J-2 Consistency Results in Topology, I: Quotable Principles 419 J-2 Consistency Results in Topology, I: Quotable Principles

J-2 Consistency Results in Topology, I: Quotable Principles 419 J-2 Consistency Results in Topology, I: Quotable Principles

hart v.2003/06/18 Prn:23/06/2003; 15:17 F:hartj02.tex; VTEX/EA p. 1 j-2 Consistency results in topology, I: Quotable principles 419 j-2 Consistency Results in Topology, I: Quotable Principles A noteworthy feature of general topology – as contrasted 1. The Continuum Hypothesis with geometry, number theory, and most other fields of mathematics – is that many of its fundamental questions are The best-known quotable principle is undoubtedly the Con- not decided by “the usual axioms of set theory”. What this tinuum Hypothesis, abbreviated as CH, which states that means is that if one codifies the set-theoretic assumptions the real line has minimum possible cardinality, i.e., c = ℵ used implicitly in ordinary mathematics and consolidates |R|=2 0 =ℵ1. It was proved consistent by Gödel in [5] these in a list of axioms – the usual axioms of set theory (see [Ku] for a proof) but before that numerous conse- – one is left with many statements that can neither be proved quences were derived from it, see, e.g., [7]. nor refuted on the basis of these axioms alone. In some It derives its strength from the facts that so many sets have sense such undecidability results are commonplace: every- cardinality c and the initial segments of ω1 are all countable; one knows that, say, the statement ϕ = (∀x)(∀y)(xy = yx) this makes for relatively easy transfinite constructions. To give the flavour we construct a Lusin set: an uncountable cannot be decided by “the usual axioms of Group Theory”. subset of R that meets every nowhere dense set in a count- Indeed, the group of integers is a model for “Group The- able set. It suffices to take an enumeration N : α<ω of ory plus ϕ”, which makes this conjunction consistent and α 1 the family of all closed nowhere dense sets and apply the ϕ non-refutable; the permutation group S3 does the same Baire Category Theorem each time to choose lα outside for ¬ϕ. { : } β<α Nα;then lα α<ω1 is the Lusin set. A similar con- A convenient list of axioms for Set Theory, one which struction will yield a strongly infinite-dimensional subspace has become standard is “ZFC” – Zermelo–Fraenkel set the- of the Hilbert cube all of whose finite-dimensional subsets ory, including the Axiom of Choice. See, e.g., [Ku] for a are countable. In [KV, Chapter 8, § 4] one finds more intri- detailed exposition. Thus there are important topological cate constructions, involving CH,ofhereditarily separable statements ϕ such that neither ϕ nor its negation follow nonLindelöf spaces (S-spaces)aswellashereditarily Lin- from ZFC. Establishing this is not as easy as in the case of delöf nonseparable spaces (L-spaces). Group Theory. While it is easy to prove the consistency of On occasion CH helps simply by counting: in combination ℵ ℵ Group Theory by providing models (groups) for it, the same with Jones’ Lemma CH (or even its consequence 2 0 < 2 1 ) is, by Gödel’s Second Incompleteness Theorem, in principle implies that in a separable normal space closed and dis- impossible for ZFC (or any other useful collection of axioms crete subsets are countable and hence that separable normal for set theory). Gödel’s theorem says that a consistency proof Moore spaces are metrizable. By contrast, CH implies there of a theory as strong as ZFC cannot be formalized in the the- is a non-metrizable normal Moore space [KV, Chapter 16]. ory itself. This explains why many results are formulated as Another counting example comes from the theory of uni- “if ZFC is consistent then so is ZFC plus ϕ” instead of sim- form spaces. It is well-known that in a uniform space the ply “ϕ is consistent with ZFC” and why, formally, we should finite uniform covers generate a uniformity again as do the be speaking of relative consistency proofs. Even if one is not countable uniform covers. The proofs can be generalized to ℵ interested in consistency results per se, it is nonetheless pru- show that any 1-sized uniform cover has a uniform star re- finement of cardinality c. Thus, CH implies that the ℵ -sized dent to be aware of them, lest one waste effort trying to prove 1 uniform covers generate a uniformity. In general, the Gen- a proposition that has a consistent negation. eralized Continuum Hypothesis (GCH), which states that There are two ways in which consistency results in to- + for all cardinals κ one has 2κ = κ , implies that for every κ pology are obtained. The first consists of proving implica- the uniform covers of cardinality less than κ generate a uni- tions between statements, where the antecedent is a state- formity. ment previously