Nonrigidity of smooth tight surfaces

arXiv:2611.17269 · math.DG, math.AP · Submitted 2026-11-12 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: We're looking at the paper: "Nonrigidity of smooth tight surfaces".

Emmy: Here's the gist of the paper. We construct smooth isometric embeddings of closed surfaces of positive genus in Euclidean 3-space which are tight, i.e., have minimal total positive curvature, but are not congruent The rigidity theorem of Alexandrov for analytic tight surfaces does not hold in the smooth category Core Finding Theorem 1.1 states that for every integer g ≥ 1, there is a Riemannian metric on the closed orientable surface M of genus g which admits uncountably many smooth (C∞) noncongruent tight isometric embeddings into R3 These embeddings coincide outside a compact set in the negatively curved region of M and enclose different volumes.

Ada: Thus Alexandrov’s rigidity theorem does not hold in the smooth category, and the answer to the Banchoff–K¨uhnel question is also negative.

Emmy: Construction of Noncongruent Embeddings The proof of Theorem 1.1 is based on constructing a negatively curved band ruled by the principal normals of a closed space curve Near this curve, the asymptotic curves transverse to the rulings are closed, which allows for the construction of infinitesimal bendings supported between any two of them.

Ada: The construction involves extending this band to a negatively curved annulus whose ends are surfaces of revolution and then doubling it to insert g copies between flat faces.

Paper summary: Ada: So to wrap up, this paper "Nonrigidity of smooth tight surfaces" constructs uncountably many noncongruent smooth isometric embeddings of closed surfaces with positive genus in three-dimensional space that are tight but not congruent.

Emmy: The implication is that the rigidity theorem of Alexandrov doesn't hold in the smooth category for these objects because they can enclose different volumes.

Kolya: It confirms that you can have a continuous family of noncongruent embeddings, and this is achieved through bending fields with nonzero flux supported in the negatively curved region.

Ada: The construction relies on building a specific ruled band, extending it to an annulus with rotationally symmetric ends, and then gluing copies to form the final surface of genus g.

Emmy: It really shows that even after adding smoothness, there's still this structural freedom in how these surfaces can be embedded. This is a result about the Banchoff–K¨uhnel question.

Kolya: So for us, it means that when we look at these tight surfaces, we have to consider the volume as something that can vary even under isometric constraints in three-dimensional space.

Conclusion: Ada: So we're wrapping up our look at "Nonrigidity of smooth tight surfaces." The core idea is that you can have these surfaces, they are smooth, they are tight—meaning their total positive curvature is exactly four pi—but you can’t keep them fixed in space.

Emmy: Exactly. The authors show that for a closed surface with a positive genus, there are uncountably many ways to embed it smoothly into three dimensional space so that the metric stays the same but the shape of the surface changes.

Kolya: It means these surfaces can enclose different volumes, which is wild if you're thinking about how they interact with gravity or anything else in physics.

Ada: That's right. The title points to this lack of rigidity in the smooth category, contrasting it with Alexandrov’s earlier work on analytic surfaces where things were much more fixed.

Emmy: It really shows that the smoothness constraint isn't enough to enforce rigidity when we deal with tight embeddings; you need to look at the geometry of bending fields instead.

Kolya: I wonder how this affects anything practical? If these volumes can be different, does that change any theorems about minimal surfaces or something?

Ada: That’s the big question. It means we have to be careful when we say that a tight surface is rigid; in the smooth world, it isn't.

Emmy: The authors achieved this by using carefully constructed bands and then perturbing them near flat spots to keep the tightness property intact while allowing for those volume differences.

Kolya: It's fascinating how they manage to maintain those properties with these infinitesimal bendings, which is where the noncongruence comes from in the first place.

Ada: So, it opens up a whole new way to think about what makes a surface rigid or flexible when we allow for smooth deformations. Next time we talk about whether these surfaces can actually move continuously—whether they are flexible themselves.

