Modularity of special cycles on Shimura varieties: a survey

François Greer, Salim Tayou · 2026

Read Modularity of special cycles on Shimura varieties: a survey by François Greer, Salim Tayou (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-2603-01251.txt.

Introduction

Greer and Tayou survey special cycles and the generating series built from their classes, asking how modularity behaves for them. Their central focus is a conjecture of Kudla, examined for orthogonal Shimura varieties and for unitary ones under toroidal compactification. The pair also put forward new conjectures covering more general quotients of period domains, and related phenomena for special cycles in other Shimura variety types.

Catalog record

CREATORFrançois Greer, Salim Tayou
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62200
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.01251

Browse the full library · About the Tower of Babel

Full text

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

Modularity of special cycles on Shimura varieties: a survey François Greer, Salim Tayou Abstract . We survey recent results on a conjecture of Kudla regarding the modularity of generating series of special cycle classes in toroidal compactifications of orthogonal and unitary Shimura varieties. Along the way, we formulate several conjectures on related phenomena for special cycles in other types of Shimura varieties, as well as on more general quotients of period domains. 1. Introduction In their groundbreaking paper, Hirzebruch and Zagier [ HZ76 ] proved that the generating series of corrected special cycles on a compactified Hilbert modular surface is a modular form of weight 2 , with values in the cohomology. Their work prompted a series of developments culminating in the theorem of Kudla and Millson [ KM90 ] , which gave a general modularity statement for special cycles on orthogonal and unitary Shimura varieties. Our goal in this note is twofold: to propose an analogue of the Kudla-Millson theorem for other classes of Shimura varieties, and to formulate a conjectural framework for how the special cycles can be completed and corrected in toroidal compactifications of these Shimura varieties. Let ({\mathbf{G}},{\mathcal{D}}) be a Shimura datum; that is, {\mathbf{G}} is a semisimple algebraic group over \mathbb{Q} , and {\mathcal{D}} is a Hermitian symmetric domain equipped with a transitive action of G={\mathbf{G}}(\mathbb{R}) : G\rightarrow\mathrm{Aut}({\mathcal{D}}), with compact stabilizer K , satisfying the axioms of Deligne [ Del71 , Section 1.5] . Of particular interest are the irreducible Hermitian symmetric domains {\mathcal{D}}=G/K which have been completely classified: they consist of four infinite families (Types I–IV), together with two exceptional cases ( E_{6} and E_{7} ). We will focus only on the infinite families here. Type G {\mathbf{G}}_{\mathbb{C}} K \dim_{\mathbb{C}}(X) moduli problem I \mathrm{SU}(a,b) \mathrm{SL}_{a+b} S(\mathrm{U}(a)\times\mathrm{U}(b)) ab pol. a.v. with CM II \mathrm{SO}^{*}(2n) \mathrm{SO}_{2n} \mathrm{U}(n) n(n-1)/2 pol. a.v. with QM III \mathrm{Sp}(2n,\mathbb{R}) \mathrm{Sp}_{2n} \mathrm{U}(n) n(n+1)/2 pol. a.v. IV \mathrm{SO}(n,2) \mathrm{SO}_{n+2} S(\mathrm{O}(n)\times\mathrm{O}(2)) n Kuga-Satake a.v. Table 1. The four infinite families, with n>0 and a\geq b>0 . The choice of a faithful representation {\mathbf{G}}\rightarrow\mathrm{GL}(V_{\mathbb{Q}}) , where V is a free \mathbb{Z} -module, determines an arithmetic subgroup \Gamma of {\mathbf{G}}(\mathbb{Q}) consisting of the elements that preserve the lattice V . The double quotient X=X_{\Gamma}=\Gamma\backslash G/K is a complex quasi-projective variety by the theorem of Baily-Borel [ BB66 ] . There are natural families of abelian varieties \mathcal{B} with polarization, endomorphism algebra, and level structure (PEL) over the Shimura varieties of Types I–III. In each of these cases, X supports a PVHS of weight 1, where the fiber at each point x\in X is isomorphic to V\simeq H^{1}(B_{x},\mathbb{Z}). The Type IV case is more subtle; one needs to take the double cover of \mathrm{SO}(n,2) , the spin group \mathrm{Spin}(n,2) which admits a representation into the linear group of \mathrm{Cl}(V) , the Clifford algebra of V . The Spin Shimura variety admits a family \mathcal{B} of polarized abelian varieties of dimension 2^{n+1} given by the Kuga-Satake construction. We have an isomorphism: \mathrm{Cl}(V)\simeq H^{1}(B_{x},\mathbb{Z}), and an embedding V\hookrightarrow\mathrm{End}(H^{1}(B_{x},\mathbb{Z})) . For each of the types I–IV, we will recall the construction of certain natural classes of special cycles