mathcal{V}_2.

Continuing the inductive process, we have open intervals O_{m_0},O_{m_0,m_1},O_{m_0,m_1,m_2},\cdots from \mathcal{U}_0,\mathcal{U}_1,\mathcal{U}_2,\cdots, respectively. Furthermore, for each n, O_{m_0,\cdots,m_n} \notin \mathcal{V}_n. Then the intersection

    \bigcap_{n \in \omega} O_{m_0,\cdots,m_n}=\bigcap_{n \in \omega} \overline{O_{m_0,\cdots,m_n}}=\{ x \}

must be an irrational number x. This is a real number that is not covered by \mathcal{V}_n for all n. This shows that \{ \mathcal{V}_n: n \in \omega \} cannot be an open cover of \mathbb{P}. This completes the proof that \mathbb{P} does not satisfy the Menger property. \square

A Homeomorphism

In defining the sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of \mathbb{P}, we have also define a one-to-one correspondence between \mathbb{P} and the product space \omega^\omega.

First, each x \in \mathbb{P} is uniquely identified with a sequence of non-negative integers f_0,f_1,f_2,\cdots.

    x \mapsto (f_0,f_1,f_2,\cdots)=f \in \omega^\omega

Observe that each irrational number x belongs to a unique element in each \mathcal{U}_n. Thus the sequence f_0,f_1,f_2,\cdots is simply the subscripts of the open intervals from the open covers \mathcal{U}_n such that x \in O_{f_0,\cdots,f_n} for each n. With the way the intervals O_{f_0,\cdots,f_n} are chosen, it must be the case that O_{f_0} \supset O_{f_1} \supset O_{f_2} \cdots. Furthermore the intervals collapse to the point x.

    \bigcap_{n \in \omega} O_{f_0,\cdots,f_n}=\bigcap_{n \in \omega} \overline{O_{f_0,\cdots,f_n}}=\{ x \}……………(1)

The mapping goes in the reverse direction too. For each f=(f_0,f_1,f_2,\cdots) \in \omega^\omega, we can use f to obtain the open intervals O_{f_0,\cdots,f_n}. These intervals collapse to a single point, which is the irrational number that is associated with f=(f_0,f_1,f_2,\cdots).

Denote this mapping by H: \mathbb{P} \rightarrow \omega^\omega. For each x \in \mathbb{P}, H(x)=f is derived in the manner described above. The function H maps \mathbb{P} onto \omega^\omega. It follows that H is a one-to-one map and that both H and the inverse H^{-1} are continuous. To see this, we need to focus on the open sets in both the domain and the range.

For the correspondence H(x)=(f_0,f_1,f_2,\cdots)=f, we consider the two sets:

    \mathcal{B}_x=\{ O_{f_0}, O_{f_0,f_1}, O_{f_0,f_1,f_2}, \cdots, O_{f_0,\cdots,f_n}, \cdots \}

    \mathcal{V}_f=\{ E(f,n): n \in \omega \}

where E(f,n)=\{ g \in \omega^\omega: f_j=g_j \ \forall \ j \le n \} for each n. Note that \mathcal{B}_x is a local base at x \in \mathbb{P}. This is because the intervals O_{f_0,\cdots,f_n} have lengths converging to zero and they collapse to the point x in the manner described in (1) above. On the other hand, \mathcal{V}_f is a local base at H(x)=f \in \omega^\omega. Furthermore, there is a clear correspondence between \mathcal{B}_x and \mathcal{V}_f. Note that H(O_{f_0,\cdots,f_n})=E(f,n). This means that both H and the inverse H^{-1} are continuous. Since \{ \mathcal{B}_x: x \in \mathbb{P} \} is a base for \mathbb{P}, it is clear the map H is one-to-one.

This previous post has a shorter (but similar) derivation of the homeomorphism between \mathbb{P} and \omega^\omega.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

Hurewicz spaces are sigma-compact-like

Menger spaces and Hurewicz spaces are situated in between \sigma-compactness and the Lindelof property.

    \sigma \text{-} \bold C \bold o \bold m \bold p \bold a \bold c \bold t  \Longrightarrow \bold H \bold u \bold r \bold e \bold w \bold i \bold c \bold z \Longrightarrow \bold M \bold e \bold n \bold g \bold e \bold r \Longrightarrow \bold L \bold i \bold n \bold d \bold e \bold l \bold o \bold f

In the preceding post, we discuss a characterization of Menger spaces that is a Lindelof-like property. The attention is now turned to Hurewicz spaces. We discuss a \sigma-compact-like characterization of Hurewicz spaces among subsets of the real line.

It seems there is a “symmetry” here. The property that is closer to Lindelof property is Lindelof-like, while the one closer to \sigma-compactness is \sigma-compact-like. In addition to proving the \sigma-compact-like characterization, we use several examples to demonstrate this property. One of the examples is the Lusin sets. We also briefly discuss the Hurewicz problem.

Let X be a space. The space X is a Menger space (has the Menger property) if for each sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \bigcup_{n \in \omega} \mathcal{V}_n is an open cover of X. The space X is a Hurewicz space (has the Hurewicz property) if for each sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \{ \cup \mathcal{V}_n: n \in \omega \} is a \gamma-cover of X, i.e. each x \in X belongs to \cup \mathcal{V}_n for all but finitely many n.

Two previous posts (here and here) discuss basic facts of Menger spaces and Hurewicz spaces. Consider the following theorems.

Theorem 1
Let X \subset \mathbb{R}. Then the following conditions are equivalent.

  1. The set X has the Hurewicz property.
  2. For every G_\delta-set G with X \subset G, there exists an F_\sigma-set F such that X \subset F \subset G.

This theorem is Theorem 5.7 in [3]. In the real line, any F_\sigma-set is \sigma-compact. Hurewicz subsets of the real line may not be \sigma-compact but are “approximated” by \sigma-compact sets in the manner described in Theorem 1. Whenever a Hurewicz subset is situated in a G_\delta-set, we can always find a \sigma-compact set cushioned in between.

