On countability and representations

The topic of this paper is the subtle interplay between countability and representations. In particular, we establish that the definition of countability of a certain set $X$ crucially hinges on the associated equivalence relation $=_{X}$. Armed with this knowledge, we study well-known and basic principles about countable sets, going back to Cantor, Sierpiński, and König, working in Kohlenbach’s higher-order Reverse Mathematics. While these principles are relatively weak in second-order Reverse Mathematics, we obtain equivalences involving countable choice and Feferman’s projection principle. The latter are essentially the strongest axioms studied in higher-order Reverse Mathematics and usually only come to the fore when dealing with the uncountable.

💡 Research Summary

The paper “On countability and representations” investigates a subtle but fundamental issue in the foundations of mathematics: the definition of a countable set depends critically on the equivalence relation that is attached to the underlying collection of objects. While the classical definition in set theory simply requires an injection into the natural numbers, this approach ignores the fact that many mathematical objects (real numbers, points in metric spaces, functions in Lp‑spaces, etc.) admit multiple representations. Two such representations may be considered equal in the ambient structure even though they are syntactically different.

To capture this nuance the author introduces Definition 1.1: a set X equipped with an equivalence relation =ₓ is called countable if there exists a map Y : X → ℕ such that Y(x)=Y(y) implies x =ₓ y. In other words, the injection must respect the intrinsic notion of equality on X. The paper motivates this definition with concrete examples: a real number is an equivalence class of fast‑converging Cauchy sequences, a point in a metric space is an equivalence class under the metric‑zero relation, and functions in Lp are identified up to changes on a null set. Consequently, a singleton {x} may contain uncountably many distinct representations, yet it is still a single element with respect to =ₓ.

Having fixed this refined notion of countability, the author works within Kohlenbach’s higher‑order Reverse Mathematics (RM). The base theory RCAω⁰ (recursive comprehension for higher‑type objects) is extended with various higher‑type functionals: ∃² (Kleene’s quantifier), ∃³, the Suslin functional S², and Feferman’s projection principle (named BOOT). The paper carefully reviews the logical strength of these functionals, noting that ∃² yields ACA₀, S² yields Π₁¹‑CA₀, and ∃³ yields Z₂.

The central technical contribution is a series of equivalences between classical theorems about countable structures (originally due to Cantor, Sierpiński, König, Ginsburg‑Sands, etc.) and the strong higher‑order principles BOOT and QF‑AC₀¹ (quantifier‑free countable choice). Roughly, the results can be summarised as follows:

1. Cantor’s characterisation of countable linear orders – When the linear order (X,≤) is defined using Definition 1.1, the statement “every countable linear order embeds into ℕ” is equivalent over RCAω⁰ to BOOT. This shows that the embedding construction implicitly requires the projection principle.

2. Sierpiński’s theorem on countable metric spaces without isolated points – Reformulated with the metric‑zero equivalence, the theorem becomes equivalent to the conjunction of BOOT and QF‑AC₀¹. The proof needs to select, for each point, a code in ℕ while respecting the metric’s equality.

3. König’s infinity lemma for countable graphs – The classical lemma, when interpreted with Definition 1.1, is shown to be equivalent to Π₁¹‑CA₀ (or stronger) together with BOOT, indicating that the existence of an infinite path cannot be proved in weaker systems.

4. Ginsburg‑Sands theorem on suprema of continuous functions on countable second‑countable spaces – The supremum principle is provable in RCAω⁰ + BOOT + QF‑AC₀¹, demonstrating that continuity plus the refined notion of countability already yields the full strength of the theorem.

5. Other related principles – The paper also analyses bounded comprehension, induction schemata (IND₃), and a higher‑order version of arithmetic comprehension (Principle 1.8), showing how they fit into the hierarchy between ATR₀ and Π₁¹‑CA₀.

A noteworthy methodological point is the use of Hunter’s conservation results. By constructing term models that extend a given Π₁¹‑AC₀ model to a higher‑type model of ACAω⁰ + Principle 1.8, the author demonstrates that many of the higher‑order principles are conservative over their second‑order counterparts. This bridges the gap between higher‑order RM and the more familiar second‑order framework.

The paper also draws a philosophical parallel with Bishop’s constructive analysis, where a set is defined together with an explicit equality test. The author argues that the refined definition of countability mirrors this constructive stance: a set exists only when we have a procedure to recognise when two of its elements are equal.

In conclusion, the work shows that the seemingly innocuous choice of how we define equality on a set dramatically influences the logical strength of statements about countability. By adopting Definition 1.1, many classical “weak” theorems about countable objects become equivalent to strong higher‑order principles such as BOOT and countable choice. This reveals a hidden layer of logical power in the study of countable structures and opens new avenues for research in higher‑order Reverse Mathematics, particularly concerning the interplay between representations, equivalence relations, and foundational axioms.