[Paper Review] Lifting countable to uncountable mathematics

Sam Sanders|arXiv (Cornell University)|Aug 15, 2019

Computability, Logic, AI Algorithms15 references4 citations

TL;DR

This paper demonstrates that proofs rooted in computability theory—originally formulated for countable structures like sequences—can be systematically 'lifted' to uncountable settings using nets (Moore-Smith sequences), yielding new, strong results in uncountable mathematics with minimal modification. The key contribution is showing that recursive counterexamples and reversals in Reverse Mathematics for countable objects directly imply powerful theorems in higher-order arithmetic, such as the monotone convergence theorem for nets implying the strong comprehension axiom BOOT.

ABSTRACT

Turing's famous 'machine' framework provides an intuitively clear conception of 'computing with real numbers'. A recursive counterexample to a theorem shows that the theorem does not hold when restricted to computable objects. These counterexamples are often crucial in establishing reversals in the Reverse Mathematics program. All the previous is essentially limited to a language that can only express countable mathematics directly. The aim of this paper is to show that reversals and recursive counterexamples, countable in nature as they might be, directly yield new and interesting results about uncountable mathematics with little-to-no modification. We shall treat the following topics/theorems: the monotone convergence theorem/Specker sequences, compact and closed sets in metric spaces, the Rado selection lemma, the ordering and algebraic closures of fields, and ideals of rings. The higher-order generalisation of sequence is of course provided by nets (aka Moore-Smith sequences ).

Motivation & Objective

• To establish a systematic method for transferring proofs from countable to uncountable mathematics using higher-order generalizations of sequences.
• To demonstrate that recursive counterexamples and reversals in Reverse Mathematics for countable objects yield new, non-trivial results in uncountable settings.
• To show that theorems about sequences (e.g., monotone convergence) can be naturally extended to nets in uncountable metric spaces, preserving logical strength.
• To explore the role of nets (Moore-Smith sequences) as the higher-order analogue of sequences in lifting proofs from countable to uncountable mathematics.
• To clarify the connection between classical Reverse Mathematics and higher-order arithmetic by showing that results in $\textup{{ACA}}_{0}^\omega$ imply strong comprehension axioms like BOOT in uncountable contexts.

Proposed method

• The paper uses higher-order arithmetic $\textup{{RCA}}_{0}^\omega$ and $\textup{{ACA}}_{0}^\omega$ as the logical framework to formalize uncountable mathematics.
• It applies the ECF-interpretation to translate higher-order statements into second-order arithmetic, enabling comparison with classical Reverse Mathematics.
• The core technique involves lifting proofs of theorems about sequences (e.g., monotone convergence) to analogous theorems about nets in uncountable metric spaces.
• It leverages the fact that continuous functionals on $2^\mathbb{N}$ admit countable representations, allowing recursive counterexamples to be transferred via ECF-translation.
• The paper uses the 'excluded middle trick'—$\exists^2 \vee \neg\exists^2$—to handle discontinuous functionals, ensuring results hold in both continuous and discontinuous cases.
• It establishes that theorems such as HBU and BOOT, when translated via ECF, reduce to known second-order principles like WKL₀ and ACA₀, respectively, under continuity assumptions.

Experimental results

Research questions

• RQ1 Can proofs from countable computability theory be systematically lifted to uncountable mathematics using nets as the higher-order analogue of sequences?
• RQ2 To what extent do recursive counterexamples in Reverse Mathematics for sequences yield new, strong results when generalized to nets in uncountable settings?
• RQ3 How do the logical strengths of theorems about sequences (e.g., MCT_seq) compare to their net-based counterparts (e.g., MCT_net) in higher-order arithmetic?
• RQ4 What is the role of the ECF-interpretation in connecting higher-order theorems (e.g., BOOT) to their second-order counterparts (e.g., ACA₀)?
• RQ5 Can the transfer of proofs from countable to uncountable mathematics be formalized as a general procedure, even when the original proof relies on discontinuous functionals?

Key findings

• The proof of $\textup{{MCT}}_{\textup{{seq}}}^{[0,1]} \rightarrow \textup{{ACA}}_{0}$ directly lifts to $\textup{{MCT}}_{\textup{{net}}}^{[0,1]} \rightarrow \textup{{BOOT}}$, showing that a net-based monotone convergence theorem implies a very strong comprehension axiom.
• The Rado selection lemma, when generalized to nets, yields a higher-order version that implies the same strong comprehension strength as in the countable case.
• For continuous functionals, the canonical covering in HBU reduces to a countable covering over rationals, showing that $[\textup{{HBU}}]_{\textup{{ECF}}}$ is equivalent to the Heine-Borel theorem for countable coverings.
• The ECF-interpretation translates $\textup{{BOOT}}$ to $\textup{{ACA}}_{0}$, and $[\textup{{HBU}}]_{\textup{{ECF}}}$ to WKL₀, confirming that the second-order counterparts of higher-order principles are well-known in Reverse Mathematics.
• The paper shows that most results in $\textup{{ACA}}_{0}^\omega$ also hold in $\textup{{RCA}}_{0}^\omega$ via the excluded middle trick, ensuring continuity-based reductions to second-order logic.
• The transfer of proofs from sequences to nets is not merely formal but yields genuinely new results in uncountable mathematics, such as new reversals in higher-order Reverse Mathematics.