A Nash-Kuiper theorem for isometric immersions beyond Borisov’s exponent

Wentao Cao, Jonas Hirsch, and Dominik Inauen · 2025

Read A Nash-Kuiper theorem for isometric immersions beyond Borisov’s exponent by Wentao Cao, Jonas Hirsch, and Dominik Inauen (2025) 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-2503-13867.txt.

Introduction

Wentao Cao, Jonas Hirsch, and Dominik Inauen wrote this 2025 mathematics paper stretching a classic theorem on bending surfaces without stretching them. The Nash-Kuiper theorem says a squeezed-short shape can still be embedded with its distances intact, wrinkled at fine scales. How rough that wrinkled map may be has a famous threshold tied to Borisov.

The team push past that threshold: for any short immersion of a bounded, simply connected region, they build an isometric immersion as close as desired, with smoothness up to the weaker exponent. In two dimensions their bound meets the Onsager exponent known from fluid turbulence. Three new devices power the proof: a fresh corrugation pattern, an integration-by-parts step, and a tailored split of the error terms.

Catalog record

CREATORWentao Cao, Jonas Hirsch, and Dominik Inauen
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE78 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-13867

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Nash-Kuiper theorem for isometric immersions beyond Borisov’s exponent Wentao Cao, Jonas Hirsch, and Dominik Inauen Abstract . Given any short immersion from an n n -dimensional bounded and simply connected domain into ℝ n + 1 \mathbb{R}^{n+1} and any Hölder exponent α < ( 1 + n 2 − n ) − 1 \alpha<(1+n^{2}-n)^{-1} , we construct a C 1 , α C^{1,\alpha} isometric immersion arbitrarily close in the C 0 C^{0} topology. This extends the classical Nash–Kuiper theorem and shows the flexibility of C 1 , α C^{1,\alpha} isometric immersions beyond Borisov’s exponent. In particular, for n = 2 n=2 , the regularity threshold aligns with the Onsager exponent 1 / 3 1/3 for the incompressible Euler equations. Our proof relies on three novelties that allow for the cancellation of leading-order error terms in the convex integration scheme: a new corrugation ansatz, an integration by parts procedure, and an adapted algebraic decomposition of these errors. 1991 Mathematics Subject Classification 58D10, 53C21, 57N35 1. Introduction The isometric immersion problem is a fundamental problem in differential geometry. It seeks an immersion u:(\mathcal{M},g)\to\mathbb{R}^{m} from an n -dimensional Riemannian manifold (\mathcal{M},g) into m -dimensional Euclidean space that preserves the length of any C^{1} curve. This condition is equivalent to the equality of the induced metric u^{\sharp}e and the intrinsic metric g . In local coordinates, this translates to the system of n_{*}=n(n+1)/2 nonlinear partial differential equations: g_{ij}=\partial_{i}u\cdot\partial_{j}u,\quad 1\leq i,j\leq n, (1.1) in m unknowns. The classical Nash–Kuiper theorem [ 33 , 28 ] establishes that for m\geq n+1 , any short immersion (or embedding) u:\mathcal{M}\to\mathbb{R}^{m} can be uniformly approximated by C^{1} isometric immersions (resp. embeddings). Here, an immersion u is called short if \partial_{i}u\cdot\partial_{j}u\leq g_{ij} as quadratic forms. In particular, if the manifold is compact, any immersion can be made short by a homothety. The Nash–Kuiper theorem therefore demonstrates that when there are no topological obstructions to immersing the manifold into \mathbb{R}^{m} (a condition satisfied for m\geq 2n-1 ), there exists an abundance of C^{1} solutions to ( 1.1 ). This abundance, often termed the flexibility of isometric immersions, is especially striking given the overdetermined nature of ( 1.1 ) for m\geq n+1 and large n . In contrast, classical rigidity results show that smooth isometric embedding into Euclidean spaces with such low codimension are typically unique. A prominent example is the rigidity theorem for the Weyl problem due to Cohn-Vossen [ 12 ] and Herglotz [ 24 ] , which states that a C^{2} isometric embedding u:(\mathbb{S}^{2},g)\to\mathbb{R}^{3} , where g has positive Gaussian curvature, is unique up to rigid motions. This strong contrast between flexibility in C^{1} and rigidity in C^{2} raises the natural question: is there a critical Hölder regularity threshold \alpha_{0} that separates flexibility from rigidity? That is, does there exist a threshold \alpha_{0} such that • if \alpha>\alpha_{0} , isometric immersions u\in C^{1,\alpha} exhibit (some form of) rigidity; • if \alpha<\alpha_{0} , the Nash–Kuiper theorem extends to C^{1,\alpha} ? The precise value of \alpha_{0} remains unresolved. Similarities between the iteration processes used to construct both isometric immersions or embeddings of regularity C^{1,\alpha} and Hölder continuous weak solutions to the incompressible Euler equations (see e.g. [ 17 ] ) suggest the Onsager exponent \alpha_{0}=1/3 as a potential threshold. Indeed, the Onsager conjecture, which describes a similar phenomenon, states that for C^{\alpha} weak solutions to the incompressible Euler equations (1) if \alpha>1/3 , it preserves the energy; (2) if \alpha<1/3 , the energy identity might be violated. The rigidity result (1) was established in [ 13 ] , while the flexibility result (2) was ultimately resolved in [ 26 ] following a sequence of works [ 19 , 20 , 4 , 5 ] that built on the groundbreaking approach of [ 18 ] , where the authors introduced a Nash-type iteration scheme to construct continuous weak solutions violating the energy identity. On the other hand, [ 15 , 8 ] demonstrate that for isometric embeddings, C^{1,\nicefrac{{1}}{{2}}} is a critical space in a suitable sense , suggesting \alpha_{0}=1/2 (see also [ 22 , Question 36] , [ 17 , Section 10] ). This paper focuses on the flexibility side of this dichotomy for general dimension n and codimension one. The case of general codimension m\geq n+1 is investigated in forthcoming work. 1.1. Flexibility of C^{1,\alpha} isometries: known results The study of C^{1,\alpha} isometric immersions dates back to the pioneering work of Yu. F. Borisov in the 1950s. Building on results of Pogorelov, Borisov proved in [ 1 ] that the Cohn–Vossen–Herglotz rigidity theorem extends to immersions of class C^{1,\alpha} when \alpha>2/3 (see [ 14 ] for an alternative proof). On flexibility, Borisov announced in [ 2 ] that the Nash–Kuiper theorem extends to C^{1,\alpha} for \alpha<\frac{1}{1+n^{2}+n}\,, (1.2) the Borisov exponent , when \mathcal{M} is an n -dimensional ball, with a potential improvement to \alpha<1/5 for n=2 . He provided a proof for n=2 and \alpha<1/7 with an analytic metric in [ 3 ] . More recently, [ 14 ] confirmed Borisov’s claims, and in [ 16 ] it was shown that for n=2 one can indeed achieve the exponent \alpha<1/5 . These results were extended to general compact manifolds in [ 10 ] . Remark 1.1 (High codimension) . If the codimension is large, more regular isometric immersions can be constructed. The breakthrough result in this direction is also due to Nash [ 34 ] , who proved that any (M,g) with g\in C^{k} , k\geq 3 , admits a C^{k} -regular isometric immersion into \mathbb{R}^{m} for sufficiently large m . Gromov [ 21 ] and Günther [ 23 ] improved the codimension bounds, the latter simplifying Nash’s intricate iteration (now known as the hard implicit function theorem ). For less regular metrics g\in C^{l,\beta} with 0<l+\beta<2 , Källén [ 27 ] demonstrated that if m is large enough (one can show that m\geq 2n_{*}+n suffices), there exists a C^{1,\alpha} immersion for any \alpha<(l+\beta)/2 . See also [ 15 , 8 , 11 ] for further results in high codimension. 1.2. Statement of the result This paper improves the achievable Hölder exponent in the codimension one setting. Our main result is: Theorem 1.1 . Let \Omega\subset\mathbb{R}^{n} be any smooth bounded and simply connected domain and g\in C^{2}(\overline{\Omega},\mathrm{Sym}_{n}^{+}) a Riemannian metric. For any short immersion \underline{u}\in C^{1}(\overline{\Omega},\mathbb{R}^{n+1}) , any \epsilon>0 , and any \alpha<\frac{1}{1+n^{2}-n}, (1.3) there exists an immersion u\in C^{1,\alpha}(\overline{\Omega},\mathbb{R}^{n+1}) such that Du^{t}Du=g\quad\text{and}\quad\|u-\underline{u}\|_{0}<\epsilon. Remark 1.2 . For n=2 , Theorem 1.1 yields flexibility of C^{1,\nicefrac{{1}}{{3}}-} isometric immersions from surfaces to \mathbb{R}^{3} , aligning with the Hölder exponent of Onsager’s conjecture. Given the parallels between convex integration solutions for Onsager’s conjecture and isometric immersions (see e.g. [ 6 , 17 ] ) and the heuristic discussion about the Hölder exponent \alpha<(1+2N)^{-1}(N\geq 1) in Section 1.3.2 , 1/3 may be optimal for codimension-one flexibility. Remark 1.3 . Following classical arguments (see e.g. [ 16 , Section 8] ), one can show that if the short map \underline{u} in Theorem 1.1 is an embedding, then u can be chosen to be an embedding as well. Remark 1.4 . With additional effort, the key iteration Proposition 3.4 can be adapted to the framework of [ 10 ] , leading to a global version of

Read the complete text