Regularized theta lifts and \((1,1)\)-currents on GSpin Shimura varieties (Q299140)

The theta lift, which is basically the integration of a theta kernel against a modular form (i.e., an automorphic form on \(\mathrm{SL}_{2}\)) and producing an automorphic form on an orthogonal group, has already been used in many branches of mathematics, proving useful for many applications. One of them is the construction of Green functions for special divisors on orthogonal Shimura varieties. The point of view on this process that lies in the context of the present paper is when \(\mathrm{SL}_{2}\) and the orthogonal group are considered as a dual reductive pair inside a larger symplectic group. Very roughly, the aim of this paper is to generalize this theory for another dual reductive pair, in which \(\mathrm{SL}_{2}=\mathrm{Sp}_{2}\) is replaced by \(\mathrm{Sp}_{4}\), and produce the corresponding Green objects. The classical theta lift requires a quadratic space over \(\mathbb{Q}\), of signature \((n,2)\) (in this normalization). We remark that the paper works in the context of a quadratic space \(V\) over a totally real field \(F\), whose signature is indeed \((n,2)\) in one real embedding of \(F\) but is positive definite in all the other embeddings. This generalization, using Whittaker forms, was defined and investigated in the \(\mathrm{SL}_{2}\) setting in [\textit{J. H. Bruinier}, J. Reine Angew. Math. 672, 177--222 (2012; Zbl 1268.11058)], and the current paper uses the same ideas concerning this generalization. The main idea here is to define the relevant theta kernel (which includes in its data a function of some level \(K\) on the points of \(V\) over the finite adéles of \(F\)), taking values on differential forms of Hodge type \((1,1)\) over the (\(n\)-dimensional complex) symmetric space \(\mathbb{D}\) associated with our vector space \(V\), and investigate the results of the (regularized) theta lift with respect to appropriate Whittaker forms. More specifically, the author shows that up to images of the differentiation maps \(\partial\) and \(\overline{\partial}\) on the corresponding Shimura variety \(X_{K}\) of level \(K\), the resulting theta lift yields the same class as the Green current of a special cycle of codimension 2 (from the theory of the Kudla program). The relation with the images of higher Chow groups under the Beilinson regulator is also studied. \smallskip The paper is divided into 4 sections, each of which consists of several subsections (called also sections in the paper), some of which are partitioned even further. We shall now describe these sections and subsections in more detail, but because of the length and the technical nature of the paper, some the results will be not described here in full detail. Subsection 1A of the Introduction includes the statement of the main results, while Subsections 1B and 1C contain an outline of the structure of the paper and some useful notation respectively. In Section 2, Subsection 2A reviews the theory of the relevant orthogonal Shimura varieties (both algebraically and analytically), including the definition of the relevant tautological line bundle, while Subsection 2B defines the special cycles that such varieties carry by the Kudla program. Section 3 is the technical heart of the paper, which includes 10 subsections, to be described in further details below. Finally, Section 4 considers the 2-dimensional example in which the Shimura variety \(X_{K}\) is the product of two Shimura curves associated with the same rational, indefinite, non-split quaternion algebra. Subsection 4A includes the relevant definitions of quaternion algebras and Shimura curves (with the relation between a GSpin group of dimension 4 and the multiplicative group of the quaternion algebra), while Subsection 4B, on its various parts, presents the explicit form that all the notions from Section 3 take in this case: The special divisors, the \((1,1)\) Green currents, and the resulting connections with the Hecke algebra. It remains to describe the notions and arguments appearing in the 10 subsections of Section 3. Subsection 3A reviews the construction, due to the reference [\textit{T. Oda} and \textit{M. Tsuzuki}, Publ. Res. Inst. Math. Sci. 39, No. 3, 451--533 (2003; Zbl 1044.11033)], of spherical functions on \(\mathbb{D}\) (depending also on a complex parameter \(s\) from an appropriate right half-plane) from hyper-geometric series, and their properties (eigenfunctions of the Laplacian, growth, etc.). In Section 3B, the argument from [Bruinier loc. cit.] through which the aforementioned functions produce, via summation on \(\Gamma\)-translates (where \(\Gamma\) is an appropriate discrete subgroup of the automorphisms of the symmetric space \(\mathbb{D}\)) and taking the constant term of the expansion in \(s\) near the central point \(\rho_{0}=\frac{n}{2}\), Green functions for special divisors on \(X_{\Gamma}=\Gamma\backslash\mathbb{D}\) is presented. The definition of the latter function depends on a totally positive vector \(v \in V\), and Subsection 3C investigates similar functions that are based on two vectors in \(V\) that span a totally positive definite plane. This subsection shows that the new functions have properties that are similar to the ones from [Oda and Tsuzuki loc. cit.] and [Bruinier loc. cit.] (e.g., local integrability of them and their derivatives), that their restriction to the special divisor associated with \(v\) (which is an orthogonal Shimura variety of dimension one less) yields Green functions for divisors on that smaller Shimura variety, and presents them as Laplace transforms of appropriate Whittaker forms. Subsection 3D uses appropriate differential \((1,1)\)-forms arising through differentiating the functions from Subsection 3C and summation over appropriate one-sided quotients of \(\Gamma\), and shows, using some technical work, that they are integrable on \(X_{\Gamma}\). The connection between the constant terms of these functions of \(s\) at \(s_{0}=\frac{n-1}{2}\) and \((1,1)\)-currents on \(X_{\Gamma}\) arising from push-forwards of Green functions on the relevant special divisors are proven in Subsection 3E: They are the same modulo images of \(\partial\) and \(\overline{\partial}\) of smooth 1-forms on \(X_{\Gamma}\). Adapting these definitions to be valid, in a coherent manner, for \(X_{K}\), whose set of complex points can be presented as the disjoint union of several \(X_{\Gamma}\)s, is the content of Subsection 3F, which also investigates the simple properties of the resulting functions and their (simple) behavior under projections from varieties of higher level. Subsection 3G considers the function associated with a Schwartz function \(\varphi\) on the finite adélic points of \(V\) and a totally positive definite symmetric \(2\times2\) matrix \(T\) over \(F\), and shows that such functions are more natural, and also allow for a good definition of inverse limits with respect to the level. In part 1 of Subsection 3H the \((1,1)\)-forms-valued theta kernel associated with \(\varphi\) from above as well as an appropriate Schwartz function on the real points of \(V\) is defined, using the Weil representation of \(\mathrm{Sp}_{4}\). Using some Whittaker functions, part 2 of that subsection investigates the properties of the theta lift of certain Whittaker forms, and shows that they coincide, on an open dense subset of \(X_{K}\), with the differential forms from Subsection 3G, thus establishing the proof of the main result. Then Subsection 3I reviews the definitions of the higher Chow group \(CH^{2}(X_{K},1)\) (using roughly divisors on \(X_{K}\) and rational functions on them, up to some conditions and divided by some relations), the Beilinson regulator map \(r_{\mathcal{D}}\) from that group to \((1,1)\)-forms on \(X_{K}\), and characterizes the image of \(r_{\mathcal{D}}\) inside those differentials forms as precisely those combinations of classes that are \(dd^{c}\)-harmonic. Finally, Subsection 3J carries out some explicit calculations, in terms of the Kudla-Millson kernel function, Whittaker forms, and other theta kernels, and deduces some relations between the \((1,1)\)-currents arising from the various theta lifts in the theory.