Dihedral Rigidity for Convex Polytopes by Smooth Approximation

Yuchen Bi

Dihedral Rigidity for Convex Polytopes by Smooth Approximation

T0 review · 0 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves the dihedral rigidity conjecture for convex polytopes: any smooth metric with nonnegative scalar curvature, nonnegative face mean curvature, and Euclidean angle bounds must be flat.

desk verdict This is the real thing: a full proof of Gromov's dihedral rigidity conjecture for all convex polytopes, with the technical estimates closing once you check them. read the letter →

arxiv 2608.06320 v1 pith:RJYEF5EC submitted 2026-08-06 math.DG

Yuchen Bi This is my paper · ORCID

classification math.DG MSC 53C2453C2153C27

keywords dihedralrigidityconvexpolytopescalarcurvaturemeanDiracoperatorspinorsmoothapproximationtheorem

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves the dihedral rigidity conjecture for convex polytopes: in every dimension $n\ge 3$, a smooth metric on a compact convex polytope that has nonnegative scalar curvature, nonnegative mean curvature on each face, and interior dihedral angles no larger than the corresponding Euclidean angles must be flat, with every face totally geodesic and all exterior dihedral angles equal to the Euclidean ones. The proof works by exhausting the polytope from the inside by smooth convex domains $P_\lambda$, constructing a degree-one sphere-valued map on each boundary whose derivative is controlled by the boundary mean curvature, and then solving a Dirac boundary value problem on each $P_\lambda$. A weighted boundary estimate shows that the error term in the Dirac energy inequality vanishes as $\lambda\to\infty$, so the limit is a nonzero parallel spinor. The existence of such a spinor forces the metric to be flat and the faces to be totally geodesic. This settles a conjecture that had previously been proved only for special classes of polytopes.

What carries the argument

The load-bearing mechanism is a Dirac boundary value problem for spinor-valued homomorphisms on a smooth inner approximation. For a domain $Y$ with boundary $\Sigma$ and a map $\eta:\Sigma\to S^{n-1}$, the boundary condition $\chi_\eta A=A$ with $\chi_\eta A=-c(\nu)A\,\omega(\eta)$ makes the Dirac operator a self-adjoint boundary problem, and Proposition 3.1 gives the energy identity $$ \int_Y|\nabla A|^2+\frac14\int_Y R_g|A|^2\le \frac12\int_\Sigma(\|d\eta\|_{\mathrm{tr}}-H^g_\Sigma)_+|A|^2. $$ For the approximate polytope $P_\lambda$ with $\eta=\eta_\lambda$, the right-hand side is controlled by Lemma 3.4, a weighted trace estimate that localizes the integrand $W_\lambda=(\|d\eta_\lambda\|_{\mathrm{tr}}-H^g_{\Sigma_\lambda})_+$ near the codimension-two and codimension-three strata and shows $\int_{\Sigma_\lambda}W_\lambda|\psi|^2\le \varepsilon_\lambda(\int_{P_\lambda}|\nabla\psi|^2+\int_B|\psi|^2)$ with $\varepsilon_\lambda\to 0$. The boundary map $\eta_\lambda$ is built from the radial map $q(x)=(x-p_0)/|x-p_0|$, which guarantees degree one, smoothly interpolated with the Euclidean normal field via a cutoff function depending on the distance to the codimension-three skeleton.

What would settle it

For a standard rectangular box, construct the approximate domains $P_\lambda$ and boundary maps $\eta_\lambda$ as in the paper and evaluate the integral $\int_{\Sigma_\lambda}W_\lambda|\psi|^2$ for a sequence of $W^{1,2}$ spinors $\psi_\lambda$ with uniformly bounded energy; Lemma 3.4 predicts decay at least as $C(\lambda^{-1/(n-1)}+\lambda^{-1/4}+|\log\lambda|^{-1/2})$, so observing any sequence for which the integral fails to tend to zero would disprove the central estimate and with it the proof.