Theorem 2
Let X \subset \mathbb{R}. If X has empty interior and has the Hurewicz property, then X is of first category.

Theorem 2 is useful in showing that Lusin sets are not Hurewicz spaces.

Examples

In order to understand more about this \sigma-compact-like property of Hurewicz subsets of the real line, consider the following examples.

Example 1
Clearly, any F_\sigma-set in the real line satisfies condition 2 in Theorem 1.

Example 2
In the real line, any G_\delta-set X that is not F_\sigma does not satisfy condition 2 of Theorem 1.

For any G_\delta-set X that is not an F_\sigma-set, there can be no F_\sigma cushioned in between X and X. An example of such a set is the set \mathbb{P} of all irrational numbers. We already know that \mathbb{P} does not have the Menger property (see Theorem 4 here) and hence does not not have the Hurewicz property.

Example 3
Not only G_\delta-sets X that are not F_\sigma do not have the \sigma-compact-like property of Theorem 1, all Borel sets that that are not F_\sigma do not satisfy condition 2 of Theorem 1. This is because such sets are not Menger. This observation is made in Example 4 of the preceding post.

Example 4
Any Bernstein set in the real line does not satisfy condition 2 of Theorem 1.

Bernstein sets are not Menger. This is discussed in Example 5 of the preceding post. Thus they are not Hurewicz. As illustration, we demonstrate that \sigma-compact sets cannot be cushioned in between Bernstein sets and G_\delta-sets. Let X \subset \mathbb{R} be a Bernstein set. The complement \mathbb{R} \backslash X is also a Bernstein set and hence dense in the real line. Let D \subset \mathbb{R} \backslash X be a countable dense subset. Then \mathbb{R} \backslash D is a dense G_\delta-set containing X.

Suppose that X \subset F \subset \mathbb{R} \backslash D for some F_\sigma-set F. Let F=\bigcup_{n \in \omega} F_n where each F_n is a closed set. Observe that each F_n is nowhere dense in the real line. To see this, let U be an open interval. Since D is dense, choose x \in U \cap D. Since x \notin F_n, choose open interval V such that x \in V, V \cap F_n=\varnothing and V \subset U.

Consider W=\mathbb{R} \backslash F. The set W is a G_\delta-set. If W is countable, then the real line is the union of countably many nowhere dense sets F_n and a countable set which means that the real line is a first category set, a contradiction. Thus W must be uncountable. Since W is an uncountable G_\delta-set, W contains a Cantor set C. Since X is a Bernstein set, X \cap C \ne \varnothing. But C cannot contain points of X since C \subset W=\mathbb{R} \backslash F. Thus there can be no F_\sigma-set cushioned between the Bernstein set X and the G_\delta-set \mathbb{R} \backslash D.

Example 5
Any Lusin set does not satisfy condition 2 of Theorem 1 and hence is not Hurewicz.

Let X \subset \mathbb{R} where X is uncountable. The set X is a Lusin set (alternative spelling: Luzin) if for each F that is of first category in the real line, X \cap F is countable.

Suppose X is a Lusin set. We show that there can be no F_\sigma-set cushioned between X and some G_\delta-set containing X. Note that the Lusin set X has empty interior. Otherwise the Lusin set would contain an open interval. This is impossible. The reason is that that open interval would contain a Cantor set, which is nowhere dense. By Theorem 2, if X is Hurewicz, then it would be of first category. But a Lusin set by definition cannot be of first category. Thus X cannot be Hurewicz and thus does not satisfy the \sigma-compact-like property.

Example 6
Let X \subset \mathbb{R} such that X is a Sierpinski set. Then X has the Hurewicz property hence satisfies the \sigma-compact-like property.

The set X is a Sierpinski set if X is uncountable and X \cap L is countable for every Lebesgue measure zero set L. See Theorem 2.3 in [10] for a proof that any Sierpinski set is Hurewicz. My note. Source is Menger’s and Hurewicz’s problems: solutions from the book and refinements.

Hurewicz’s Problem

Example 5 concerns Lusin sets and is a good lead-in to the Hurewicz’s Problem.

In 1924, Karl Menger [4] conjectured that in metric spaces, the Menger property is equivalent to \sigma-compactness. In 1928 Sierpinski [9] pointed out that when continuum hypothesis (CH) holds, Menger’s conjecture is false. The counterexample was the Lusin sets, which were shown to exist under CH by N. Lusin in 1914. Thus Lusin sets are Menger spaces that are not \sigma-compact (the steps are worked out in Example 4 in this previous post). Lusin sets do not exist under Martin’s axiom and the negation of CH. Thus Lusin sets are only consistent counterexamples to Menger’s conjecture. The first ZFC counterexample to Menger’s conjecture was given by Miller and Fremlin [5] (the steps are worked out in Example 5 in this previous post).

Example 5 above shows that Lusin sets are not Hurewicz spaces. This shows that Lusin sets are solutions to another problem. The Hurewicz problem is the following question:

    Is there a metric space that is Menger but not Hurewicz?

Lusin sets answer this question in the affirmative (first solved by Sierpinski). The first ZFC example of a Menger space that is not Hurewicz is found in Chaber and Pol, which is unpublished source. Also see Theorem 5.3 in [10] for a combinatorial proof that contains the essence of the proof of Chaber and Pol. The proof in Chaber and Pol’s unpublished note and Theorem 5.3 in [10] are dichotomic. So no explicit examples are provided in these proofs. These proofs exploit the opposing cases of a set-theoretic statement. For example, when \mathfrak{b}<\mathfrak{d}, an example of a Menger but not Hurewicz space can be derived. On the other hand, when \mathfrak{b}=\mathfrak{d}, another example of a Menger but not Hurewicz space can be derived. The first explicit ZFC example of a Menger but not Hurewicz space can be found in [11].

