A sphericity criterion for strictly pseudoconvex hypersurfaces in ℂ 2 \mathbb{C}^{2} via invariant curves
Florian Bertrand, Giuseppe Della Sala, Bernhard Lamel · 2025
Read A sphericity criterion for strictly pseudoconvex hypersurfaces in ℂ 2 \mathbb{C}^{2} via invariant curves by Florian Bertrand, Giuseppe Della Sala, Bernhard Lamel (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-2501-08842.txt.
Introduction
Florian Bertrand, Giuseppe Della Sala, and Bernhard Lamel proved in 2025 a clean test for when a curved surface in two-dimensional complex space is secretly a piece of a sphere. Their surfaces are strictly pseudoconvex, a strong bending condition. Special curves called chains lie on such surfaces, and flat discs can be hung from them. The theorem says: if every chain on the surface bounds one of these stationary discs, the surface must be locally spherical, the most symmetric shape possible. One geometric habit thus certifies perfect roundness.
Catalog record
| CREATOR | Florian Bertrand, Giuseppe Della Sala, Bernhard Lamel |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-08842
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A sphericity criterion for strictly pseudoconvex hypersurfaces in ℂ 2 \mathbb{C}^{2} via invariant curves Florian Bertrand, Giuseppe Della Sala, Bernhard Lamel Abstract We prove that if every chain on a strictly pseudoconvex hypersurface M M in ℂ 2 \mathbb{C}^{2} coincides with the boundary of a stationary disc, then M M is locally spherical. Introduction Every strictly pseudoconvex hypersurface M\subset\mathbb{C}^{2} bounding a domain \Omega\subset\mathbb{C}^{2} carries two natural, biholomorphically invariant families of real curves: the so-called chains and boundaries of stationary discs . These come from very different types of geometrical constructions. Chains have been introduced by Chern and Moser [ 5 ] as the CR geometry analogue of geodesics in Riemannian geometry. Stationary discs, on the other hand, are the solutions to the Euler-Lagrange equations of Kobayashi extremal discs in \Omega . If M is in addition real-analytic, it carries a third natural biholomorphically invariant family of real curves: traces of Segre varieties. A theorem of Faran [ 6 ] shows that if the traces of Segre varieties agree with the chains, then M is locally spherical. In a former paper [ 1 ] we showed that if the traces of Segre varieties agree with the traces of stationary discs, then M is also locally spherical. In this paper, we address the remaining question: If the traces of stationary discs coincide with chains, is M also necessarily spherical? We have been asked this repeatedly when presenting the results in [ 1 ] , and it turns out that the answer is also yes. The natural setting for this question is for sufficiently smooth hypersurfaces. Theorem 1 . Assume that M is a strictly pseudoconvex hypersurface of class {C}^{12} in \mathbb{C}^{2} . If the chains of M are boundaries of stationary discs, then M is locally spherical. We remark that M in Theorem 1 is not assumed to be closed (so the theorem is a local result). In order to prove this theorem, we cannot utilize the cited results. Instead, we rely on Fefferman’s characterization of chains as projections of light rays of an associated Lorentz metric and analyzing its Hamiltonian. We construct a special family of chains centered at the origin and show that if each of the members of this family is the trace of a stationary disc, then the origin is an umbilical point. The organization of this paper is to review the basics in section 1 , summarize facts about the chains in section 2 , and give the proof of a (slightly sharpened) version of the theorem in section 3 . 