Upper bound for the total mean curvature of spin fill-ins
Christian Bär · 2026
Read Upper bound for the total mean curvature of spin fill-ins by Christian Bär (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-2601-06713.txt.
Introduction
Christian Bar proves an upper bound for total mean curvature of spin fill-ins in 2026. Gromov conjectured such an estimate for compact Riemannian manifolds. The bound should depend only on the boundary metric, Gromov proposed.
It should also depend on a lower bound for the fill-in's scalar curvature. Bar proves Gromov's conjecture when the manifolds are spin. He also assumes the mean curvature H is non-negative throughout. Total mean curvature measures how the boundary bends on average everywhere. Bounding it from boundary data alone is a strong rigidity statement. geometers working on scalar curvature will use this result directly. Spin geometry here delivers what general methods could not yet. The proof closes a named conjecture in a natural class.
Catalog record
| CREATOR | Christian Bär |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-06713
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Upper bound for the total mean curvature of spin fill-ins Christian Bär Abstract . Gromov conjectured that the total mean curvature of the boundary of a compact Riemannian manifold can be estimated from above by a constant depending only on the boundary metric and on a lower bound for the scalar curvature of the fill-in. We prove Gromov’s conjecture if the manifolds are spin and the mean curvature H H is non-negative. One can also allow H H to take negative values, but then the constant depends on a (negative) lower bound for H H . 2020 Mathematics Subject Classification 53C20, 53C27 1. Introduction Given a compact Riemannian spin manifold M without boundary and a number \lambda\geq 0 , we may consider all compact connected manifolds X that bound M as a Riemannian spin manifold and satisfy the scalar curvature bound \mathrm{scal}_{X}\geq-\lambda^{2} . Given such an X , let H be the unnormalized mean curvature of the boundary \partial X=M . Our sign convention for the mean curvature is such that the boundary of an (n+1) -dimensional Euclidean ball of radius r has mean curvature H=n/r . Of course, M may not bound any compact manifold X , but if it does so, it follows from [ 14 ] *Theorem 1.1 by Shi, Wang, and Wei that the given metric on M can be extended to a metric on X with \mathrm{scal}_{X}>0 . Their result gives no control on the mean curvature of the boundary. Theorem 3.7 in [ 3 ] by the author and Hanke implies that the metric on X can be deformed in such a way that its scalar curvature remains positive, it still induces the given metric on M , and the mean curvature of the boundary becomes arbitrarily negative. Thus \inf\int_{M}H=-\infty where the infimum is taken over X with \mathrm{scal}_{X}>0 and \partial X=M (as Riemannian manifolds). Gromov conjectured that \int_{M}H cannot be made arbitrarily large. More precisely, the following is believed to hold: Conjecture (Gromov [ 10 ] *p. 232) . Let M be a compact Riemannian manifold without boundary which bounds a compact manifold and let \sigma\in\mathbb{R} . Then there exists a constant C(M,\sigma) such that for each compact Riemannian manifold X with boundary \partial X=M and scalar curvature \mathrm{scal}_{X}\geq\sigma we have \int_{M}H\leq C(M,\sigma). If \dim(M)=1 and \sigma=0 , then this follows from the Gauss-Bonnet theorem. Indeed, assuming without loss of generality that X is connected, the Gauss-Bonnet theorem gives us \int_{M}H=2\pi\chi(X)-\frac{1}{2}\int_{X}\mathrm{scal}_{X}\leq 2\pi. If we are more modest and replace the integral of the mean curvature by its minimum H_{\min}=\min_{M}H , then we indeed have H_{\min}\leq C(M,\sigma) (1) for spin manifolds. This follows from [ 11 ] *Theorem 2 by Hijazi, Montiel, and Roldán. Here the constant C(M,\sigma) depends on the first non-negative eigenvalue of the Dirac operator of M and on \sigma . Combining this with an upper Dirac eigenvalue estimate by the author ( [ 5 ] *Main Theorem), one obtains ( 1 ) where the constant depends on the hyperspherical radius