Note that the Hurewicz problem is not the same as the Hurewicz’s conjecture. The latter is a conjecture made by W. Hurewicz in 1925 that a non-compact metric space is \sigma-compact if, and only if, it is a Hurewicz space. Sierpinski sets pointed out in Example 6 provide consistent counterexample to this conjecture. The first ZFC counterexample is provided in [3]. The proof in [3] is a dichotomic proof. See [10] for a more detailed discussion of Hurewicz’s conjecture.

The Proof Section

We now prove Theorem 1 in the following form. Proof of Theorem 2 follows the proof of Theorem 1.

Theorem 1
Let X \subset \mathbb{R}. Then the following conditions are equivalent.

  1. The set X has the Hurewicz property.
  2. For each sequence \{ \mathcal{U}_n: n \in \omega \} of \gamma-covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \{ \bigcup \mathcal{V}_n: n \in \omega \} is a \gamma-cover of X.
  3. For every G_\delta-set G with X \subset G, there exists an F_\sigma-set F such that X \subset F \subset G.

The direction 1 \rightarrow 2 is straightforward. Condition 2 is the Hurewicz property restricted to sequences of \gamma-covers. Note that any \gamma-cover is also an open cover. So if a space satisfies the full Hurewicz property, it will then satisfy the restricted property.

2 \rightarrow 3
Suppose that X \subset G such that G=\bigcap_{n \in \omega} O_n where each O_n is an open subset of the real line. We can also require that O_0 \supset O_1 \supset O_2 \supset \cdots. From the G_\delta-set G, we construct a sequence \{ \mathcal{U}_n: n \in \omega \} of \gamma-covers of X.

To define \mathcal{U}_n, do the following. For each x \in X, choose an open interval E_x^n such that x \in E_x^n and \overline{E_x^n} \subset O_n. For the open cover \{E_x^n: x \in X \}, we can find a countable subcover \{ E_{x_{n,k}}^n: k \in \omega \}. To simplify notation, we rename E_{x_{n,k}}^n=F_{k}^n. For each k, define H_k^n=\bigcup_{j \le k} F_{j}^n. Let \mathcal{U}_n=\{ H_k^n: k \in \omega \}. Note that \mathcal{U}_n is a \gamma-cover of X and that for each k, \overline{H_k^n} \subset O_n.

For each \mathcal{U}_n, note that the open sets H_k^n are increasing, i.e. for i<j, H_i^n \subset H_j^n. So picking a finite subset of \mathcal{U}_n is the same as picking one element (the largest one in the finite set). Using condition 2, we can choose, for each n, H_{t(n)}^n \in  \mathcal{U}_n such that \{ H_{t(n)}^n: n \in \omega \} is a \gamma-cover of X.

For each n, define L_n=\bigcap_{m \ge n} \overline{H_{t(m)}^m}. Note that each closure in the intersection is contained in the open set O_m. To see this,

    \overline{H_{t(m)}^m}=\overline{\bigcup_{j \le t(m)} F_j^m}=\bigcup_{j \le t(m)} \overline{F_j^m}=\bigcup_{j \le t(m)} \overline{E_{x_{m,j}}^m} \subset O_m

Thus L_n=\bigcap_{m \ge n} \overline{H_{t(m)}^m} \subset \bigcap_{m \ge n} O_m \subset \bigcap_{k \in \omega} O_k=G. Observe there is now an F_\sigma-set cushioned between X and G.

    X \subset \bigcup_{n \in \omega} L_n \subset G

3 \rightarrow 1
Let \{ \mathcal{U}_n: n \in \omega \} be a sequence of open covers of X such that elements of each \mathcal{U}_n are open subsets of the real line. By condition 3, there is an F_\sigma-set L such that

    X \subset L=\bigcup_{n \in \omega} L_n \subset \bigcap_{n \in \omega} (\bigcup \mathcal{U}_n)

Since the real line is \sigma-compact, we can assume each L_n is compact. For each n, we can choose finite \mathcal{V}_n \subset \mathcal{U}_n such that L_n \subset \bigcup \mathcal{V}_n. It follows that \bigcup_{n \in \omega} \mathcal{V}_n is an open cover of X. This completes the proof that X is a Hurewicz space. \square

Theorem 2
Let X \subset \mathbb{R}. If X has empty interior and has the Hurewicz property, then X is of first category.

Suppose X has empty interior and has the Hurewicz property. Since X has empty interior, \mathbb{R} \backslash X is a dense subset of the real line. Let D be a countable dense subset of \mathbb{R} \backslash X. Then \mathbb{R} \backslash D is a G_\delta-set and X \subset \mathbb{R} \backslash D. By Theorem 1, there exists an F_\sigma-set F such that X \subset F \subset \mathbb{R} \backslash D. Let F=\bigcup_{n \in \omega} F_n where each F_n is a closed set. By the same argument in Example 4 and Example 5, each F_n is a nowhere dense set. Then X is a set of first category since it is a subset of F. \square

