How statement and proof provenance work
The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.
- Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
- AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
- AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.
These labels describe origin, not correctness: citations and verification chips remain separate evidence.
Choice Strength in Baire, Urysohn, Stone, and Tychonoff: Examples and Counterexamples
1 · Prerequisites
- Arithmetization, Incompleteness, and Relative Consistency
- Binary Operations, Monoids, Groups and Subgroups
- Boolean Algebras, Stone Duality, and the Prime Ideal Theorem
- Cardinal Arithmetic, Cofinality and the Alephs
- Choice Strength in Baire, Urysohn, Stone, and Tychonoff
- Compactness
- Compactness in Metric Spaces
- Completeness, Completion, and Uniform Continuity
- Condensation, GCH, and Diamond in L
- Construction of the Natural Numbers
- Construction of the Real Numbers via Cauchy Sequences
- Construction of the Real Numbers via Dedekind Cuts
- Convergence: Nets and Filters
- Countability and Uncountability
- Countability Axioms and Cardinal Functions
- Deduction, Soundness, Completeness, and Compactness
- Filters and Ultrafilters
- Finite Counting, Factorials and Binomial Coefficients
- Forcing Orders, Names, and Generic Extensions
- Formal Set-Theoretic Syntax, Structures, and Satisfaction
- Foundations of the Real Numbers for Analysis
- Group Actions, Orbits, Stabilisers and Cayley's Theorem
- Hausdorff via the Diagonal
- Metric Spaces
- Monotone Sequences, Bolzano-Weierstrass, and Cauchy Completeness
- Order, Zorn's Lemma, and the Axiom of Choice
- Ordinal Arithmetic and the First Uncountable Ordinal
- Ordinals, Cardinals, and Transfinite Recursion
- Partitions of Unity and Paracompactness
- Permutation Models and Transfer to ZF
- Ramsey Theory
- Reflection, Absoluteness, and Elementary Submodels
- Relations, Functions, and Quotients
- Roots, Rational Powers, and Classical Inequalities
- Separation Axioms: the Hierarchy
- Sequences and Limits
- Subspaces, Products, and Quotients
- Suprema and Infima
- The Arithmetical Hierarchy and Post's Theorem
- The Constructible Hierarchy and Inner Models
- The Forcing Theorem and Formal Consistency Transfer
- The ZFC Axioms and the Basic Set Constructions
- Topological Spaces and Continuity
- Urysohn's Lemma and the Tietze Extension Theorem
- Weak Choice Principles and Sierpiński's Theorem
- Well-Founded Relations, Rank, and the Cumulative Hierarchy
2 · Summary
Worked examples and counterexamples for the choice-strength pair. The page computes the canonical least-ball selection in the separable Baire proof, the finite-menu intersection in the DMC Urysohn construction, the failure of closedness for the cofinite coordinate set, and the isolated-point repair that recovers a choice function from a compact T1 product; it also records the false statement that the Boolean prime ideal principle proves Stone's theorem for metric spaces, refuted by the transferred Corson model.
3 · Logical flowchart
4 · Definitions, theorems and proofs
None yet.
5 · Examples, counterexamples and false statements
Canonical least-ball selection removes choice
Example
A coded least-pair rule is an alternative to the lexicographic rule in Separable complete metric spaces are Baire in ZF. Given a nonempty open and , call admissible when , and . Choose the admissible pair of least code under . The following three-stage calculation permits repetitions in the dense enumeration and spends no choice.
Facts & Assumptions
Given: Work in ZF. Let be a nonempty separable complete metric space, let have dense range, let be a specified sequence of dense open subsets of , and let be nonempty open. Set and use radius bound at stage .
Positive-radius balls contain their centers; open balls are open and finite intersections of open sets are open (Open ball, closed ball and sphere in a metric space, The metric topology: a set is open when every one of its points has a ball around it inside the set; closed means open complement, Arbitrary unions and finite intersections of open sets are open, open balls are open and closed balls are closed).
