A Foundation for the Core Mathematician
David Mumford, Sy-David Friedman · 2026
Read A Foundation for the Core Mathematician by David Mumford, Sy-David Friedman (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-03868.txt.
Introduction
David Mumford and Sy-David Friedman argue that the usual Zermelo-Fraenkel-Choice axioms do not settle the foundations of everyday mathematics. Set theory admits many models with different truths, and even classical questions such as the measurability of projective sets can vary between them. The core of mathematics, they contend, lies in structures built from the real numbers.
They propose a foundation for core mathematics consisting of both a system of axioms and a definite model of those axioms, in which essentially all core mathematics is incorporated. Because the model is definite, every core mathematical assertion receives a definite truth value, true or false, rather than depending on which set-theoretic universe one happens to inhabit.
Catalog record
| CREATOR | David Mumford, Sy-David Friedman |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.03868
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Foundation for the Core Mathematician David Mumford, Sy-David Friedman Abstract A large majority of research mathematicians consider the issue of the “Foundations of Mathematics” long settled, accepting the axiom system of Zermelo-Fraenkel with Choice (ZFC) as adequate for everything they do. Alternatives are the various intuitionist, constructivist and finitist systems as well as the more recent homotopy-theoretic univalent foundations. But most mathematicians are happy to use ZFC. However, the very word foundations means a firm and fixed non-contro-versial basis, which supports all further work. In particular, foundations should support the need of research mathematicians to believe that all conjectures they work on have an answer, right or wrong (the Truth-Platonist viewpoint). On the other hand, the dramatic developments in set theory regarding large cardinal axioms (see [ Ka ] ) and Cohen’’s method of forcing new models have resulted in a proliferation of set theories and models for the universe of sets. What set theory you adopt has significant consequences for questions in classical descriptive set theory. There is no consensus on how ‘‘high” you can go with infinities nor on other possible axioms 1 1 1 Another example is the axiom of determinacy (AD), stating that certain games involving 2 players making alternate choices from some set ad infinitum must have winning strategies, has dramatic consequences for descriptive set theory, but this is adopted by some set-theorists and rejected by others. There are also new forcing axioms, such as PFA (the proper forcing axiom) which have dramatic consequences. This vigorous debate over how best to extend ZFC to accommodate the developments alluded to above continues to the present day. Gödel believed in the Platonistic view that set theory has a unique and well-defined universe of “real world” sets, a view still supported by some set-theorists ( [ Wo ] ), while others feel that due to its openness and tolerance of different interpretations, set theory must accept a “multiverse” with a range of possible universes of set theory ( [ SFr ] , [ AFHT ] ). For these reasons, ZFC does not provide an adequate foundation for mathematics as a whole. But what do most mathematicians deal with in their work? The set \mathbb{R} of real numbers occupies a special place among all sets: it is used in the real world to measure space, time, mass and derivative quantities, using a unit and taking ratios. It is not unreasonable to say that a large majority of mathematical research concerns (i) the reals and/or (ii) the integers, together with structures derived from them both. Therein lies the “central core” of mathematical experience. The aim of this paper is to provide a better foundation for this “central core” of mathematics. In §1, we review how far you get without set theory. “Second order arithmetic” or Z_{2} turns out to be sufficient for a very large part of mathematics, including the use of trees to construct all Borel sets. But it lacks, for example, the set of all Borel sets. It feels like mental gymnastics to define topological spaces via a basis for open sets instead of a set of points. In §2, we introduce a major feature of our proposal, the use of intuitions based on probability theory and relativity as expressed in Freiling’s axiom A^{\mathbb{N}}_{null} [ CFr ] for the existence of certain sets of random numbers. He found that the infinite case of his axiom contradicted the uncountable use of the axiom of choice. We point out how unrestricted choice constructs many weird unworldly sets (and magic tricks) anyway. This paper chooses Freiling’s A^{\mathbb{N}}_{null} instead of uncountable choice. In §3, we spell out our axioms but, as pointed out above, we need a specific model because in models all statements are either true or false. In our approach, we take the set of real numbers as a given, modeling the physical world in the time-honored way, just as the set of natural. numbers \mathbb{N} is given by counting. Thus we adopt Gödel’s Platonistic view, not for arbitrary sets, but specifically for the set \mathbb{R} of real numbers. We include Freiling and countable choice. In §4, we describe the proposed model, in which all sets are given by constructions from \mathbb{N} and \mathbb{R} . But, as usual since Gödel’s work, these constructions can use transfinite induction based on ordinals. We include a quick primer on the theory of ordinals. We propose to stop higher induction at the first “fully \mathbb{R} -admissible ordinal”, i.e. as soon as the resulting model satisfies ZF– (Zermelo-Fraenkel without uncountable choice or the power set axiom). This is our main proposal for a model where every statement has a ground truth: it is reasonable to believe that all its sets can be treated as “real world” objects. Some of the arguments needed to prove this are technical and are given in an appendix. in §5, we look at some alternatives. One is the more restricted set theory of Kripke and Platek with “bounded quantifiers” which may be sufficient for core mathematics. The other is category theory which does not fit cleanly into ZFC. We describe the two approaches that have been suggested, to accommodate categorical approaches. Finally in §6, we look at a truly remarkable model of our axioms due to Solovay [ So ] in which \mathbb{R} is countable and and all its subsets are measurable, hence Freiling’s axioms hold. This construction requires higher set theory, the existence of a model with an “inaccessible” cardinal. However, it shows that if ZFC plus the existence of an inaccessible is consistent, then so are our axioms. Solovay’s approach also leads to an alternative “minimalist” foundation given in this section. Other models of Freiling’s axioms can also be obtained without an inaccessible using iterations of random real forcing (see the Appendix). Both authors have considered the issue of the optimal model for the foundations of mathematics [ M1 , M2 ] and for set theory [ SFr , AFHT ] , respectively, and this led to our collaboration. The present article attempts to include enough set-theoretic background to be readable by non-specialists. We will review most definitions but good references for non-specialists who seek further clarification are the books [ Je , De , Ba ] . §1. Do we need set theory? As mentioned in the Introduction, the natural numbers and the real numbers occupy special places among all sets. The simplest approach to a foundation for the core mathematician would be to formulate a theory with distinct natural number unknowns and real unknowns. This leads to so-called second order arithmetic . First, we need to clarify the treatment of \mathbb{R} . It’s simplest to describe a real number in the interval [0,1] by its binary expansion. There is an annoying need to identify the expansions 0.\,\vec{a}\,0\,1\,1\,1\,1\cdots=0.\,\vec{a}\,1\,0\,0\,0\,0\cdots where \vec{a} is any finite initial sequence, but these are only a countable set of identifications. Without this identification, the set of 1’s in a binary expansion gives us an arbitrary subset of \mathbb{N} , an element of the power set \mathcal{P}(\mathbb{N}) . The set of all sequences of 0’s and 1’s itself is homeomorphic to the Cantor set which we will call \mathcal{C} . As far as