Watch

Extended reading notes

Core claim

The central discovery is Theorem 1.1: let $P=\bigcap_{a=1}^m\{u_a\le 0\}\subset\mathbb{R}^n$ be a compact convex polytope with irredundant defining inequalities, $g$ a smooth metric on a neighborhood of $P$, $F_a=P\cap\{u_a=0\}$, $\nu_a=\nabla^g u_a/|du_a|_g$, and $\alpha_{ab}\le\alpha^g_{ab}$ the exterior dihedral angles along codimension-two faces. If $R_g\ge 0$ on $P$, $H^g_{F_a}\ge 0$ on each face, and $\alpha^g_{ab}\ge\alpha_{ab}$ along every codimension-two face, then $g$ is flat, every $F_a$ is totally geodesic, and $\langle\nu_a,\nu_b\rangle_g=\langle N_a,N_b\rangle$ on $F_a\cap F_b$. The proof establishes this by smooth inner approximation: it constructs domains $P_\lambda=\{F_\lambda\le 1\}$ with $F_\lambda=\sum_a\Phi(\lambda u_a)$, boundary maps $\eta_\lambda:\partial P_\lambda\to S^{n-1}$ of degree one that agree with the Euclidean normal on each face away from edges, and spinor fields $A_\lambda$ solving the Dirac equation $D A_\lambda=0$ in $P_\lambda$ with boundary condition $\chi_{\eta_\lambda}A_\lambda=A_\lambda$. The key step is a weighted boundary-error estimate showing that $\int_{\Sigma_\lambda}(\|d\eta_\lambda\|_{\mathrm{tr}}-H^g_{\Sigma_\lambda})_+|A_\lambda|^2\to 0$, because the excess is confined to $O(\lambda^{-1})$-neighborhoods of the codimension-two edges and $O(\lambda^{-1/4})$-neighborhoods of the codimension-three skeleton, where the trace of a $W^{1,2}$ spinor has negligible mass. Passing to the limit yields a nonzero parallel spinor, which forces the curvature tensor to vanish and the faces to be totally geodesic.

Load-bearing premise

The argument stands or falls on the claim that the contribution of the thin regions near edges and corners to the boundary integral vanishes in the limit; if that failed, the limit spinor would not be parallel and rigidity would not follow.

Editorial extensions

If this is right

This proves the dihedral rigidity conjecture for convex polytopes in all dimensions $n\ge 3$.
The metric must be flat, the faces totally geodesic, and the exterior dihedral angles exactly equal to their Euclidean values, so the inequalities in the hypotheses are equalities in the rigid case.
The scalar curvature $R_g$ must vanish identically; a metric satisfying the three hypotheses cannot carry any region of positive scalar curvature.
The result applies uniformly to odd and even dimensions through the product reduction $P\times[-1,1]$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The quantitative decay in Lemma 3.4 may yield a stability version: the rate $\varepsilon_\lambda=O(\lambda^{-1/4}+|\log\lambda|^{-1/2})$ could translate into an explicit closeness-to-Euclidean estimate for metrics that nearly satisfy the three inequalities.
The same smooth-approximation and Dirac-spinor scheme might adapt to other polyhedral rigidity problems, such as flat corner domination for manifolds with polyhedral boundary, provided an analogous weighted trace estimate for the strata holds.
A natural stress test is to check whether the $\lambda^{-1/4}$ width near the codimension-three skeleton is optimal; a sharper rate would extend the method to polytopes with more general stratifications.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this paper proves the full dihedral rigidity conjecture for convex polytopes in all dimensions n ≥ 3, removing the acute-angle restriction that limited Brendle–Wang. The new ingredient is the hybrid map ηλ, which patches the face-normal interpolation ζλ to a degree-one radial map q near the codimension-three skeleton, and the weighted boundary-error estimate Lemma 3.4 that makes the Dirac energy inequality close in the limit. I read the proof carefully and the main estimates hold up.

What is actually new: the theorem is genuinely absent from the literature. Brendle handled equality of angles, Brendle–Wang assumed acute angles, Li treated prisms and cones, and Wang–Xie–Yu gave dimension three. The hybrid map is a clean solution to the angle problem, and the paper is well-written, with detailed proofs of the geometric estimates. The reliance on Brendle's Dirac boundary existence result (Proposition 3.5, from his Invent. Math. paper) is external and properly cited, not circular. The even-dimensional reduction via product with an interval works: the new faces have right dihedral angles and zero mean curvature, so the odd-dimensional case applies.

The softest spot is indeed Lemma 3.4, as your reader noted. The three ingredients—Hölder on the edge neighborhood, the slice trace near K_P, and the log-cutoff weight—all close, with final rate λ^{-1/(n-1)} + λ^{-1/4} + |log λ|^{-1/2} → 0. I also checked the coordinate computation near F_I: the annulus integral is uniform, and the trace inequality (3.8) is consistent. The other imported premise, Proposition 3.5, comes from a published paper and is the kind of result one can rely on. I did not find a circular step, a missing proof, or a post-hoc assumption. The only real caveat is that the argument is long and delicate, with no formal verification, so a referee should check the estimates line by line, especially the cancellation in the boundary identity behind Proposition 3.1 and the slice trace argument.

Who this is for: geometric analysts working on scalar curvature rigidity, Dirac operator methods, and Gromov's polyhedral comparison questions. It deserves a serious referee. I recommend sending it to a strong journal with a referee who knows Brendle's framework. I'd be surprised if it didn't pass.