1 Preliminaries 1.1 Intrinsic geometry of strictly pseudoconvex hypersurfaces In this section we give a quick review of the basic local biholomorphic equivalence theory for strictly pseudoconvex hypersurfaces in \mathbb{C}^{2} . We thank one of the referees for the suggestion to include some background material in this paper, and hope the reader will enjoy it. A hypersurface M\subset\mathbb{C}^{2} inherits the complex structure of \mathbb{C}^{2} on its complex tangent spaces T^{c}_{p}M=T_{p}M\cap iT_{p}M . The complex tangent bundle T^{c}M is defined by the vanishing of a not uniquely determined 1 -form \theta , which we usually call characteristic form ; the annihilator (T^{c}M)^{\perp}=T^{0}M=\mathcal{N}M\subset T^{*}M is called the characteristic or conormal bundle of M . If d\theta induces a hermitian inner product on T^{c}M by h(X,Y)=i\theta([X,\bar{Y}])=-id\theta(X,\bar{Y}) , called the Levi form of M with respect to \theta , then we say that M is strictly pseudoconvex . Thus, geometrically speaking, a strictly pseudoconvex hypersurface M\subset\mathbb{C}^{2} can be thought of as a 3 -dimensional manifold with a contact structure with some additional compatibility conditions. The choice of a contact form gives rise to a pseudohermitian structure (M,\theta) . The ambivalence in choosing a contact form gives rise to fascinating mathematics which mixes aspects of complex, conformal, contact, and symplectic geometry. The equivalence problem for strictly pseudoconvex hypersurfaces in \mathbb{C}^{2} was solved in a series of groundbreaking papers by E. Cartan [ 3 , 4 ] applying his method of equivalence; later, Tanaka [ 16 ] and Chern-Moser [ 5 ] gave solutions of the problem for strictly pseudoconvex and Levi-nondegenerate hypersurfaces in higher dimensions. For the convenience of the reader, we recall some of the necessary background, with a view towards the Fefferman construction of chains which we are going to use. If we start with an arbitrary characteristic form \theta we can consider a real line bundle E over M consisting of the multiples u\theta , where u>0 . The form \omega=u\theta is intrinsically defined, and (on E ) we have d\omega=ig_{1\bar{1}}\omega^{1}\wedge\bar{\omega}^{1}+\omega\wedge\varphi with a real one form \varphi ; the forms \omega , \omega^{1} , \bar{\omega}^{1} , and \varphi span \mathbb{C}T^{*}E . Since we assume that M is strictly pseudoconvex, we have g_{1\bar{1}}\neq 0 , and we will assume for simplicity that g_{1\bar{1}}=1 ; the more general case where g_{1\bar{1}} is not assumed to be constant follows in a similar but more involved way. Since this short review is only meant to recall how the main computations work, we opted to keep the simple variant. We follow the notation of Chern-Moser [ 5 ] so that the reader can pick up the necessary modifications in that source easily. Now any other frame of \mathbb{C}T^{*}E satisfying the condition above is given by \begin{pmatrix}\tilde{\omega}\\ \tilde{\omega}^{1}\\ \bar{\tilde{\omega}}^{1}\\ \tilde{\varphi}\end{pmatrix}=\begin{pmatrix}1&0&0&0\\ \lambda&\mu&0&0\\ \bar{\lambda}&0&\bar{\mu}&0\\ s&i\mu\bar{\lambda}&-i\bar{\mu}\lambda&1\end{pmatrix}\begin{pmatrix}\omega\\ \omega^{1}\\ \bar{\omega}^{1}\\ \varphi\end{pmatrix} where |\mu|^{2}=1 . The group of matrices of this form is denoted by G_{1} , and we can form the principal G_{1} -bundle Y over E ; thus s , \lambda , and \mu are fiber coordinates. One has the integrability condition