Density of the range of means it meets every nonempty open set (Separability: the existence of an at most countable dense subset). The metric triangle inequality and symmetry hold (Metric space: iff , symmetry, and the triangle inequality; pseudometric and ultrametric); for every some integer has (For every in a complete ordered field there is a natural with ).
The explicit in is a bijection . Every nonempty subset of has a least element (The well-ordering principle), so the least element of decodes to a unique member of every nonempty admissible set .
A specified self-map of a set and an initial state determine a unique natural-number sequence by The recursion theorem.
Verification
For nonempty open and , fix and with . By [F2] take with and with . For the triangle inequality gives . Thus , proving the admissible set nonempty. This is one finite existence argument for arbitrary , not a simultaneous choice of witnesses.
Density of makes nonempty, and it is open. Thus step 1.1 with bound supplies an admissible pair.
Decode the least code to and put , . Then . The set is nonempty by density of , because the open ball contains ; it is open by finite intersection.
Apply the same rule to with bound , and then to with bound . At each stage density of the specified next keeps nonempty open, and step 1.1 and [F3] supply the unique next pair. Repeated values of do not affect uniqueness of the index-radius pair.
For a concrete example take , , for every , and for every . The metric axioms hold, every sequence converges to , and the range of is dense, so the given hypotheses hold. All positive-radius open and closed balls equal . At stages admissibility is exactly . Since , with equality exactly when , the least pairs are , with codes and radii . All centers are the same point , as required for an enumeration with repetitions.
For the general recursion use the set of states with nonempty open, together with a default state. The unique least-code pair defines the successor state on each such state; let the default state map to itself. The preceding nonemptiness argument makes this a total self-map. Apply [F4] from and take the uniquely defined centers and radii. This is set recursion, not a choice of points from an arbitrary family; it gives the same closed-ball inclusions and radius bounds needed in the cited theorem, though its pairs need not equal the lexicographically least pairs there.
Kelley's cofinite set is not closed
Statement refuted
The following two related claims both fail, but they are not equivalent instances of one claim:
- in the cofinite space on an infinite set , every infinite subset with infinite complement is closed; and
- the coordinate set is closed in the cofinite topology on when is infinite (The discrete, indiscrete, cofinite, cocountable, particular-point and Sierpinski topologies, Topology on a set, open and closed sets, clopen sets, the closed-set axiomatisation, and the coarser/finer comparison).
Counterexample
Take with the cofinite topology (The natural numbers (von Neumann), Finite, countably infinite, countable, uncountable) and let be the set of even naturals. For the second claim, give its cofinite topology.
Facts & Assumptions
Given: The cofinite spaces on and , and the set of even naturals.
In the cofinite topology on a set , the open sets are and the sets with finite complement, and the closed sets are and the finite subsets; hence every finite set, in particular every singleton, is closed (The discrete, indiscrete, cofinite, cocountable, particular-point and Sierpinski topologies, A space is if and only if every singleton is closed, if and only if every finite subset is closed, if and only if its topology contains the cofinite topology, (Kolmogorov) and (Frechet) spaces).
The repaired coordinate of The isolated-point repair of Kelley's choice space is a different space: there is closed because the added point is isolated, which is why the cofinite presentation on is not the coordinate used in the product argument.
Verification
is infinite: the map is injective from onto , so is countably infinite.
, the set of odd naturals, is infinite: is injective from into it, so it is not finite.
In the cofinite space , the coordinate set is not closed. Indeed, its complement is the singleton ; this set is nonempty but is not open because its complement is infinite.
is not closed: if were closed then its complement would be open. It is nonempty because is odd, and it is not cofinite because its complement is infinite by step 1.1. Thus is neither empty nor cofinite, contrary to [F1].
Step 2.1 refutes the first claim using an infinite subset whose complement is infinite, whereas step 1.3 separately refutes the coordinate claim using a subset whose complement is finite. The isolated-point repair of [F2] meets the latter closedness obligation by making open.