Referee Report

0 major / 4 minor

Summary. The paper proves Gromov's dihedral rigidity conjecture for compact convex polytopes in all dimensions n≥3. For a smooth metric g on a neighborhood of a polytope P with nonnegative scalar curvature, nonnegative mean curvature on each facet, and exterior dihedral angles no smaller than the Euclidean ones, the conclusion is that g is flat, each facet is totally geodesic, and the g-unit normals along intersecting facets have the same inner product as the Euclidean normals. The strategy follows Brendle's smooth inner approximation: a family P_λ of smooth convex domains approximates P from inside, and sphere-valued maps η_λ:∂P_λ→S^{n-1} are constructed so that the trace norm of dη_λ is bounded by the mean curvature of ∂P_λ plus an error term supported near the codimension-two edges and the codimension-three skeleton. A weighted boundary estimate (Lemma 3.4) shows the error is negligible when tested against the normalized solutions of the Dirac boundary problem. The limiting section is parallel, and Clifford algebra/Schur's lemma arguments give the rigidity conclusions. Even dimensions are reduced to odd ones by taking a product with an interval.

Significance. This is a full proof of a conjecture of Gromov, extending previous work by Brendle (matching angles) and Brendle–Wang (acute angles) to arbitrary dihedral angles. The argument is essentially self-contained after quoting Brendle's Dirac boundary existence result, and it is parameter-free: assumptions are used only in the form of inequalities, with no fitted constants entering the final rigidity statement. The main technical achievement is the error analysis near the skeleton, in particular the weighted trace estimate (3.5), which is proved by explicit measure and slicing arguments. The paper is clearly written and the estimates are checkable. If the proof is correct, it is a significant contribution to scalar-curvature rigidity.

minor comments (4)

[§3.3, Proposition 3.6] The sentence 'the strong L² convergence and (3.10) give ∫_P |A|² = 1' is not automatic, because the extensions E_λ A_λ may place mass in the shell P\P_λ; this should be justified, for instance by Hölder's inequality on P\P_λ together with the uniform W^{1,2} bound on E_λ A_λ, which yields a shell mass of order λ^{-2/n}.
[§3.2, Lemma 3.2] In the case I_λ(x)={a}, the assertion H^g_{Σλ}=H^g_{F_a}≥0 should be justified by noting that u_b(x)≤-2/λ for all b≠a places x in the portion of Σ_λ that coincides with F_a; as written this is terse.
[§3.3, even-dimension reduction] For the even-dimensional reduction, it would be helpful to state explicitly that P×[-1,1] with g+dt² satisfies the hypotheses; this follows from R_{g+dt²}=R_g and the vanishing mean curvature of the added faces, but it is not written.
[§1, header] The header on the first page contains obvious artifacts ('POL YTOPES', 'approxima tion' from line-breaking) that should be cleaned.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the proof is a self-contained approximation argument with an external, non-self Dirac existence input.

full rationale

The derivation chain is self-contained relative to its cited inputs. Theorem 1.1 is not assumed in the proof: the hypotheses R_g >= 0, H^g_{F_a} >= 0, and alpha^g_{ab} >= alpha_ab enter as conditions in the Dirac boundary estimate and in the construction of the approximating maps, while the conclusions of flatness, total geodesicity, and equality of face-pair inner products are derived from the limiting parallel spinor obtained in Proposition 3.6. Lemma 3.4 is a genuine weighted trace estimate with explicit vanishing constants, not a reformulation of the rigidity conclusion. Proposition 3.5 is imported from Brendle [1, Prop. 2.15] as an external Dirac existence theorem; it is not a self-citation, is not used as a uniqueness or rigidity statement, and does not presuppose the polytope rigidity conjecture. The smooth approximation scheme follows Brendle and Brendle-Wang but is re-derived in the present paper through Lemmas 2.1-2.9, including the explicit normal map estimate in Lemma 2.5. No fitted parameter is renamed as a prediction, no uniqueness theorem by the same authors is invoked, and no known empirical pattern is merely repackaged. The conclusion therefore does not reduce to its inputs by construction.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central argument introduces no fitted parameters and no new physical or geometric entities. Constants C, c, C0 and δλ=λ^{-1/4} are asymptotic choices independent of the target metric, not tuned to the conclusion. The load-bearing prior input is Brendle's Dirac boundary framework and the imported Proposition 3.5; the remaining background is standard spin geometry and convex analysis.