MOHAMMAD GHOMI, HONGDA QIU, MATTEO RAFFAELLI, YUETONG ZHANG

math.DG, math.AP

Submitted: 2026-11-12

Updated: 2026-11-12

Comments: 16 pages, 5 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 75/100

The gist: We construct smooth isometric embeddings of closed surfaces of positive genus in Euclidean 3-space which are tight, i.e., have minimal total positive curvature…

Key concepts

Tight Embedding
A tight embedding is a surface where the total positive curvature equals $4 imes ext{genus} imes ext{pi}$. This condition relates the intrinsic geometry of the surface to its extrinsic embedding in 3D space. The paper constructs surfaces that satisfy this condition while allowing for non-rigid deformations.
Infinitesimal Bending Field
This is a small, continuous deformation applied to an embedding that preserves the induced metric up to first order. If the field has nonzero flux (a net flow through a region), it causes the enclosed volume of the surface to change, leading to noncongruent embeddings even if they share the same metric.
Nonzero Flux
Flux measures how much of a bending field passes through a specific region. The paper constructs bending fields supported in negatively curved regions that have nonzero flux. This nonzero flux is the mechanism that ensures the resulting embeddings enclose different volumes, proving noncongruence.

Terminology

Summary

We construct smooth isometric embeddings of closed surfaces of positive genus in Euclidean 3-space which are tight, i. p. 0e., have minimal total positive curvature, but are not congruent The rigidity theorem of Alexandrov for analytic tight surfaces does not hold in the smooth category

Core Finding

Theorem 1.1 states that for every integer g ≥ 1, there is a Riemannian metric on the closed orientable surface M of genus g which admits uncountably many smooth (C∞) noncongruent tight isometric embeddings into R3 These embeddings coincide outside a compact set in the negatively curved region of M and enclose different volumes. p. 0 Thus Alexandrov’s rigidity theorem does not hold in the smooth category, and the answer to the Banchoff–K¨uhnel question is also negative. p. 0

Construction of Noncongruent Embeddings

The proof of Theorem 1.1 is based on constructing a negatively curved band ruled by the principal normals of a closed space curve Near this curve, the asymptotic curves transverse to the rulings are closed, which allows for the construction of infinitesimal bendings supported between any two of them. p. 1 The construction involves extending this band to a negatively curved annulus whose ends are surfaces of revolution and then doubling it to insert g copies between flat faces. p. 1

Mechanism for Noncongruence

The noncongruence arises from an infinitesimal bending field Y supported in the negatively curved region that has nonzero flux. The embeddings X ± εY induce the same metric on M while their enclosed volumes differ to first order by 2εR⟨Y, ν⟩, where ν is the outward unit normal of X. p. 1 Uncountably many such embeddings are obtained from a sequence of bending fields with disjoint supports. p. 1

Extension to Genus g Surfaces

The construction proceeds by first constructing the annulus A, which is the graph of −h∗ which is saddle, and then perturbing it near its flat regions so that 0 becomes a regular value of its curvature. By gluing g copies of this modified annulus Ae, a smooth closed surface Σ of genus g is formed which is tight because its total positive curvature agrees with that of C which is 4π. p. 0

Perturbation and Tightness Preservation

The perturbation near the flat regions involves modifying the height function by subtracting a term related to λy2 / 2, leading to a surface whose curvature is negative for r rλ, ensuring 0 is a regular value of the curvature. The total positive curvature of the new surface is calculated as 4π(1 − g) + 4πg = 4π, confirming tightness. p. 0

Volume Variation and Uncountability

The volumes enclosed by X± satisfy V (X+) − V (X−) = 2εF + o(ε), where F is the nonzero flux. By applying Proposition 2. p. 3 p. 33 to a sequence of disjoint subintervals of I, one obtains bending fields Yj with disjoint supports and nonzero flux. p. 1 This leads to embeddings Xσ which enclose different volumes, establishing uncountably many noncongruent tight surfaces. p. 0

Final Embedding Construction

The final embedding is constructed by bending the initial surface Σ along the fields Y of nonzero flux chosen in Section 3.2. The resulting embeddings X± are smooth and tight because the curvature remains negative where the support of Y is compact, and their volumes differ due to the nonzero flux F. p. 0 The construction yields uncountably many such embeddings by applying Proposition 2. p. 03 to a sequence of disjoint subintervals of I. p. 1

Acknowledgement

Initial drafts of the main proofs were developed with the assistance of Claude (Anthropic) and ChatGPT (OpenAI), building on our earlier work [5]. These drafts required extensive clarification, revision, and verification before reaching their present form. p. 5

References

[1] M. Aigner and G. M. Ziegler, Proofs from The Book, Springer-Verlag, Berlin, 1999

[2] A. D. Alexandrov, On a class of closed surfaces, Mat. p. 0 Sbornik 4 (1938), 69–77