Reference

  1. Bartoszyński T., Tsaban B., Hereditary topological diagonalizations and the Menger–Hurewicz Conjectures, Proc. Amer. Math. Soc., 134 (2), 605-615, 2005.
  2. Hurewicz W., Uber die Verallgemeinerung des Borelschen Theorems, Mathematische
    Zeitschrift, 24, 401-425, 1925.
  3. Just W., Miller A. W., Scheepers M., Szeptycki P. J., Combinatorics of
    open covers (II)
    , Topology Appl., 73, 241–266, 1996.
  4. Menger K., Einige Uberdeckungssatze der Punltmengenlehre, Sitzungsberichte Abt. 2a, Mathematik, Astronomie, Physik, Meteorologie und Mechanik (Wiener Akademie, Wien), 133, 421-444, 1924.
  5. Miller A. W., Fremlin D. H., On some properties of Hurewicz, Menger, and Rothberger, Fund. Math., 129, 17-33, 1988.
  6. Scheepers M., Combinatorics of open covers I: Ramsey theory, Topology Appl., 69, 31-62, 1996.
  7. Scheepers M., Selection principles and covering properties in topology, Note di Matematica, 22 (2), 3-41, 2003.
  8. Scheepers M., Selection principles in topology: New directions, Filomat, 15, 111-126, 2001.
  9. Sierpinski W., Sur un probleme de M. Menger, Fund. Math., 8, 223-224, 1926.
  10. Tsaban B., Menger’s and Hurewicz’s Problems: Sulutions From “The Book” and Refinements, Contemporary Math. 533, 211–226, 2011.
  11. Tsaban B., Zdomskyy L., Scales, fields, and a problem of Hurewicz, J. Eur. Math. Soc. 10, 837–866, 2008.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

Menger spaces are Lindelof-like

This post extends two previous posts on Menger spaces (here and here). We show that the Menger property, though stronger than the Lindelof property, is equivalent to a Lindelof-like property.

Let X be a space. The space X is a Menger space (has the Menger property) if for each sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \bigcup_{n \in \omega} \mathcal{V}_n is an open cover of X. It is sometimes easier to work with an equivalent definition of Menger spaces (found in Theorem 2 here).

The space X is a Hurewicz space (has the Hurewicz property) if for each sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \{ \cup \mathcal{V}_n: n \in \omega \} is a \gamma-cover of X, i.e. each x \in X belongs to \cup \mathcal{V}_n for all but finitely many n.

It is clear from definitions that

    \displaystyle \begin{aligned} &  \sigma \text{-} \bold C \bold o \bold m \bold p \bold a \bold c \bold t  \Longrightarrow \bold H \bold u \bold r \bold e \bold w \bold i \bold c \bold z \Longrightarrow \bold M \bold e \bold n \bold g \bold e \bold r \Longrightarrow \bold L \bold i \bold n \bold d \bold e \bold l \bold o \bold f   \end{aligned}

The Menger property is strictly stronger than Lindelof. See here for examples that are Lindelof but not Menger. One of them is the space \mathbb{P} of all irrational numbers. These examples show that having the property that every open cover has a countable subcover does not quite rise to the Menger property. Consider the following theorem.

Theorem 1
Let X be a space. Then the space X is a Menger space, if, and only if, for every Menger subspace M of X, and for every G_\delta-subset G_M of X such that M \subset G_M, the cover \{ G_M: M \subset X \text{ and } M \text{ is Menger} \} has a countable subcover.

The proof of Theorem 1 is found at the end of the post. Given a cover described in Theorem 1, every Menger subset is contained in one element of the cover. The covers in Theorem 1 can be called G_\delta covers since the elements of such covers are G_\delta sets. This is a “Lindelof” property using G_\delta covers. To get better intuition about this “Lindelof” characterization, we look at some examples.

Example 1
Any \sigma-compact space satisfies the covering property in Theorem 1.

In a \sigma-compact space, each of the countably many compact sets that makes up the space is a Menger subspace. Suppose that X=\bigcup_{n \in \omega} X_n where each X_n is compact. For any G_\delta cover \{ G_M: M \subset X \text{ and } M \text{ is Menger} \}, \{ G_{X_0}, G_{X_1}, G_{X_2}, \cdots \} would be a countable subcover. Similarly, any space that is the union of countably many Menger subspaces would satisfy the property in Theorem 1.

Example 2
Let X \subset \mathbb{R}. The set X is \sigma-compact if and only if X is an F_\sigma-set. This is because the real line is \sigma-compact. Thus any F_\sigma-subset of the real line satisfies the covering property in Theorem 1.

Example 3
The space \mathbb{P} of all irrational numbers is not a Menger space. Thus it does not satisfy the covering property in Theorem 1.

The set \mathbb{P} is homeomorphic to \omega^\omega, which is a dominating set and thus not Menger. See Theorem 4 in this post. The set of irrational numbers \mathbb{P} is a G_\delta-set that is not F_\sigma. The next example discusses other Borel sets that are not F_\sigma.

Example 4
Among the Borel sets in the real line, F_\sigma sets are the only ones that have the Menger property. Thus any Borel set that is not F_\sigma is not Menger. The only Borel sets satisfying the covering property in Theorem 1 are precisely the F_\sigma sets.

The Borel sets in the real line are constructed using the open sets and closed sets through the operations of countable unions and countable intersections. The following shows the Borel sets from the first few steps.

    open sets, \text{ } closed sets, F_\sigma, \text{ } G_\delta, \text{ } G_{\delta \sigma}, \text{ } F_{\sigma \delta}, \text{ } F_{\sigma \delta \sigma}, \text{ } G_{\delta \sigma \delta}, \text{ } \cdots

In the above steps, the subscript \sigma denotes union and the subscript \delta denotes intersection. After the steps for the finite ordinals are completed (n < \omega), the construction continues on through all the countable ordinals \alpha<\omega_1 (through taking countable unions and taking countable intersections of all the Borel sets that come before the \alphath step). We claim that the only Borel sets that are Menger are the F_\sigma-sets, which of course include the open sets and the closed sets. This fact follows from the following theorem [3].

    Theorem 2
    Let X \subset \mathbb{R}. If the set X is an analytic set, then either X is \sigma-compact or X contains a closed copy of the space \mathbb{P} of all irrational numbers.

A subset X of the real line is an analytic set if it is the continuous image of \mathbb{P}, the space of all irrational numbers. Any Borel set is an analytic set. Closed subsets of any Menger space are Menger. Thus for any analytic set X, if X is Menger, it cannot have \mathbb{P} as a closed subspace and thus must be \sigma-compact. It follows that Borel sets that are not F_\sigma are not Menger spaces.

