Kuga-Satake construction on families of K3 surfaces of Picard rank 14 14

Flora Poon · 2025

Read Kuga-Satake construction on families of K3 surfaces of Picard rank 14 14 by Flora Poon (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-2504-08270.txt.

Introduction

Flora Poon connects two landscapes of algebraic geometry that look unrelated. On one side sit K3 surfaces with a fixed lattice of curves, their Picard rank fourteen; on the other, polarized abelian eightfolds whose symmetries obey quaternion rules. A numerical coincidence between their governing domains hints the two moduli spaces should talk.

Poon builds the bridge with the Kuga-Satake construction, which manufactures an abelian variety from a K3 surface's geometry. She shows the domain coincidence induces a genuine map between the moduli spaces, not just a formal analogy. The paper then works the construction through for concrete families, including ones where the surfaces carry even richer curve lattices.

Catalog record

CREATORFlora Poon
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE84 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-08270

Browse the full library · About the Tower of Babel

Full text

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

Kuga-Satake construction on families of K3 surfaces of Picard rank 14 14 Flora Poon Abstract The isometry between the type IV 6 and the type II 4 hermitian symmetric domains suggests a possible relation between suitable moduli spaces of K3 surfaces of Picard rank 14 14 and of polarised abelian 8 8 -folds with totally definite quaternion multiplication. We show how this isometry induces a geometrically meaningful map between such moduli spaces using the Kuga-Satake construction. Furthermore, we illustrate how the the modular mapping can be realised for any specific families of K3 surfaces of Picard rank 14, which can be specialised to families of K3 surfaces of higher Picard rank. \DeclareCaptionType diagram[Diagram][] 1 Introduction Consider a lattice P of signature (1,r-1) with 1\leq r\leq 20 which primitively embeds into the K3 lattice \Lambda_{K3} . We will consider algebraic K3 surfaces over \mathbb{C} with quasi-polarisation by P , and we denote the moduli space of such K3 surfaces as \mathcal{K}_{P} . It is known [ D ] that each irreducible component of \mathcal{K}_{P} is a locally symmetric variety (LSV), which is a Riemannian manifold that is locally reflectionally symmetric around any point. Algebraically, they can be expressed [ He , Theorem VIII.7.1] as biquotients in the form \Gamma\backslash G/K , where G is a connected simple Lie group, K is a non-discrete maximal compact subgroup of G , and \Gamma is an arithmetic subgroup of G . In this situation, G/K is a Hermitian symmetric domain (HSD). Considering the different options for the Lie group G in this characterisation gives rise to a classification for both HSDs and LSVs: see [ He , Table X.6.V] . Under this classification, certain types of LSVs can be viewed as moduli spaces [ Lo , Section 3] : the HSD G/K serves as the period domain, and the left translation action of the arithmetic subgroup \Gamma<G on G/K identifies period points that correspond to isomorphic polarised varieties. In particular, the period domain of \mathcal{K}_{P} , which is the set of weight two Hodge structures on the second integral cohomology group of a P -polarised K3 surface, is the union of two copies of the HSD (of type IV 20-r in the classification in [ He ] ) given by {\mathop{\mathrm{SO}}}^{+}(2,20-r)/({\mathop{\mathrm{SO}}}(2)\times{\mathop{\mathrm{SO}}}(20-r)). The lattice polarisation by P determines an arithmetic subroup \Gamma(P)<{\mathop{\mathrm{SO}}}^{+}(2,20-r) containing Hodge isometries that correspond to isomorphisms of P -polarised K3 surfaces. We are especially interested in the type IV 20-r series of LSVs, because for r large, \emph{i.e.} close to 20 , the HSD overspace is isometric to the period domain of a different modular variety as Riemannian manifolds. For example [ GHS ] , the type IV 4 and type IV 5 HSDs ( r=16 and r=15 ) coincide with the period domains of a modular variety of deformation of generalised Kummer varieties and of hyperkähler manifolds of OG6 type respectively. When r=14 , the type IV 6 HSD is isometric to the type II 4 HSD [ He , Exercise X.D.2(b)] . The latter HSD, characterised as the Lie group quotient {\mathop{\mathrm{SO}}}^{*}(8)/\mathop{\hbox{}\mathrm{U}}(4) , can be identified to the set of weight one Hodge structures on the first integral cohomology group of a polarised abelian 8 -fold with totally definite quaternion multiplication, which is also the period domain of a modular variety \mathcal{A}_{\mathcal{M},\mathcal{T}} of polarised abelian 8 -folds whose PEL structures are controlled by certain attributes \mathcal{M} and \mathcal{T} . In fact, \mathcal{A}_{\mathcal{M},\mathcal{T}} is a type II 4 LSV, \emph{i.e.