[3] T. F. Banchoff, Non-rigidity theorems for tight polyhedra, Arch. p. 0 Math. (Basel) 21 (1970), 416–423

[4] T. F. Banchoff and W. p. 0 K¨uhnel, Tight submanifolds, smooth and polyhedral, Tight and taut submanifolds (Berkeley, CA, 1994), 1997, pp. p. 0 51–118

[5] H. Banks, M. Ghomi, H. Qiu, and M. p. 0 Raffaelli, A counterexample to Nirenberg’s conjecture on closed asymptotic curves, 2026

[6] A. I. Bobenko, T. Hoffmann, and A. O. Sageman-Furnas, Compact Bonnet pairs: isometric tori with the same curvatures, Publ. p. 0 Math. Inst. Hautes Etudes Sci. ´ 142 (2025), 241–293

[7] A. L. Cauchy, Sur les lignes asymptotiques et leur repr´esentation sph´erique, Bull. p. 1 Sci. Math. (2) 1888, 126–128

[8] T. E. Cecil and S.-S. Chern (eds.), Tight and taut submanifolds, Mathematical Sciences Research Institute Publications, vol. p. 0 32, Cambridge University Press, Cambridge, 1997

[9] S. Y. Cheng and S. p. 7 T. Yau, On the regularity of the Monge-Amp’ere equation det(∂ 2u/∂xi∂xj) = F(x, u), Comm. p. 7 Pure Appl. Math. 30 (1977), no. 1, 41–68

[10] S. Cohn-Vossen, Zwei Sätze über die Starrheit der Eiflächen, Nachr. p. 1 Ges. Wiss. Göttingen, Math. p. 0-Phys. Kl. (1927), 125–134

[11], Unstarre geschlossene Flächen, Math. Ann. 102 (1930), 10–29

[12] R. Connelly, A counterexample to the rigidity conjecture for polyhedra, Publ. p. 1 Math. Inst. Hautes Etudes Sci. ´ 47 (1977), 333–338

[13] M. Ghomi, Gauss map, topology, and convexity of hypersurfaces with nonvanishing curvature, Topology 41 (2002), no. p. 7 1, 107–117

Conclusion

The paper demonstrates that the rigidity theorem of Alexandrov does not hold in the smooth category for tight surfaces. The existence of uncountably many noncongruent smooth tight isometric embeddings into R3, which enclose different volumes, settles the Banchoff–K¨uhnel question negatively. p. 0 This result is achieved through a detailed construction involving ruled bands and infinitesimal bending fields with nonzero flux. p. 2 The paper also provides a method for extending the support function of the surface to R squared without losing its saddle property. p. 5 The final result confirms that smooth tight surfaces are not rigid under isometric embedding constraints. p. 0

--- Page 5 ---

NONRIGIDITY OF SMOOTH TIGHT SURFACES

M. GHOMI, H. QIU, M. RAFFAELLI, AND Y. p. 0 p. 0 ZHANG

however, whether these surfaces are flexible, i.e., admit a continuous one-parameter family of noncongruent isometric embeddings

--- Page 6 ---

M. GHOMI, H. QIU, M. RAFFAELLI, AND Y. ZHANG

The proof of Theorem 1.1 is based on the following observation, which goes back to Cohn-Vossen [11]; see also Rembs [26] Let X: M → R3 be an embedding of a closed surface with induced metric IX:= ⟨dX, dX⟩. p. 1 A vector field Y along X is an infinitesimal bending [27] if ⟨dX(v), dY (v)⟩ = 0 for all tangent vectors v. p. 1 Then (IX+εY = IX−εY = IX + ε2/⟨dY, dY⟩. p. 1 For small ε, the embeddings X ± εY therefore induce the same metric, while their enclosed volumes differ to first order by 2εR⟨Y, ν⟩, where ν is the outward unit normal of X. p. 1 Thus it suffices to construct a tight embedding X: M → R3 carrying an infinitesimal bending of nonzero flux R M⟨Y, ν⟩ supported in the negatively curved region, for then X ± εY remain tight for small ε. p. 1

--- Page 7 ---

Construction of the Band

We begin by constructing a negatively curved ruled band X which, near its central curve, is foliated by closed asymptotic curves, and show that it carries in its interior compactly supported infinitesimal bending fields Y with nonzero flux. The band is ruled by the principal normals of a closed space curve α, which is determined by its binormal indicatrix B and its speed v. p. 2