Example 5
Borel sets are “nice” sets. For example, they are measurable sets. Let’s look at a pathological example. We turn to the Bernstein sets. We show that any Bernstein set does not have the Menger property and hence does not satisfy the covering property in Theorem 1. A subset B of the real line is a Bernstein set if both B and the complement of B intersect every uncountable closed subset of the real line. Such a set can be constructed using transfinite induction (see here for an illustration).

Let B \subset \mathbb{R} be a Bernstein set. Note that the complement \mathbb{R} \backslash B is also a Bernstein set. Any Bernstein set is dense in the real line. Choose a countable dense subset D of \mathbb{R} \backslash B. Note that \mathbb{R} \backslash D is a dense G_\delta-set and is homeomorphic to \mathbb{P}, the set of all irrational numbers. The set \mathbb{P} is in turn homeomorphic to \omega^\omega. With B \subset \mathbb{R} \backslash D, we can consider B as a subset of \omega^\omega.

Let B \subset \omega^\omega be a Bernstein set. We show that it is a dominating set. By Theorem 4 in this post, it is not a Menger set. To this end, let f \in \omega^\omega. Consider the compact subset C of \omega^\omega where C is the following set.

    C=\{ f(0)+1, f(0)+2 \} \times \{ f(1)+1, f(1)+2 \} \times \{ f(2)+1, f(2)+2 \} \times \cdots

The set C is a Cantor set and hence is an uncountable closed set. Thus B \cap C \ne \varnothing. Let g \in B \cap C. It is clear that f \le^* g. Thus B is a dominating set and hence not Menger.

Remarks

There exist uncountable subsets of the real line that have the Menger property but are not \sigma-compact (see here). These are ZFC counterexamples to the conjecture of Menger. In 1924, Karl Menger conjectured that in metric spaces, Menger spaces are precisely the \sigma-compact spaces. Based on the discussion for the above examples, counterexamples to Menger’s conjecture are not to be found among Borel sets or analytic sets.

As an application of Theorem 1, consider a notion called Alster spaces. Let X be a space. A collection \mathcal{G} of G_\delta-subsets of X is called an Alster cover if for any compact subset K of X, K \subset G for some G \in \mathcal{G}. Such a collection of G_\delta-sets is necessarily a cover of the space X. A space X is said to be an Alster space if for every Alster cover \mathcal{G} of X, there is a countable \mathcal{U} \subset \mathcal{G} such that \mathcal{U} is a cover of X. In other words, such a space has the property that every Alster cover has a countable subcover. Alster spaces were introduced by K. Alster [1] to characterize the class of productively Lindelof spaces (a space L is productively Lindelof if L \times Y is Lindelof for every Lindelof space Y).

With the G_\delta covering of Menger spaces (as described in Theorem 1), there is a clear connection between Alster spaces and Menger spaces. Observe that any G_\delta covering as described in Theorem 1 is also an Alster cover. Thus it follows that every Alster space is a Menger space. Menger spaces are D-spaces [2]. Thus Alster spaces are D-spaces. Thus the notion of Menger spaces is connected to the notion of Alster spaces and other notions that are used in the study of productively Lindelof spaces.

The Proof Section

We now give a proof of Theorem 1.

Theorem 1
Let X be a space. Then the space X is a Menger space, if, and only if, for every Menger subspace M of X, and for every G_\delta-subset G_M of X such that M \subset G_M, the cover \{ G_M: M \subset X \text{ and } M \text{ is Menger} \} has a countable subcover.

The direction \Rightarrow is clear. If X is Menger, then G_X belongs to any G_\delta cover \{ G_M: M \subset X \text{ and } M \text{ is Menger} \}.

\Leftarrow
Let \{ \mathcal{U}_n: n \in \omega \} be a sequence of open covers of X. For each Menger M \subset X, each \mathcal{U}_n is a open cover of M. Thus we can choose, for each n, a finite \mathcal{V}_n^M \subset \mathcal{U}_n such that for each x \in M, x \in \bigcup \mathcal{V}_n^M for infinitely many n. We are using a characterization of Menger space found here (see Theorem 2). Define G_M=\bigcap_{n \in \omega} W_n^M where each W_n^M is defined as follows:

    W_n^M=\bigcup_{j \ge n} (\bigcup \mathcal{V}_j^M)

Note that for each n, M \subset W_n^M. Thus M \subset G_M. Consider the G_\delta cover just defined.

    \{ G_M: M \subset X \text{ and } M \text{ is Menger} \}

By assumption it has a countable subcover \{ G_{M_0},G_{M_1},G_{M_2},\cdots \}. For each n, define \mathcal{V}_n by \mathcal{V}_n=\bigcup_{k \le n} \mathcal{V}_n^{M_k}. Note that each \mathcal{V}_n is finite and \mathcal{V}_n \subset \mathcal{U}_n. We now show that for each x \in X, we have x \in \bigcup \mathcal{V}_n for infinitely many n.

Let x \in X. Then x \in G_{M_j}=\bigcap_{n \in \omega} W_n^{M_j} for some j. By the definition of W_n^{M_j}, we have x \in \bigcup \mathcal{V}_m^{M_j} for infinitely many m. Then it follows that x \in \bigcup \mathcal{V}_m^{M_j} for infinitely many m where m \ge j. Note that x \in \bigcup \mathcal{V}_m^{M_j} for m \ge j means that x \in \bigcup \mathcal{V}_m. Therefore, x \in \bigcup \mathcal{V}_m for infinitely many m. \square

Reference

  1. Alster K., On the class of all spaces of weight not greater than \omega_1 whose cartesian product with every Lindelof space is Lindelof, Fund. Math., 129, 133–140, 1988.
  2. Aurichi L., D-Spaces, Topological Games, and Selection Principles, Topology Proc., 36, 107–122, 2010.
  3. Kechris A. S., Classical descriptive set theory, Graduate Texts in Mathematics, 156, Springer-Verlag, New York, 1995.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma

Counterexamples to the conjecture of Menger

This post is a continuation of the preceding post on Menger and Hurewicz spaces. We aim to provide the details on Example 4 and Example 5 in that post. These are counterexamples to Karl Menger’s conjecture concerning Menger spaces.

