[Paper Review] The real numbers - a survey of constructions

Ittay Weiss|arXiv (Cornell University)|May 18, 2015

Computability, Logic, AI Algorithms14 references3 citations

TL;DR

This paper presents a comprehensive, unified survey of 20+ mathematical constructions of the real numbers, ranging from Stevin's decimal expansions to modern approaches like Knopfmacher's series-based methods and Arthan's completion of ordered groups. It systematically compares these constructions by framing each as a bijective correspondence between reals and structured sets built from rationals or integers, highlighting their algebraic, order, and topological properties to enable direct comparison and clarify foundational nuances in constructive and computational contexts.

ABSTRACT

We present a comprehensive survey of constructions of the real numbers (from either the rationals or the integers) in a unified fashion, thus providing an overview of most (if not all) known constructions ranging from the earliest attempts to recent results, and allowing for a simple comparison-at-a-glance between different constructions.

Motivation & Objective

• To provide a comprehensive, unified overview of known constructions of the real numbers, enabling direct comparison across diverse approaches.
• To clarify foundational distinctions between constructions, especially in constructive and computational mathematics where isomorphism does not imply practical equivalence.
• To survey constructions from historical origins (e.g., Stevin, 1585) to modern variants (e.g., Knopfmacher, 1989), including nonstandard and surreal number-based approaches.
• To emphasize that while all constructions yield isomorphic complete ordered fields, differences emerge in effectiveness, computability, and implementability in automated systems.
• To present each construction uniformly—via a bijective correspondence between reals and structured sets built from rationals or integers—facilitating side-by-side analysis.

Proposed method

• Adopt a uniform framework: each construction is presented as a bijective correspondence between the reals and a set derived from simpler entities (e.g., rationals, integers), with order and arithmetic defined a posteriori.
• Use the extended natural numbers ℕ⁺ = ℕ ∪ {ω}, where ω behaves like ∞, to treat finite sequences as infinite by padding with ω, simplifying the treatment of series and expansions.
• For each construction, define the set of reals as a subset of sequences or equivalence classes satisfying specific algebraic or order-theoretic constraints (e.g., growth conditions in Engel or Sylvester-based constructions).
• Apply known theorems (e.g., Cantor’s theorem, Hölder’s theorem, Dedekind-MacNeille completion) to justify the field and order structure of the resulting sets.
• Present arithmetic and order operations explicitly for key constructions (e.g., in ℤ[√2] via (a,b)+(c,d)=(a+c,b+d), (a,b)(c,d)=(ac+2bd,ad+bc)), ensuring consistency with ℝ.
• Distinguish between constructions yielding ℝ₊ and those yielding ℝ, noting that the former often simplifies technicalities, and that both are considered complete if the full field structure is recoverable.

Experimental results

Research questions

• RQ1 How do various constructions of the real numbers—ranging from classical (Dedekind, Cantor) to modern (Knopfmacher, Schanuel)—compare in terms of foundational assumptions and structural properties?
• RQ2 To what extent do different constructions yield isomorphic complete ordered fields, and in what contexts do these isomorphisms fail to be computable or effective?
• RQ3 What role do specific number-theoretic theorems (e.g., Engel’s, Sylvester’s, alternating Sylvester’s) play in generating real numbers via infinite series or expansions?
• RQ4 How do nonstandard or surreal number constructions relate to the classical reals, and in what ways do they collapse to known constructions when restricted to the reals?
• RQ5 Can a single, uniform framework be applied to all constructions to enable direct comparison, particularly in the context of constructive mathematics and automated theorem proving?

Key findings

• All surveyed constructions yield a complete ordered field isomorphic to ℝ, confirming the categoricity of the axioms of the reals, though the isomorphism is not always effective or computable.
• The construction via Dedekind-MacNeille completion of a dense, archimedean, ordered commutative group (e.g., ℤ[√2]) yields ℝ as a field, provided multiplication can be effectively defined—this underpins Arthan’s approach.
• Knopfmacher and Knopfmacher’s multiple constructions (using Cantor’s, Engel’s, Sylvester’s, and alternating theorems) all produce ℝ by defining reals as infinite sequences satisfying growth or convergence conditions, with rationals identified via finite termination.
• De Bruijn’s additive expansion construction and Faltin et al.’s wreath product method both yield ℝ by encoding reals as infinite expansions or algebraic structures over ℤ, with arithmetic defined via recursive rules.
• Schanuel’s construction using approximate endomorphisms of ℤ produces ℝ as a completion of ℤ under a suitable metric, with the field structure derived from the endomorphism ring.
• Conway’s surreal numbers contain a copy of ℝ, but isolating the reals within them reduces to the Dedekind cut construction, indicating that surreal numbers do not offer a fundamentally new construction of ℝ.