The Asymptotic Curves

By (3), the latter curves are given by s 7→ X(s, u(s)) with u(v′−κ′u)−2vu′ = 0. In particular α itself is asymptotic. p. 3 For u ≠ 0, the substitution 1/u makes this equation linear, and multiplying by √v then gives (4) √v/u′ = κ′2/√v. p. 3 The level sets ξ = c, c > 0, are closed asymptotic curves that foliate the region 0 < u < √v/ω < δ where ω is chosen so large that √v/ω < δ. p. 3

The Bending Field Y

Proposition 2.3 shows that for every nonempty open interval I ⊂ (0, ∞) there is a function ϕ ∈ C∞ c(I) whose bending field (5) has nonzero flux. p. 4 The support of Y lies in the region c1 < ξ < c2 in the interior of the domain of X where √v/ω < δ, and so Y extends smoothly by zero to a smooth infinitesimal bending of X with nonzero flux F:= R M⟨Y, ν⟩ ≠ 0. p. 4

Flux Calculation

Lemma 2.2 shows that the flux of the bending field Y over the band is given by (6) Z⟨Y, ν⟩ = 2 Z I ϕ(c) Z L0 κ v3/2 (c + ω) 4 − (1 + κ2) v(c + ω) 5 ds dc. p. 2 Proposition 2.3 proves that for every nonempty open interval I there is a function ϕ ∈ C∞ c(I) whose bending field (5) has nonzero flux. p. 4

Closing Conditions

The construction of the spherical curve B and speed v involves periodic conditions on L/q-periodic functions, which are satisfied by a carefully chosen speed function v0 > 0 satisfying the first condition in (2). The second condition is satisfied by perturbing v0 using smooth cutoffs to ensure Π(vσ) = 0 for some σ. p. 6

Embeddedness

The embeddedness of X is shown when the frequency q is sufficiently large and the thickness δ is sufficiently small, as this ensures that α winds once around the origin and dα/dθ is a positive multiple of −J dβ/dθ. The uniform estimate for α gives θ2 − θ1 = O(1/q), which implies that α, and consequently α, is embedded. p. 7

Extension of the Band

The band X is extended to an annulus with rotationally symmetric ends by working with the support function h of X. This involves showing that h extends to a saddle function on R squared without losing its saddle property. p. 5 The extended support function is the graph of −h∗ over the annulus a < y < R0, which forms an embedded negatively curved annulus A. p. 8

The surface Σ is constructed by removing disks from a convex surface C and gluing in g copies of Ae, resulting in a tight surface Σ of genus g. The perturbation near the flat regions ensures that 0 becomes a regular value of its curvature, and the total positive curvature remains at 4π, preserving tightness. p. 1

Final Embeddings

The final embeddings Xσ are smooth embeddings that are tight and enclose different volumes because the contributions of Yj to the enclosed volume add up. The convergence in every C k norm ensures that Xσ − XC 2 ≤ ε, and the surfaces enclose different volumes by V (Xσ) − V (Xσ′) = P j(σj − σ'j)δj > 0.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved system can construct smooth isometric embeddings of closed surfaces of positive genus in Euclidean 3-space that are tight but not congruent, which allows it to settle problems related to Nirenberg and Banchoff–K¨uhnel regarding the rigidity theorem of Alexandrov for analytic tight surfaces.

  2. Improved system can determine the existence and properties of uncountably many smooth noncongruent tight isometric embeddings into R3 by constructing negatively curved bands with nonzero flux, as detailed in Theorem 1.1 and Section 4.3.

  3. Improved system can identify the necessary conditions for a surface to be nonrigid by analyzing how infinitesimal bending fields induce different volumes, specifically utilizing the relation the enclosed volumes differ to first order by 2epsilon R , where nu is the outward unit normal of X.

  4. Improved system can generate specific geometric constructions used in the proof, such as constructing a negatively curved ruled band ruled by principal normals of a closed space curve and extending it into an annulus with rotationally symmetric ends.

  5. Improved system can perform complex analytic calculations related to the support function h and its Legendre transform h star, enabling the extension of saddle functions across boundaries using Lemma 3.3 and Lemma 3.4 to construct surfaces that agree with X on the bending region but have different asymptotic behavior at infinity.

Sources

Related papers