In 1924, Karl Menger introduced a basis covering property in metric spaces. He conjectured that among metric spaces this basis covering property is equivalent to \sigma-compactness. In 1925 Witold Hurewicz introduced a selection principle that led to a definition called Menger space as defined in the preceding post and below, which he showed to be equivalent to Menger’s basis covering property. In 1928, Sierpinski showed that Continuum Hypothesis (CH) implies that Menger’s conjecture is false. The counterexample is the Lusin sets. This is Example 4 in the preceding post, which is further discussed here. Lusin sets can be constructed using CH. On the other hand, Lusin sets do not exist under Martin’s axiom and the negation of CH. Thus Lusin sets, as counterexample to Menger’s conjecture, are only consistent counterexamples.

In a 1988 paper, Miller and Fremlin [3] gave the first ZFC counterexample to Menger’s conjecture. This is the subject of Example 5 in the preceding post. We plan to give a more detailed account of this example here.

A space X is a Menger space (or has the Menger property) if for every sequence \{ \mathcal{U}_n: n \in \omega \} of open covers of X, there exists, for each n, a finite \mathcal{V}_n \subset \mathcal{U}_n such that \bigcup_{n \in \omega} \mathcal{V}_n is an open cover of X. If in the definition the \mathcal{V}_n can always be made to consist of only one element, then the space is said to be a Rothberger space. It is clear from definition that \sigma-compact \rightarrow Menger \rightarrow Lindelof. On the other hand, Rothberger \rightarrow Menger. The following diagram is from the preceding post.

Figure 1

      \displaystyle \begin{aligned} &  \sigma \text{-} \bold C \bold o \bold m \bold p \bold a \bold c \bold t  \\&\ \ \ \ \ \downarrow \\& \bold H \bold u \bold r \bold e \bold w \bold i \bold c \bold z  \\&\ \ \ \ \ \downarrow \\& \bold M \bold e \bold n \bold g \bold e \bold r  \ \ \ \ \ \leftarrow \ \ \ \bold R \bold o \bold t \bold h \bold b \bold e \bold r \bold g \bold e \bold r \\&\ \ \ \ \ \downarrow \\& \bold L \bold i \bold n \bold d \bold e \bold l \bold o \bold f  \end{aligned}

Example 4 – Lusin Sets

In this example, we show that any Lusin set is a Menger space that is not \sigma-compact. This fact is established through a series of claims listed below. The proofs are found at the end of the post.

Let X \subset \mathbb{R} be uncountable. The set X is said to be a Lusin set (alternative spelling: Luzin) if the intersection of X with any first category subset of the real line is countable, i.e. for every first category subset A of \mathbb{R}, X \cap A is countable. In other words, an uncountable subset X of the real line is a Lusin set if every first category subset of the real line can contain at most countably many points of X. Thus any Lusin set (if it exists) must necessarily be a subset of the real line that is of second category. Sets of first category are also called meager sets. For more information about Lusin sets, see [2].

Let X \subset \mathbb{R}. Let B \subset \mathbb{R}. The set X is said to be concentrated on the set B if for any open set U with B \subset U, X \backslash U is countable. In other words, X is concentrated on the set B if any open set containing the B contains all but countably many points of X.

Claim 1
Let X \subset \mathbb{R}. If X is a Lusin set, then X is concentrated on any countable dense subset of X.

Claim 2
Let X \subset \mathbb{R}. Suppose that D is a countable dense subset of X. Then if X is concentrated on the set D, then X is a Rothberger space and hence a Menger space.

Claim 3
Let X \subset \mathbb{R}. If X is an uncountable \sigma-compact, then X contains a Cantor set.

Note. By Cantor set here, we mean any subset of the real line that is homeomorphic to the middle third Cantor in the unit interval or the product space \{0, 1 \}^\omega.

Claim 4
Let X \subset \mathbb{R}. If X is a Lusin set, then X is a Menger space that is not \sigma-compact.

Lusin sets can be constructed under CH while they cannot exist under Martin’s axiom and the negation of CH. Thus any Lusin set is a consistent counterexample to Menger’s conjecture.

Example 5 (Miller and Fremlin)

The example makes use of two cardinals \mathfrak{b} and \mathfrak{d} that are associated with subsets of \omega^\omega, which is the set of all functions from \omega into \omega. Define an order \le^* on \omega^\omega as follows. For any f, g \in \omega^\omega, we say f \le^* g if f(n) \le g(n) for all but finitely many n. A subset A of \omega^\omega is a bounded set if there exists some f \in \omega^\omega such that for all g \in A, g \le^* f, i.e. f is an upper bound of A according to \le^*. The set A is said to be unbounded if it is not bounded. The set A is a dominating set if for each f \in \omega^\omega, there exists g \in A such that f \le^* g.

The cardinal \mathfrak{b} is called the bounding number and is the least cardinality of an unbounded subset of \omega^\omega. The cardinal \mathfrak{d} is called the dominating number and is the least cardinality of a dominating subset of \omega^\omega. It is always the case that \omega_1 \le \mathfrak{b} \le \mathfrak{d} \le \mathfrak{c} where \mathfrak{c} is the cardinality of the continuum. For more information on bounded and dominating sets and the associated cardinals, see the preceding post or [4].