The isolated-point repair recovers a choice function
Example
Take the three nonempty sets , and , and form the repaired coordinates of The isolated-point repair of Kelley's choice space. Assume the compact- product hypothesis: every product of compact spaces is compact. In the product the closed constraints have the finite intersection property, so compactness of produces a point whose three coordinates are a choice tuple.
Facts & Assumptions
Given: The three sets, their repaired coordinates , and the hypothesis that every product of compact spaces is compact.
Each is compact and is a closed subspace of it (The isolated-point repair of Kelley's choice space, (Kolmogorov) and (Frechet) spaces).
In a product, the cylinder is closed when is closed in the factor, and a space is compact exactly when every family of closed sets with the finite intersection property has nonempty intersection (The product set of functions choosing a point in each factor, the projections, the box topology, and the product topology as the initial topology of the projections; the empty product is a one-point space, A space is compact exactly when every family of closed subsets with the finite intersection property has nonempty intersection, Finite intersection property).
The hypothesis that every product of compact spaces is compact makes compact (The compact T1 product theorem is equivalent to AC, Open cover, subcover, and compact topological space; a compact subset is a subspace that is compact in its own right).
Let . If a product point satisfies for every , then for each let be the least with and define . The least index exists because occurs in the displayed finite list, and . Hence has domain and is a choice function on (Choice function, Every natural-number-indexed list of nonempty sets has a choice function on its family of values).
Verification
The three cylinders are closed in by [F1] and [F2].
Each single cylinder is nonempty: the point with one coordinate (or any other element of ) and the artificial values at the other two coordinates lies in it; the artificial values are available because for every .
For the pair the point with prescribed values in and and in the remaining coordinate lies in , so the family has the finite intersection property; for the triple the point lies in .
By [F3] the product is compact, so by [F2] the intersection is nonempty. Choose in this intersection. Then for all three indices, and the function defined in [L1] has domain and satisfies for every . Thus is the required choice function.
For a finite list of sets, the same least-index construction converts a point in the closed cylinders into a choice function on the underlying set-family. In the general AC argument the factors are instead indexed by the family itself, so a product point with directly defines the choice function ; compactness supplies such a point after finite choice verifies the cylinders' finite-intersection property. [step 3.1, F2, L1, Every natural-number-indexed list of nonempty sets has a choice function on its family of values] ∎
False: BPI proves Stone's theorem for metric spaces
Statement
False: over , BPI implies that every metrizable space is paracompact.
More precisely, the universal implication from BPI to Stone's theorem for metric spaces is not provable over : relative to there is a model of containing a metrizable space that is not paracompact (The Boolean prime ideal principle, Paracompactness: every open cover has a locally finite open refinement, with no separation axiom built into the word, Metric space: iff , symmetry, and the triangle inequality; pseudometric and ultrametric).
Refutation
Facts & Assumptions
Given: The relative-consistency theorem for BPI with a metrizable nonmetacompact space, and the assumed consistency of .
Relative to there is a model of containing a metrizable nonmetacompact space; by definition, that space has an open cover with no point-finite open refining cover (Relative consistency of BPI with failure of Stone's theorem, Metacompactness: every open cover has a point-finite open refinement, Refinements, locally finite families, point-finite families, and star refinements).
A paracompact space is one in which every open cover has a locally finite open refinement, and a locally finite family is point-finite (Paracompactness: every open cover has a locally finite open refinement, with no separation axiom built into the word, Refinements, locally finite families, point-finite families, and star refinements).
Proof
Assume, for the sake of contradiction, that over BPI implies that every metrizable space is paracompact, and assume .
In the model of [F1] the theory holds, so by the assumed implication every metrizable space in that model is paracompact; in particular its metrizable nonmetacompact space would be paracompact.
By [F1], some open cover of has no point-finite open refining cover. Paracompactness would give a locally finite open refinement covering , and would be point-finite by [F2], a contradiction.
The contradiction shows that BPI does not imply Stone's theorem for metric spaces over , conditionally on ; the refutation is relative-consistency based and does not exhibit an outright counterexample in ZF.