The degeneration of Calabi-Yau 3-folds via 3-forms and the SYZ conjecture

Teng Fei · 2025

Read The degeneration of Calabi-Yau 3-folds via 3-forms and the SYZ conjecture by Teng Fei (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-2503-11992.txt.

Introduction

Teng Fei studies how Calabi-Yau three-folds degenerate, using the language of 3-forms on a symplectic six-manifold. Different 3-forms fall into orbit types with distinct geometry. Fei shows that unstable 3-forms, which arise naturally from specific ways Calabi-Yau structures break down, carry rich geometric structure. That structure, in turn, offers an angle on the SYZ conjecture, a central open idea about how these spaces can be understood through fibrations. The note is short, but it links a class of forms to a major question in geometry.

Catalog record

CREATORTeng Fei
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE78 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-11992

Browse the full library · About the Tower of Babel

Full text

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

The degeneration of Calabi-Yau 3-folds via 3-forms and the SYZ conjecture Teng Fei Abstract In this paper, we investigate the geometries associated with 3-forms of various orbital types on a symplectic 6-manifold. We demonstrate that certain unstable 3-forms, which naturally emerge from specific degenerations of Calabi-Yau structures, exhibit remarkably rich geometric properties. This, in turn, offers a novel perspective on the SYZ conjecture. 1 Introduction The classification of 3-forms in real dimension six has been well-established [ Rei07 , Gur35 , Gur64 , Cha98 , Bry06 ] . A notable feature in this dimension is the existence of two open (also known as stable) orbits under the natural action of the general linear group. This fact has been utilized by many authors, including Hitchin [ Hit00 ] and Donaldson-Lehmann [ DL24b , DL24c , DL24a ] , to investigate geometric and deformation problems in complex and Calabi-Yau geometries. In this paper, we systematically investigate various geometric structures associated with 3-forms on a 6-dimensional symplectic manifold, including those from unstable orbits under the general linear group action. We introduce several natural integrability conditions on 3-forms and explore the relationships among them. In particular, we focus on the geometry of certain unstable 3-forms that naturally arise from the degeneration of Calabi-Yau structures. We demonstrate that such geometries possess a canonical Lagrangian foliation, which emerges as a consequence of the degeneration of complex structures. Moreover, we prove that the leaves of this foliation naturally admit a pair of dual Hessian structures, where the local potentials of the metric automatically satisfy a real Monge-Ampère equation. When the Lagrangian foliation is a Lagrangian torus fibration, we further show that the fibration is automatically semi-flat, and the base of the fibration is naturally equipped with an L^{2} -metric and a pair of dual Hessian structures. Additionally, we provide explicit examples of this type of Calabi-Yau 3-fold degeneration. This paper is organized as follows. In Section 2, we review the classification of 3-forms in dimension six under both the general linear group and the symplectic group actions. We introduce homogeneous equivariant polynomials to describe the orbits of 3-forms. Section 3 is dedicated to integrability conditions on 3-forms. In particular, we introduce the concept of F -harmonic 3-forms as an ideal integrability condition. In Section 4, we introduce the notion of wonderful degenerations of Calabi-Yau structures as a convenient set-up for our discussion. Section 5 is devoted to investigation of the geometry associated with an F -harmonic 3-form in the orbit \mathcal{O}_{0}^{+} . Such 3-forms naturally arise from certain degenerations of Calabi-Yau structures and induce remarkably rich geometric structures on the underlying symplectic manifold. These include a canonical Lagrangian foliation, where the leaves are equipped with a pair of dual Hessian structures and a parallel volume form. Furthermore, when the Lagrangian foliation is a torus fibration, we show that almost all the geometric predictions of the SYZ conjecture are automatically satisfied. Finally in Section 6, we present a non-trivial example of wonderful degeneration and its associated SYZ fibration. Acknowledgment: The author would like to thank Robert Bryant, Tristan Collins, Yu-Shen Lin, Duong H. Phong, Li-Sheng Tseng and Xiangwen Zhang for valuable discussions. This work is generously supported by the Simons Collaboration Grant 830816 and the NSF LEAPS Grant 2418918. 