Furthermore, the cardinal \text{non}(\text{Menger}) is the least cardinality of a non-Menger subset of the real line. It is shown in the preceding post that \mathfrak{d}=\text{non}(\text{Menger}). As a result, any subset of the real line with cardinality less than \mathfrak{d} must be a Menger space. It is consistent with ZFC that \mathfrak{b} < \mathfrak{d}. When this happens, any subset of the real line with cardinality \omega_1 is a Menger space that is not \sigma-compact. But this by itself is only a consistent example and not a ZFC example. We continue to fine tune until we get a ZFC example.

Another notion that is required is that of a scale. Let X \subset \omega^\omega. The set X is said to be a scale if it is a dominating set and is well ordered by \le^*. The following claims will produce the desired example. In proving the following claims, we make of the fact that the set of irrational numbers, as a subset of the real line, is homeomorphic to the product space \omega^\omega. Thus in dealing with a set of irrational numbers, we sometimes think of it as a subset of \omega^\omega.

Claim 5
Let X \subset \omega^\omega. If X is a dominating set of cardinality \omega_1, then there exists Y \subset X such that Y is a scale.

Claim 6
Let Y \subset \omega^\omega be compact. Then Y is a bounded set.

Claim 7
Let X \subset \omega^\omega. If X is a scale of cardinality \omega_1, then X is concentrated on \mathbb{Q} where \mathbb{Q} is the set of all rational numbers.

Claim 8
If X is an uncountable \sigma-compact subset of the real line, then X contains a Cantor set C such that C contains no rational number.

Note. By Cantor set here, we mean any subset of the real line that is homeomorphic to the middle third Cantor in the unit interval or the product space \{0, 1 \}^\omega.

Claim 9
There exists a subset X of the real line with \lvert X \lvert=\omega_1 such that X is a Menger space that is not \sigma-compact.

\text{ }

\text{ }

\text{ }

The Proof Section

Proof of Claim 1
Let X be a Lusin set. Let D be any countable dense subset of X. Let U be any open subset of the real line such that D \subset U. Let W=U \cup (\mathbb{R} \backslash \overline{X}). Note that W is a dense open subset of the real line. Then \mathbb{R} \backslash W is a closed nowhere dense subset of the real line. Since X is a Lusin set, \mathbb{R} \backslash W can contain at most countably many points of X. Note that \mathbb{R} \backslash \overline{X} contains no points of X. The set \mathbb{R} \backslash U can contain at most countably many points of X. It follows that X \backslash U is countable. This shows that X is concentrated on any countable dense subset of X. \square

Proof of Claim 2
Let X \subset \mathbb{R}. Suppose that D is a countable subset of X such that D is a dense subset of X. Suppose X is concentrated on the set D. We show that X is a Rothberger space. To this end, let \{ \mathcal{U}_n: n \in \omega \} be a sequence of open covers of X. Enumerate the set D=\{ d_0, d_2, d_4, \cdots \} (indexed by the even integers). For each even n, choose V_n \in \mathcal{U}_n such that d_n \in V_n. Each V_n is an open subset of X. Find W_n, open in the real line, such that V_n=W_n \cap X. Let W=\bigcup_{\text{even } n} W_n. Note that D \subset W. Since X is concentrated on D, there are at most countably many points of X not in W. Say, the set of these points is labeled as \{d_1, d_3, d_5, \cdots \} (indexed by the odd integers). Then for each odd n, choose V_n \in \mathcal{U}_n such that d_n \in V_n. Overall, we can choose, for each n, one set V_n \in \mathcal{U}_n such that \{V_0,V_1,V_2,V_3,\cdots \} is a cover of X. This shows that X is a Rothberger space, hence a Menger space. \square

Proof of Claim 3
Let X be an uncountable \sigma-compact subset of the real line. Let X=\bigcup_{n \in \omega} Y_n where each Y_n is a compact subset of the real line. Then for at least one n, Y_n is uncountable. Let Y_n be one such. It is well known that any uncountable closed subset of the real line contains a Cantor set. For an algorithm of how to construct a Cantor set, see here. \square

Proof of Claim 4
This is where we put the preceding three claims together to prove that any Lusin set is a desired counterexample. Let X be a Lusin set. By Claim 1, X is concentrated on any countable dense subset of X. Since there is some countable subset D \subset X such that D is dense in X, by Claim 2, X is a Menger space.

If X is \sigma-compact, then by Claim 3 X would contain a Cantor set C. The set C is a nowhere dense subset of the real line. Because X is a Lusin set, X can have at most countably many points in C (a contradiction). Thus X cannot be \sigma-compact. \square

Proof of Claim 5
Let X=\{ x_\alpha: \alpha<\omega_1 \} be a dominating set of size \omega_1. We now derive Y=\{ y_\alpha: \alpha<\omega_1 \} \subset X such that Y is also a dominating set and such that for any \alpha<\beta, we have y_\alpha \le^* y_\beta. We derive by induction. First let y_0=x_0. Suppose that for \beta<\omega_1, Y_\beta=\{ y_\alpha \in X: \alpha<\beta \} has been chosen such that for any \gamma<\tau<\beta, we have y_\gamma \le^* y_\tau. The set Y_\beta \cup \{ x_\beta \} is a countable subset of \omega^\omega. As a countable set, it is bounded. There exists f \in \omega^\omega such that for each g \in Y_\beta \cup \{ x_\beta \}, we have g \le^* f. Since X is dominating, there exists x_\mu \in X such that f \le^* x_\mu. Let y_\beta=x_\mu. Now Y=\{ y_\alpha: \alpha<\omega_1 \} has been inductively derived. It is clear that Y is a scale. \square

Proof of Claim 6
Let Y \subset \omega^\omega be compact. Let B_n be the projection of Y into the nth coordinate \omega. The set B_n must be a finite set since \omega is discrete. Note that Y \subset \prod_{n \in \omega} B_n. Furthermore, \prod_{n \in \omega} B_n is a bounded set. \square

Proof of Claim 7
Let X=\{ x_\alpha: \alpha<\omega_1 \} be a scale of size \omega_1. Let U be an open subset of the real line such that \mathbb{Q} \subset U. For each positive integer n, consider the set K_n=[-n,n] \backslash U. The set K_n is a compact subset of the real line consisting entirely irrational numbers. Thus we can consider each K_n as a subset of \omega^\omega. By Claim 6, any compact subset of \omega^\omega is bounded. Thus for each n, there exists f_n \in \omega^\omega such that f_n is an upper bound of K_n, i.e. for each h \in K_n, we have h \le^* f_n. For each n, let g_n \in X such that f_n \le^* g_n (using the fact that X is dominating).

