# Rings of integers and existentially defined subfields The paper [Existential Definitions of Rings of Integers and Subfields](../../papers/number_fields/number_field_definability_20261109_v12.tex) gives complete proofs of the following statements. Counts include distinguished inputs. - For every number field K, its ring of integers is Diophantine in K if and only if Z is Diophantine in K. The implication from the ring of integers uses the established Denef–Lipshitz theorem, cited to Alpöge–Bhargava–Ho–Shnidman, Corollary 1.2. - Assuming Z is Diophantine in Q, these statements are equivalent to Q being Diophantine as a subfield of K. - For every finite characteristic-zero extension L/K, defining K in L is equivalent to defining the relative norm image, norm graph, or trace graph as subsets of powers of L. The norm-image implication uses a finite-difference identity. For quadratic extensions, defining the conjugation graph is equivalent as well. - A definition of Z inside a degree-n number field with N total coordinates gives a definition of its ring of integers with nN total coordinates. - An N-coordinate definition of O_L in L descends along a degree-d extension L/K to a definition of O_K in K with 1+d(N−1) total coordinates. A genuine counterexample therefore remains a counterexample in every finite extension. - Rank stability gives an infinite Diophantine subset of the rational subfield, while the cited ring-transfer theorem quantifies over rings of integers and the field-transfer theorem retains one universal quantifier. Two separate notes prove the related obstructions. The [abscissa-product obstruction](../../negative_results/papers/number_fields/abscissa_product_obstruction_20261116_v42.tex) uses the positive-rank curve y²=x³−2: every rational abscissa has even 5-adic valuation, so no multiplicative combination can equal 5. The [unchanged-formula obstruction](../../negative_results/papers/number_fields/unchanged_formula_obstruction_20261116_v42.tex) proves that a fixed finite family of existential formulas cannot define the rings of integers across all finite extensions. This concerns reuse of formulas, not separately chosen formulas for individual fields. The paper does not claim that every number field has a Diophantine ring of integers in its field of fractions, and it does not supply a counterexample. ## Exact verification Run: ```sh python check_norm_image_reduction.py ``` The script uses only Python's standard library. It independently checks every coefficient of ```text sum_{j=0}^{n-1} (-1)^(n-1-j) binom(n-1,j) (T+j)^n = n! T + n!(n-1)/2 ``` for n=1,...,64, plus exact rational and quadratic examples. It also enumerates the reductions of y²=x³−2 at 5 and 7 in two independent ways, confirming group orders 6 and 7, and checks the exact double of (3,5). It writes `norm-image-reduction-verification.json`. The proof for every n is in the paper; the finite checks are supplementary and do not establish any new defining polynomial for a ring of integers. ## Primary bibliography - L. Alpöge, M. Bhargava, W. Ho, A. Shnidman, *Rank stability in quadratic extensions and Hilbert's tenth problem for the ring of integers of a number field*, Inventiones Mathematicae 243 (2026), 1129–1139, Corollary 1.2. - B. Poonen, *Rank stability makes rings of integers diophantine*, 3 July 2026, sections 1.2, 4, 5. - Stacks Project, Chevalley's theorem, Tag 00FE. - G. Cornelissen, A. Shlapentokh, *Defining the integers in large rings of a number field using one universal quantifier*, J. Math. Sci. 158 (2009), 713–726, Theorem 4 and Remark 6.3. ## Correspondences and bounded denominators The paper finite_correspondences_and_denominators_20261109_v14.tex extends the abelian-image obstruction to morphisms whose geometric fibers have finitely many output values, including quasi-finite covers. Semiabelian Mordell–Lang shows that the rational outputs meet any finitely generated multiplicative subgroup in a finite set. For bounded-denominator outputs, further obstructions cover soluble unirational fibers and smooth proper genus-one families with an ample line bundle. The latter uses Siegel and uniform torsion, with a finite map to the relative Jacobian and no assumed section. Higher-genus families and arbitrary exceptional fibers are outside the result. Run python3 computation/number_fields/check_exceptional_correspondence.py from the repository root. The exact rational checker verifies a smooth surface birational to a product of elliptic curves whose four exceptional affine lines represent every rational input. This example explains why generic finiteness alone is insufficient. The general arithmetic theorems are proved in the paper, not by the finite check. Additional primary references are P. Vojta's survey on integral points and semiabelian Mordell–Lang, L. Merel's uniform torsion theorem, and the Jacobian construction of S. Y. An and coauthors with its published corrigendum. Complete references and exact theorem uses are in the paper.