2 Classification of three-forms In this section, we establish a collection of basic facts about the classification of 3-forms on a 6-dimensional real vector space. A few of the statements presented here are probably new. However, due to their elementary nature, we leave the proof as exercises for the interested readers. Let V be a real vector space of dimension 6 . We denote by \bigwedge^{3}V^{*} the space of 3-forms on V . The group \mathrm{GL}(V) naturally acts on \bigwedge^{3}V^{*} . This action is “stable” in the sense that it has open orbits. In fact, all the orbits of this action have been classified classically, see for example [ Rei07 , Gur35 , Gur64 , Cha98 , Bry06 ] . To better describe these orbits, we introduce the following definitions. Definition 2.1 . A 3-form \varphi\in\bigwedge^{3}V^{*} is called stable if it lies in an open orbit under the \mathrm{GL}(V) -action. For any 3-form \varphi\in\bigwedge^{3}V^{*} , its kernel \ker\varphi is defined to be \ker\varphi=\{v\in V:\iota_{v}\varphi=0\}. Definition 2.2 . A 3-form \varphi is called non-degenerate if \ker\varphi=\{0\} . As \varphi induces a non-degenerate 3-form \bar{\varphi}\in\bigwedge^{3}(V/\ker\varphi)^{*} , it is easy to see that the only possible dimensions for \ker\varphi are 0 , 1 , 3 , and 6 . For later use, we introduce several equivariant homogeneous polynomials on the space \bigwedge^{3}V^{*} : (a). K is an equivariant homogeneous degree 2 polynomial of \varphi such that K(\varphi)\in\mathrm{End}~V\otimes\bigwedge^{6}V^{*} for any \varphi . It is defined by K(\varphi)(v)=-\iota_{v}\varphi\wedge\varphi\in\bigwedge\nolimits^{5}V^{*}\cong V\otimes\bigwedge\nolimits^{6}V^{*} for any v\in V . (b). F is an equivariant homogeneous degree 3 polynomial of \varphi such that F(\varphi)\in\bigwedge^{3}V^{*}\otimes\bigwedge^{6}V^{*} for any \varphi . It is defined by F(\varphi)(v_{1},v_{2},v_{3})=-2\varphi(K(\varphi)(v_{1}),v_{2},v_{3}) for any v_{1},v_{2},v_{3}\in V . Modulo tensoring the one dimensional space \bigwedge^{6}V^{*} , F(\varphi) is also a 3-form. Therefore it makes sense to talk about \ker F(\varphi) . By this definition, we know that \ker F(\varphi)=K(\varphi)^{-1}(\ker\varphi\otimes\bigwedge\nolimits^{6}V^{*}). (c). Q is an equivariant homogeneous degree 4 polynomial of \varphi such that Q(\varphi)\in(\bigwedge^{6}V^{*})^{\otimes 2} for any \varphi . It is defined by Q(\varphi)=-\varphi\wedge F(\varphi). Definition 2.3 . An element of (\bigwedge^{6}V^{*})^{\otimes 2} is called positive if it can be written as \tau^{2} for some 0\neq\tau\in\bigwedge^{6}V^{*} . An element of (\bigwedge^{6}V^{*})^{\otimes 2} is called negative if its negative is positive. It has been known for a long time that the \mathrm{GL}(V) -action on \bigwedge^{3}V^{*} has 6 distinct orbits. Two of them are stable, which are exactly the open sets defined by Q(\varphi)<0 and Q(\varphi)>0 respectively. These two open orbits will be denoted by \mathcal{O}_{-} and \mathcal{O}_{+} throughout this paper. The union of the other 4 orbits is the hypersurface Q(\varphi)=0 . In fact, these 4 orbits can be distinguished by the dimension of their kernels. As a consequence, we shall denote these orbits by \mathcal{O}_{0} , \mathcal{O}_{1} , \mathcal{O}_{3} , \mathcal{O}_{6}=\{0\} , where the subscript labels the dimension of their kernel spaces. Regarding these orbits, we have the following table: Orbit Dimesion Normal Form \mathcal{O}_{-} 20 e^{135}-e^{146}-e^{236}-e^{245} \mathcal{O}_{+} 20 e^{123}+e^{456} \mathcal{O}_{0} 19 e^{146}+e^{236}+e^{245} \mathcal{O}_{1} 15 e^{135}+e^{245} \mathcal{O}_{3} 10 e^{135} \mathcal{O}_{6}=\{0\} 0 0 Definition 2.4 . For \varphi\in\bigwedge^{3}V^{*} , the space \mathrm{Ann}~\varphi is defined to be \mathrm{Ann}~\varphi:=\{\alpha\in V^{*}:\alpha\wedge\varphi=0\}. Dually we define (\mathrm{Ann}~\varphi)^{\perp}:=\{v\in V:\alpha(v)=0\textrm{ for any }\alpha\in\mathrm{Ann}~\varphi\}. Noticing that K(\varphi) is a linear map from V to V\otimes\bigwedge^{6}V^{*} and that V\cong V\otimes\bigwedge^{6}V^{*} canonically up to scaling, we may view both \ker K(\varphi) and \mathrm{Im}~K(\varphi) as subspaces of V . Therefore associated with each \varphi\in\bigwedge^{3}V^{*} , there are four natural subspaces of V : \ker\varphi , \ker K(\varphi) , \mathrm{Im}~K(\varphi) , (\mathrm{Ann}~\varphi)^{\perp} and one subspace of V^{*} : \mathrm{Ann}~\varphi . Regarding dimensions of these vector spaces, it is easy to obtain the following table using the normal forms above. Orbit \dim\ker\varphi \dim\ker K(\varphi) \dim\mathrm{Im}~K(\varphi) \dim~(\mathrm{Ann}~\varphi)^{\perp} \mathcal{O}_{-} 0 0 6 6 \mathcal{O}_{+} 0 0 6 6 \mathcal{O}_{0} 0 3 3 6 \mathcal{O}_{1} 1 5 1 5 \mathcal{O}_{3} 3 6 0 3 \mathcal{O}_{6}=\{0\} 6 6 0 0 Moreover, one can check case by case that if any two spaces in the same line has the same

Read the complete text