For each n, let X_n=X \cap K_n. Note that X \backslash U=\bigcup_{n \ge 1} X_n. Furthermore, for each h \in X_n, we have h \le^* f_n \le^* g_n. Since X is a scale of size \omega_1, there are only countably many points in X that are \le^* g_n. Thus each X_n is countable. As a result, X \backslash U is countable. This proves that the scale X is concentrated on the set \mathbb{Q} of rationals. \square

Proof of Claim 8
This is similar to Claim 3, except that this time we wish to obtain a Cantor set containing no rational numbers. Let enumerate the rational numbers in a sequence \{ r_0, r_1, r_2, \cdots \}. In the algorithm indicated here, at the nth step of the inductive process, we can make sure that the chosen intervals miss r_n. The selected intervals would then collapse in a Cantor set consisting entirely of irrational numbers. \square

Proof of Claim 9
This is the step where we put everything together to obtain ZFC example of Menger but not \sigma-compact. Recall the dominating number \mathfrak{d}. We consider two cases. Case 1. \omega_2 \le \mathfrak{d}. Case 2. \mathfrak{d}=\omega_1. The first case is that \mathfrak{d} is at least \omega_2, the second uncountable cardinal. The second case is that \mathfrak{d} is less than \omega_2, hence is \omega_1, the first uncountable cardinal. In either case, we can obtain a subset of the real line that is Menger but not \sigma-compact.

Case 1. \omega_2 \le \mathfrak{d}
Then any subset of the real line with cardinality \omega_1 must be Menger (see Theorem 10 in the preceding post). Such a set cannot be \sigma-compact. If it is, it would contain a Cantor set which has cardinality continuum (see Claim 3). In this scenario, \omega_1 is strictly less than continuum since \omega_1 < \omega_2 \le \mathfrak{d} \le \mathfrak{c}.

Case 2. \mathfrak{d}=\omega_1
Let Y \subset \omega^\omega be a dominating set of cardinality \omega_1. This set is not a Menger space. If a subset of \omega^\omega is a Menger space, it must be a non-dominating set (see Theorem 4 in the preceding post). By Claim 5, we can assume that Y is a scale. Let X=Y \cup \mathbb{Q}. By Claim 7, X is concentrated on \mathbb{Q}. By Claim 2, X is a Menger space.

We claim that X is not \sigma-compact. Suppose X=\bigcup_{n \in \omega} H_n where each H_n is compact. Then H_m is uncountable for some m. By Claim 8, there is a Cantor set C such that C \subset H_n and that C contains no rational numbers. Let U=\mathbb{R} \backslash C. Here U is an open set containing the rational numbers but uncountably many points of X are outside of U. This means that X is not concentrated on \mathbb{Q}, going against Claim 7. Thus X is not \sigma-compact. \square

Remarks on Example 5

At heart, the derivation of Example 5 is deeply set-theoretic. It relies heavily on set theory. Just that when the set theory goes one way, there is a counterexample and when the set theory goes the opposite way, there is also a counterexample. Taken altogether, it appears no extra set theory is needed. However, when each case is taken by itself, extra set theory is required. Case 2 says there is a scale of cardinality \omega_1. This is possible when \mathfrak{b}=\mathfrak{d}=\omega_1. This equality is consistent with ZFC. It is also consistent that \mathfrak{b}=\mathfrak{d}=\kappa where \kappa is a regular cardinal and \kappa>\omega_1. It is also consistent that \omega_1=\mathfrak{b}<\mathfrak{d} or \omega_1<\mathfrak{b}<\mathfrak{d}. In general, a scale exists when \mathfrak{b}=\mathfrak{d} (the scale is then of cardinality \mathfrak{b}). So Case 2 is possible because of set theory. For more information about the bounding number \mathfrak{b} and the dominating number \mathfrak{d}, see [4].

Case 1, being the opposite of Case 2, is also deeply set-theoretically sensitive. However, the two cases taken together have the effect that set theory beyond ZFC is not used. This argument is a dichotomy. The dichotomic proof covers all bases. The manipulation of set theory to produce a ZFC result is clever. It is indeed a very nifty proof! Though the proof is not constructive, the example shows that Menger’s conjecture is false in ZFC. There are better counterexamples, e.g. [1]. However, the example of Miller and Fremlin shows that Menger’s conjecture is indeed false in ZFC.

Reference

  1. Bartoszyński T., Tsaban B., Hereditary topological diagonalizations and the Menger–Hurewicz Conjectures, Proc. Amer. Math. Soc., 134 (2), 605-615, 2005.
  2. Miller, A. W., Handbook of Set-Theoretic Topology (K. Kunen and J. E. Vaughan, eds), Elsevier Science Publishers B. V., Amsterdam, 201-233, 1984.
  3. Miller A. W., Fremlin D. H., On some properties of Hurewicz, Menger, and Rothberger, Fund. Math., 129, 17-33, 1988.
  4. Van Douwen, E. K., The Integers and Topology, Handbook of Set-Theoretic Topology (K. Kunen and J. E. Vaughan, eds), Elsevier Science Publishers B. V., Amsterdam, 111-167, 1984.

\text{ }

\text{ }

\text{ }

Dan Ma topology

Daniel Ma topology

Dan Ma math

Daniel Ma mathematics

\copyright 2020 – Dan Ma