of M and on \sigma , as has been observed by Brendle, Tsiamis, and Wang in [ 6 ] . See also [ 7 ] *Theorem 1.5 by Cecchini, Hirsch, and Zeidler for ( 1 ) in the case \sigma=0 . Clearly, estimate ( 1 ) is only meaningful if the mean curvature is positive. We prove Gromov’s conjecture for spin manifolds if the mean curvature is non-negative. More precisely, we show the following: Theorem . Let X be a compact Riemannian spin manifold of dimension n+1\geq 2 with smooth boundary \partial X=M . Let \lambda\geq 0 be such that the scalar curvature of X satisfies \mathrm{scal}_{X}\geq-\lambda^{2} . Let \eta\geq 0 be such that the mean curvature H of the boundary satisfies H\geq-\eta . Then we have \int_{M}H\leq C(M)+\Big(\sqrt{\tfrac{n}{n+1}}\,\lambda+\eta\Big)\mathsf{vol}(M), where C(M) is a constant depending only on the Riemannian spin manifold M . All previously known results need the assumption that H\geq 0 or even H>0 . If \lambda=0 , H>0 , X is spin, and each connected component of M admits an isometric embedding as a strictly convex hypersurface in \mathbb{R}^{n+1} , then this has been proved by Shi and Tam in [ 13 ] *Theorems 1 and 4.1. The constant C(M) is in this case given by the total mean curvature of the isometric embedding of M into \mathbb{R}^{n+1} . The condition that each connected component of M embeds isometrically as a strictly convex hypersurface into \mathbb{R}^{n+1} has been relaxed by Eichmair, Miao, and Wang in [ 8 ] to require the components to have non-negative scalar curvature and admit embeddings as star-shaped hypersurfaces. The case where M is diffeomorphic to a sphere with \lambda=0 , H>0 , and X spin has been treated by Shi, Wang, and Wei in [ 14 ] *Theorem 4.1. Mantoulidis and Miao investigated in [ 12 ] the case where n=2 and M has arbitraty topology, again if \lambda=0 and H>0 . If M=T^{n} is a torus, X=D^{2}\times T^{n-1} , n\leq 6 , and H>0 , then Gromov’s conjecture has been proved by Wang in [ 15 ] *Theorems 1.6 and 1.8. In this case, \lambda is allowed to be negative. If the torus M carries a flat metric, then C(M) is related to the systole of M . In the mean time, Frenck, Hanke, and Hirsch have presented a different proof of Gromov’s conjecture if H\geq 0 , using surgery and deformation methods ( [ 9 ] ). Their approach has the advantage that it also works in the non-spin case if n\leq 6 . Acknowledgments: This work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project ID 569831821. 2. A boundary value problem Let X be a compact connected Riemannian spin manifold of dimension n+1\geq 2 with boundary \partial X=M . The spinor bundle \Sigma X of X is a Hermitian vector bundle with metric connection. We choose the convention that the fiberwise scalar product of \Sigma X is antilinear in the first argument and linear in the second argument. Let \gamma\colon TX\to\End(\Sigma X) denote the Clifford multiplication. It satisfies the Clifford relations \gamma(v)\gamma(w)+\gamma(w)\gamma(v)=-\langle v,w\rangle\id_{\Sigma X} for all tangent vectors v,w\in T_{p}X and p\in X . The Dirac operator of X is denoted by D . Denote the inward unit normal vector field along the boundary by \nu . Then s:=i\gamma(\nu)\in\End(\Sigma X) is a self-adjoint involution. It anticommutes with Clifford multiplication by tangent vectors of M . Hence, the eigenspaces corresponding to the eigenvalues \pm 1 of s have the same dimension. We obtain an orthogonal vector bundle decomposition \Sigma X|_{M}=\Sigma^{+}M\oplus\Sigma^{-}M, (2) where s acts on \Sigma^{\pm}M as \pm 1 . We denote the orthogonal projections onto these subbundles by P^{\pm}\colon\Sigma X|_{M}\to\Sigma^{\pm}M. If n is even, then \Sigma X|_{M} can be canonically identified with the spinor bundle \Sigma M of M . If n is odd, then \Sigma X|_{M} can be canonically identified with \Sigma M\oplus\Sigma M (which is not the same decomposition as the splitting in ( 2 )). Denoting the intrinsic Dirac operator of M by \tilde{D}_{M} and using these identifications, we set D_{M}:=\tilde{D}_{M} if n is even and D_{M}=\begin{pmatrix}\tilde{D}_{M}&0\\ 0&-\tilde{D}_{M}\end{pmatrix} if n is odd. Then D_{M} anticommutes with s in both cases (see [