} it is an arithmetic quotient of {\mathop{\mathrm{SO}}}^{*}(8)/\mathop{\hbox{}\mathrm{U}}(4) (see [ BL , Section 9.7] ). Let us denote the isometry between the HSDs in the case of r=14 by \widetilde{F} . If \widetilde{F} is also equivariant with the natural conjugation actions of the Lie groups {\mathop{\mathrm{SO}}}^{+}(2,6) and {\mathop{\mathrm{SO}}}^{*}(8) , then \widetilde{F} descends to a map from \mathcal{K}_{P} to a certain moduli space \mathcal{D}_{\mathcal{M},\mathcal{T}} of abelian 8 -folds with totally definite quaternion multiplication. {\mathop{\mathrm{SO}}}^{+}(2,6)/({\mathop{\mathrm{SO}}}(2)\times{\mathop{\mathrm{SO}}}(6)) \tilde{F} {\mathop{\mathrm{SO}}}^{*}(8)/\mathop{\hbox{}\mathrm{U}}(4) /\Gamma(P) /\widetilde{F}_{*}(\Gamma(P)) \mathcal{K}_{P} \mathcal{A}_{\mathcal{M},\mathcal{T}} F {diagram} \widetilde{F} , if equivariant, descends. The goal of this work is to realise the map F as a modular mapping. More specifically, by identifying the domain and the target of F as (an irreducible component of) the modular varieties \mathcal{K}_{P} and \mathcal{A}_{\mathcal{M},\mathcal{T}} , we will explicitly describe how F takes a K3 surface X in K_{P} to an abelian 8 -fold A in \mathcal{A}_{\mathcal{M},\mathcal{T}} . The Global Torelli Theorem which associates X with its weight 2 polarised Hodge structure on H^{2}(X,\mathbb{Z}) , and A with its weight 1 polarised Hodge structure on H^{1}(A,\mathbb{Z}) , reduces the problem to a purely lattice-theoretic one. This allows us to give a geometric interpretation of the map F in terms of the Kuga-Satake construction [ KS ] , a process that produces a weight 1 Hodge structure from the Clifford algebra of a weight 2 Hodge structure of K3 type. The r=14 case is special. For slightly larger r , \emph{i.e.} r=15 and 16 , the question of finding an explicit modular mapping induced from the isometry of the HSDs is premature, because no explicit families of deformation of generalised Kummer varieties or OG6 varieties are known. Moreover, the case r=14 is the smallest r such that there exists an isometry from a type IV 20-r HSD to a different HSD. More importantly, the two HSDs have the same dimension, so it is possible that the isometry induces an isomorphism F of modular varieties upon choosing suitable arithmetic subgroup \Gamma(P) in {\mathop{\mathrm{SO}}}^{+}(2,6) . Furthermore, Satake has proved [ Sa ] that r=14 is one of the two cases when the isometry between the period domains is an equivariant holomorphic embedding, so [ KK ] the induced map F between the corresponding arithmetic quotients extends to their Baily-Borel compactifications. The layout of the paper is as follows: In Section 2 , we will give the necessary set up for the construction of F , which includes a brief introduction of the moduli spaces \mathcal{K}_{P} and \mathcal{A}_{\mathcal{M},\mathcal{T}} (Section 2.1 ), the notion of Clifford algebras and some related concepts (Section 2.2 ), and the classical Kuga-Satake construction (Section 2.3 ). In Section 3 , we will give an explicit description of a geometrically meaningful map F from any irreducible \mathcal{K}_{P} to \mathcal{A}_{\mathcal{M},\mathcal{T}} . We will also prove that the constructed map descends from the diffeomorphism \widetilde{F} between the HSD overspaces. In Section 4 , we will describe some technicalities of the construction of F by working on an example. Finally in Section 5 , we will also observe some special behaviour of the map F as we specialise our construction to subfamilies of \mathcal{K}_{P} which parametrise K3 surfaces with richer geometric properties. 2 Background We will use the following notations: If L/K is a finite extension of fields, and G is an algebraic group defined over K , then we denote by G(L) the corresponding algebraic group defined over L . If in another situation that R is a ring, V is an R -module, and K is an R -algebra, we often write V_{K} for V\otimes_{R}K . 2.1 Moduli spaces We will recall a minimal list of facts and properties of the moduli spaces \mathcal{K}_{P} and \mathcal{A}_{\mathcal{M},\mathcal{T}} for the construction of the map F , most

Read the complete text