on X , which are always effective, and how to assemble them into generating series. Some of these constructions go back to the original work of Kudla, Millson, Zagier, and Hirzebruch, while others have only recently been considered by various groups of authors. The first goal of this survey is to place these constructions into a unified framework. The simplest class of special cycles are indexed by integers m\in\mathbb{N} , and we denote them by {\mathcal{Z}}(m) . The classes [{\mathcal{Z}}(m)] in the relevant cohomology or Chow group of X constitute 1 1 1 This is a theorem in Types (I, IV) and a conjecture in Types (II, III) the Fourier coefficients of a classical modular form of weight k (recorded below) and level determined by the arithmetic group \Gamma . \Phi(q)=\sum_{m\geq 0}[{\mathcal{Z}}(m)]q^{m}\in\mathrm{CH}^{c}(X)\otimes_{\mathbb{Q}}\mathrm{Mod}_{k}(\mathrm{SL}_{2}(\mathbb{Z})) The constant term [{\mathcal{Z}}(0)] is defined separately as the top Chern class of a tautological bundle on X , so it only makes sense as a cycle class, and it is not effective in general. Type G \dim_{\mathbb{C}}(X) c=\mathrm{codim}(Z(m)) weight k moduli description I \mathrm{SU}(a,b) ab a a+b Noether–Lefschetz II \mathrm{SO}^{*}(2n) n(n-1)/2 n-1 2n isogeny split III \mathrm{Sp}(2n,\mathbb{R}) n(n+1)/2 n-1 2n isogeny split IV \mathrm{SO}(n,2) n 1 (n+2)/2 Noether–Lefschetz Table 2. Special cycles on Shimura varieties As we will explain, the definitions of these cycles are qualitatively different between Types (I, IV) and Types (II, III). For the former two types, each {\mathcal{Z}}(m) is a union of sub-Shimura varieties of that same type, whereas for the latter two types, each {\mathcal{Z}}(m) is a Hecke translate of a particular sub-Shimura variety {\mathcal{Z}}(1) that comes from a product of two irreducible Hermitian symmetric domains of that same type. For each fixed g\geq 1 , one can generalize the definitions above to a class of special cycles {\mathcal{Z}}(M) of codimension gc , in Types (I, IV), indexed now by M an integral positive definite symmetric (or Hermitian) matrix. Their classes [{\mathcal{Z}}(M)] constitute the Fourier coefficients of a Siegel modular form of the same weight k from Table 2 . \Phi^{g}(q)=\sum_{M\geq 0}[{\mathcal{Z}}(M)]q^{M}\in\mathrm{CH}^{gc}(X)\otimes_{\mathbb{Q}}\mathrm{Mod}_{k}(\mathrm{Sp}_{2g}(\mathbb{Z})) The expressions [{\mathcal{Z}}(M)] when M is only semi-positive are interpreted using an appropriate power of the tautological top Chern class [{\mathcal{Z}}(0)] ; see Section 2.1 . In Section 3 , we will propose a conjectural Siegel modular series of higher codimension cycles in Type III. 1.1. Compactifications For each Shimura variety X , the canonical Satake-Baily-Borel compactification X^{*} is a projective algebraic variety whose boundary is stratified by Shimura varieties of lower dimensions. Each boundary stratum X^{\prime}\subset X^{*} corresponds to a unique \Gamma -conjugacy class of maximal parabolic subgroups {\mathbf{P}}\subset{\mathbf{G}} defined over \mathbb{Q} . The drawback of X^{*} is that it is singular in general. A nice class of smooth compactifications was constructed in [ AMRT10 ] ; these are called the toroidal compactifications X^{\Sigma} , obtained by blowing up the boundary of X^{*} . They depend on a choice of admissible fan \Sigma inside a certain cone \mathscr{C} associated to the parabolic {\mathbf{P}} . Let \epsilon:X^{\Sigma}\to X^{*} denote this blow up map. Over each stratum X^{\prime} in the boundary, \epsilon restricts to a proper surjective morphism which factors as \epsilon^{-1}(X^{\prime})\overset{\tau}{\to}\mathcal{Y}\overset{\pi}{\to}X^{\prime}, where \tau is a toric fibration, and \pi is an abelian fibration. The second goal of this note is to explore the structure of the boundary of these compactifications in the cases of Table 1 , and to describe the limits of the special cycles from Table 2 in the boundary of X^{\Sigma} . The Zariski closures \overline{{\mathcal{Z}}}(m)\subset X^{\Sigma} can be assembled into a generating series valued in the relevant cohomology or Chow group of X^{\Sigma} , but they are no longer modular forms; they are expected to be mixed mock modular forms in general. Kudla conjectured that in types (I, IV), the modularity property can be restored after subtracting a series of effective cycle classes supported in the boundary of X^{\Sigma} . Special cycles on Shimura varieties are examples of Hodge loci and it is natural to explore generating series in the more general setting of polarized variations of (mixed) Hodge structure over an arbitrary base, which we explore in

Read the complete text