assumptions (4)

standard math Proposition 3.5: for every smooth compact convex domain Ω⊂R^n and every smooth degree-one η:∂Ω→S^{n-1}, the Dirac boundary problem DA=0, χ_η A=A has a nonzero smooth solution.
Imported from Brendle [1, Prop. 2.15]; used in Proposition 3.6 to obtain the approximating solutions Aλ for each smooth domain Pλ.
standard math Proposition 3.1: the Schrödinger-Lichnerowicz identity and boundary integration give the energy inequality (3.1) for any solution of DA=0 with χ_η A=A.
Imported from Brendle [1, Section 2]; this is the estimate that converts scalar curvature and boundary energy into a bound on ∇A.
standard math The spin representation of so(n) is faithful for n≥3, and an irreducible Hermitian Clifford module Δ_n exists with the stated normalization.
Used in the proof of Theorem 1.1 to pass from a unitary parallel spinor to Rm_g=0.
standard math For a compact convex polytope with an irredundant representation, the normal cone formula and uniform angle bounds in Lemma 2.1 hold.
Used to control d_g(x,F_I) and the level sets of Fλ near faces throughout Section 2.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Dihedral Rigidity for Convex Polytopes by Smooth Approximation." pith.science (2026). https://pith.science/paper/RJYEF5EC

@misc{pith2026260806320,
  author       = {Pith},
  title        = {Pith review of: Dihedral Rigidity for Convex Polytopes by Smooth Approximation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RJYEF5EC}},
  note         = {Machine review of arXiv:2608.06320}
}

read the original abstract

Following Brendle's smooth approximation approach, we give a proof of Gromov's dihedral rigidity conjecture for convex polytopes.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 8 canonical work pages

[1]

Brendle,Scalar curvature rigidity of convex polytopes, Invent

S. Brendle,Scalar curvature rigidity of convex polytopes, Invent. Math.235(2024), 669–708

work page 2024
[2]

C. Bär, S. Brendle, T.-K. A. Chow, and B. Hanke,Rigidity results for initial data sets satisfying the dominant energy condition, J. Reine Angew. Math. (2025), doi:10.1515/crelle-2025-0080

work page doi:10.1515/crelle-2025-0080 2025
[3]

Brendle and Y

S. Brendle and Y. Wang,On Gromov’s rigidity theorem for polytopes with acute angles, J. Reine Angew. Math.827(2025), 191–225

work page 2025
[4]

Gromov,Dirac and Plateau billiards in domains with corners, Cent

M. Gromov,Dirac and Plateau billiards in domains with corners, Cent. Eur. J. Math. 12(2014), 1109–1156

work page 2014
[5]

Li,A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent

C. Li,A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent. Math.219(2020), 1–37

work page 2020
[6]

Li,Correction to: A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent

C. Li,Correction to: A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent. Math.228(2022), 535–538

work page 2022
[7]

Li,The dihedral rigidity conjecture forn-prisms, J

C. Li,The dihedral rigidity conjecture forn-prisms, J. Differential Geom.126(2024), 329–361

work page 2024
[8]

Wang and Z

J. Wang and Z. Xie,On Gromov’s flat corner domination conjecture and Stoker’s conjecture, arXiv:2203.09511

arXiv

Show all 10 references

[9]

J. Wang, Z. Xie, and G. Yu,On Gromov’s dihedral extremality and rigidity conjectures, arXiv:2112.01510

arXiv
[10]

J. Wang, Z. Xie, and G. Yu,Gromov’s dihedral rigidity conjecture in dimension three, arXiv:2606.30130. (Yuchen Bi)Mathematical Institute, Department of Pure Mathematics, Uni- versity of Freiburg, Ernst-Zermelo-Straße 1, D-79104 Freiburg im Breisgau, Germany Email address:yuche...

arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Brendle,Scalar curvature rigidity of convex polytopes, Invent

S. Brendle,Scalar curvature rigidity of convex polytopes, Invent. Math.235(2024), 669–708

work page 2024

[2] [2]

C. Bär, S. Brendle, T.-K. A. Chow, and B. Hanke,Rigidity results for initial data sets satisfying the dominant energy condition, J. Reine Angew. Math. (2025), doi:10.1515/crelle-2025-0080

work page doi:10.1515/crelle-2025-0080 2025

[3] [3]

Brendle and Y

S. Brendle and Y. Wang,On Gromov’s rigidity theorem for polytopes with acute angles, J. Reine Angew. Math.827(2025), 191–225

work page 2025

[4] [4]

Gromov,Dirac and Plateau billiards in domains with corners, Cent

M. Gromov,Dirac and Plateau billiards in domains with corners, Cent. Eur. J. Math. 12(2014), 1109–1156

work page 2014

[5] [5]

Li,A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent

C. Li,A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent. Math.219(2020), 1–37

work page 2020

[6] [6]

Li,Correction to: A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent

C. Li,Correction to: A polyhedron comparison theorem for 3-manifolds with positive scalar curvature, Invent. Math.228(2022), 535–538

work page 2022

[7] [7]

Li,The dihedral rigidity conjecture forn-prisms, J

C. Li,The dihedral rigidity conjecture forn-prisms, J. Differential Geom.126(2024), 329–361

work page 2024

[8] [8]

Wang and Z

J. Wang and Z. Xie,On Gromov’s flat corner domination conjecture and Stoker’s conjecture, arXiv:2203.09511

arXiv