d\omega^{1}=\omega^{1}\wedge\alpha+\omega\wedge\beta, for every frame as above, with some not uniquely determined forms \alpha and \beta . One then obtains a uniquely determined frame \omega,\omega^{1},\bar{\omega}^{1},\varphi,\alpha,\beta,\psi of \mathbb{C}T^{*}Y satisfying a number of identities: Theorem 2 ( [ 3 ] ) . There exists a unique frame \omega,\omega^{1},\bar{\omega}^{1},\varphi,\alpha,\beta,\psi of \mathbb{C}T^{*}Y and invariantly defined functions Q,R such that \displaystyle d\omega \displaystyle=i\omega^{1}\wedge\bar{\omega}^{1}+\omega\wedge\varphi \displaystyle d\omega^{1} \displaystyle=\omega^{1}\wedge\alpha+\omega\wedge\beta \displaystyle\varphi \displaystyle=\alpha+\bar{\alpha} \displaystyle d\varphi \displaystyle=i\omega^{1}\wedge\bar{\beta}-i\bar{\omega}^{1}\wedge\beta+\omega\wedge\psi \displaystyle d\alpha \displaystyle=i\bar{\omega}^{1}\wedge\beta+2i\omega^{1}\wedge\bar{\beta}-\frac{\psi}{2}\wedge\omega \displaystyle d\beta \displaystyle=\bar{\alpha}\wedge\beta-\frac{\psi}{2}\wedge\omega^{1}+Q\bar{\omega}^{1}\wedge\omega \displaystyle d\psi \displaystyle=\varphi\wedge\psi+2i\beta\wedge\bar{\beta}+(R\omega^{1}+\bar{R}\bar{\omega}^{1})\wedge\omega Definition 3 . A curve \gamma is called a chain if it solves the system of ODEs \omega^{1}=\beta=0 . We note that the equations for the real forms \omega , \varphi , and \psi along a chain simply read d\omega=\omega\wedge\varphi,\quad d\varphi=\omega\wedge\psi,\quad d\psi=\varphi\wedge\psi. One can use these equations to therefore introduce a canonical parameter along the chain defined up to a linear fractional map. We now give, as promised, the details for the construction of the canonical forms in Theorem 2 , basically to set the stage. We also point the reader to the book by Jacobowitz [ 13 ] . First, one observes that exterior differentiation of the frame conditions \displaystyle d\omega \displaystyle=i\omega^{1}\wedge\bar{\omega}^{1}+\omega\wedge\varphi (1.1) \displaystyle d\omega^{1} \displaystyle=\omega^{1}\wedge\alpha+\omega\wedge\beta yields \displaystyle 0 \displaystyle=id\omega^{1}\wedge\bar{\omega}^{1}-i\omega^{1}\wedge d\bar{\omega}^{1}+d\omega\wedge\varphi-\omega\wedge d\varphi (1.2) \displaystyle=i\left(\omega^{1}\wedge\alpha+\omega\wedge\beta\right)\wedge\bar{\omega}^{1}-i\omega^{1}\wedge\left(\bar{\omega}^{1}\wedge\bar{\alpha}+\omega\wedge\bar{\beta}\right) \displaystyle+\left(i\omega^{1}\wedge\bar{\omega}^{1}+\omega\wedge\varphi\right)\wedge\varphi-\omega\wedge d\varphi \displaystyle=i(-\alpha-\bar{\alpha}+\varphi)\wedge\omega^{1}\wedge\bar{\omega}^{1}+\left(-d\varphi+i\beta\wedge\bar{\omega}^{1}-i\bar{\beta}\wedge\omega^{1}\right)\wedge\omega \displaystyle 0 \displaystyle=d\omega^{1}\wedge\alpha-\omega^{1}\wedge d\alpha+d\omega\wedge\beta-\omega\wedge d\beta \displaystyle=\omega\wedge\beta\wedge\alpha-\omega^{1}\wedge d\alpha+(i\omega^{1}\wedge\bar{\omega}^{1}+\omega\wedge\varphi)\wedge\beta-\omega\wedge d\beta \displaystyle=(-d\alpha+i\bar{\omega}^{1}\wedge\beta)\wedge\omega^{1}+(\beta\wedge\alpha+\varphi\wedge\beta-d\beta)\wedge\omega, It follows from the first equation in ( 1.2 ) that -\alpha-\bar{\alpha}+\varphi=A\omega^{1}+\bar{A}\bar{\omega}^{1}+C\omega with C=\bar{C} , i.e. with the choice \tilde{\alpha}=\alpha+A\omega^{1}+\frac{C}{2}\omega we have \varphi=\tilde{\alpha}+\bar{\tilde{\alpha}} . It is easy to see that \varphi with this property are unique up to multiples of \omega , and for such a choice of \varphi , we have -d\varphi+i\beta\wedge\bar{\omega}^{1}-i\bar{\beta}\wedge\omega^{1}=-\omega\wedge\psi for a real 1 -form \psi . Summarizing, we have imposed the