Theories of real numbers and interpretability (Q744846)

This is a remarkable monograph, devoted to a beautiful subject that has never received this kind of in-depth, book-length coverage. It grew out of the author's doctoral dissertation in philosophy at the Humboldt University in Berlin. The subject matter is the interpretability of theories, with special emphasis on the theory of real closed fields (RCF). Here one thinks first and foremost of the interpretability of RCF in elementary geometry, be that Euclidean, hyperbolic, or elliptic, as championed by the classical works of \textit{W. Schwabhäuser} et al. [Metamathematische Methoden in der Geometrie. Teil I: Ein axiomatischer Aufbau der euklidischen Geometrie. Teil II: Metamathematische Betrachtungen. Hochschultext. Berlin etc.: Springer-Verlag (1983; Zbl 0564.51001)]. It turns out that this is far from being the whole story. In fact, there is much more to say about RCF, and the author has done an admirable job of collecting all results in the literature that refer to (non-)interpretability of RCF in some other theory or of some theories in RCF. The main motivation holding together the various results is a very simple one. Arithmetic, the algebra of real numbers, and geometry seem to be related, for the natural numbers are a part of the real numbers (the latter having been built from the former), and geometry allows, via some sort of segment calculus, the introduction of magnitudes and their addition and multiplication, which satisfy the axioms of ordered fields; with the assumption of a liberal dose of continuity, those ordered fields are real closed fields. So, when formalized, do these theories communicate in the manner reminiscent of the genesis of their standard models? There are multiple flavors of interpretability, such as higher-dimensional interpretability (with a mention of the negative results of \textit{S. Świerczkowski} [Trans. Am. Math. Soc. 322, No. 1, 315--328 (1990; Zbl 0715.03004)]), identity-modifying interpretability, and interpretability with parameters. There are interesting comparisons between RCF and various elementary arithmetics, such as the fact that PRA (primitive-recursive arithmetic) proves the consistency of RCF, that RCF can be identity-modifyingly interpreted in Robinson's $Q$ (as proved in [\textit{A. M. Fernandes} and \textit{F. Ferreira}, J. Symb. Log. 67, No. 2, 557--578 (2002; Zbl 1015.03056)]). It is also shown that RCF has definable Skolem functions (cp. [\textit{L. van den Dries}, J. Symb. Log. 49, 625--629 (1984; Zbl 0596.03032)]). There are many negative results, such as the fact that theories that have a pairing function and which prove the existence of at least two elements are not relatively interpretable in RCF, that neither the theory of atomic Boolean algebras with infinitely many atoms nor the theory of discrete orders is relatively interpretable in RCF (the latter result stemming from \textit{S. Świerczkowski} [Fundam. Math. 143, No. 3, 281--285 (1993; Zbl 0794.03018)]). It is shown, following \textit{R. Montague} [Deterministic theories. New Haven: Yale University Press (1974)], that both PA and $\text{RCF}^{\text{OP}}$ (which is an extension of RCF with a pairing function) are relatively interpretable RCFN, an extension of RCF, in a language with, in addition to $\{0, 1, +, \cdot, \leq\}$, a unary predicate $N$ ($N(x)$ having the intended interpretation `$x$ is a positive integer'). There are a number of results regarding second-order theories of real numbers and relative interpretability, as well as a discussion of the philosophical significance of these results.