proven consistent with ZFC. Thus, the con- The Continuum Hypothesis also implies that there is an sequent itself is proven consistent with ZFC as well. The almost disjoint family of uncountable subsets of ω1 of car- ℵ present article deals with results like these and in partic- dinality 2 1 or, equivalently, that the cellularity of the space ℵ ular with the best-known combinatorial principles that oc- U(ω1) of uniform ultrafilters is the maximum possible 2 1 . cur as antecedents. In the second kind of consistency result Here ‘almost disjoint’ means that intersections are countable will be dealt with in the companion article; here one needs and an ultrafilter u on a set X is uniform if every element an intimate knowledge of how one actually proves (relative) has full cardinality, i.e., |U|=|X| for all U ∈ u. consistency results. The principal subjects in that article are See also the article on βN and βR for other applications forcing and large cardinals. of CH. hart v.2003/06/18 Prn:23/06/2003; 15:17 F:hartj02.tex; VTEX/EA p. 2 420 Section J: Influences of other fields 2. Martin’s Axiom less than c are meager and of measure zero. We discuss the consequences of CH mentioned above. Every consequence of CH is a potential theorem, as it can no Lusin sets no longer exist as every set of reals of cardi- longer be refuted. Some consequences are in fact equivalent nality less than c is of first category. It becomes harder to to CH, e.g., the existence of the special infinite-dimensional find S-andL-spaces. Indeed, MA +¬CH denies the exis- space above; such consequences are then automatically non- tence of compact such spaces and it is even consistent with provable because Cohen proved the consistency of the nega- MA +¬CH that no S-spaces exist. tion of CH (see the next article for an indication of the proof On the positive side: MA +¬CH implies that separable and [Ku] for a full treatment). To decide other consequences normal nonmetrizable Moore spaces exist. It also implies one would need proofs that avoid CH or principles that imply that there is a uniform space with a uniform cover of car- their negations. dinality ℵ1 without an ℵ1-sized uniform star refinement, to Martin’s Axiom (MA) is such a principle. It can be wit the subspace {x:x=1 (∀α)(xα 0)} of ∞(ℵ1) [6]. viewed as an extension of the Baire Category Theorem: it The latter paper does not mention MA +¬CH directly but the states that if X is a compact Hausdorff space with the count- proof needs a family A,ofsizeℵ2, of uncountable subsets able chain condition (the ccc for short) then the union of of ω1 such that for some fixed cardinal κ ℵ2 every sub- fewer than c many nowhere dense in X still has a dense com- family A of A of size κ has a finite intersection. To make plement. As such it is a consequence of CH but the conjunc- such a family one starts with an almost disjoint family B,of tion MA +¬CH is also consistent. size ℵ2, of uncountable subsets of ω1 (almost disjoint means Its original formulation, though more extensive, is ulti- distinct elements have a countable intersection); by repeated mately more useful: if P is a partially ordered set (a poset) application of [4, 42I] one shrinks the elements of B to pro- with the ccc and D is a family of fewer than c dense sets then duce the desired family A (with κ = 2). there is a filter G on P that meets all members of D.Twoel- The ccc is the weakest in a line of properties, the best- ements p and q of P are compatible if there is an element r known of these are σ -centered (corresponding to separa- with r p,q,andincompatible otherwise. An antichain is ble compact spaces) and countable (corresponding to com- a set of mutually incompatible elements and “P has the ccc” pact metrizable spaces). These in turn give rise to weaken- means every antichain is countable. A set D ⊆ P is a dense ings of MA that have generated interest of their own, since set if for every p there is d ∈ D with d p. Finally, a filter they provide the possibility of denying some of MAs conse- on P is a subset G that satisfies: if p,q ∈ G then there is quences while retaining others. r ∈ G with r p,q,andifp ∈ G and q p then q ∈ G. Martin’s Axiom for σ -centered partially ordered sets The equivalence between the two formulations becomes was shown to be equivalent to the purely combinatorial state- somewhat apparent when one thinks of P as representing the ment known variously as P(c) or p = c:ifF is a family, of open sets of the space X – a proof may be devised along cardinality less than c, of subsets of N with the strong finite the lines of the Stone Representation Theorem of Boolean intersection property, i.e., the intersection of every finite algebras.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    4 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us