Geometric Theta Correspondence

• Geometric theta correspondence is a framework connecting cycles, cohomology classes, and arithmetic invariants on symmetric spaces to modular and automorphic forms via explicit theta kernels.
• The method is exemplified in modular curves and Shimura varieties where integration against theta kernels translates geometric data into Fourier coefficients and modular functions.
• Recent developments extend this correspondence to singular theta lifts and arithmetic settings, offering deeper insights into L-functions, intersection theory, and functoriality in the Langlands program.

The geometric theta correspondence is a fundamental structure in the theory of automorphic forms, arithmetic geometry, and representation theory, connecting cycles or cohomology classes on arithmetic varieties with spaces of modular or automorphic forms via explicitly constructed theta kernels. Originating from the classical theta correspondence and the Weil representation, the geometric variant interprets correspondences at the level of cycles, cohomology, and arithmetic Chow groups, and produces deep connections between geometry, representation theory, and number theory. This article surveys the main paradigms of the geometric theta correspondence, including its realization for modular curves, Hilbert modular surfaces, unitary and orthogonal Shimura varieties, and its local and global forms in the context of the geometric Langlands program.

1. Conceptual Framework

The geometric theta correspondence associates geometric or homological data from locally symmetric spaces (Shimura varieties, modular curves, etc.) to automorphic forms using theta kernels derived from the Weil representation. The general setup involves a reductive dual pair $(G,H)$ (e.g., $(\mathrm{GL}_n,\mathrm{GL}_m)$, $(\mathrm{U}(n),\mathrm{U}(m))$, or orthogonal/unitary groups) and a locally symmetric space $X$ equipped with a suitable arithmetic group $\Gamma$ acting on a symmetric domain $D$. The theta kernel is a differential form or current $\Theta(\tau,z)$, modular in $\tau$ (on some parameter space, typically the upper half-plane) and valued in forms on $X$.

Given a cycle or homology class $Z$ on $X$, the geometric theta lift evaluates as the integral or pairing

$\theta(Z)(\tau) = \int_Z \Theta(\tau,z),$

which yields a modular or automorphic form, whose Fourier coefficients encode intersection numbers, periods, or arithmetic invariants of $Z$. Conversely, pairing automorphic forms with explicit theta kernels recovers geometric or arithmetic cycles.

2. Geometric Theta Correspondence for Modular Curves

For modular curves $Y_1(N)$ with Borel–Serre compactification $Y$, the first homology decomposes as $H_1(Y;\mathbb{Z})\cong C\oplus MS_0$, where $C$ is generated by modular caps (loops at cusps) and $MS_0$ by degree-0 modular symbols $\{r,s\},\ r,s\in\mathbb{P}^1(\mathbb{Q})$ (Branchereau, 1 Feb 2026).

The geometric theta correspondence is implemented via a closed $1$-form $$E(z,\tau)\in\Omega^1(Y_1(N))\otimes M_2(\Gamma_1(N))$$, holomorphic in $\tau$, whose boundary restriction is a weight-$2$ Eisenstein series. Integration defines the theta map:

$$\Theta: H_1(Y;\mathbb{Z})\to M_2(\Gamma_1(N)),\qquad Z\mapsto \int_Z E(z,\tau).$$

Explicitly:

• Modular symbols map to products of weight-$1$ Eisenstein series.
• Caps map to weight-$2$ Eisenstein series.
• Hyperbolic cycles (associated to real quadratic fields) map to diagonal restrictions of Hilbert–Eisenstein series.

The image of $\Theta$ recovers the span of weight-$2$ Eisenstein series and products of weight-$1$ forms, leading to transparent geometric proofs of relations among Eisenstein series, notably the Borisov–Gunnells relations. The cycle-by-cycle structure provides a unified geometric explanation for previously algebraic constructions and suggests generalizations to higher-rank settings (Branchereau, 1 Feb 2026).

3. Geometric Theta Lifting for Unitary and Orthogonal Shimura Varieties

On higher-dimensional locally symmetric spaces such as unitary and orthogonal Shimura varieties, the geometric theta correspondence is realized via the Kudla–Millson–Borcherds theory (Li, 2024, Funke et al., 2011). For unitary groups, the setup considers $G=U(W)$ and $H=U(V)$, with signature dictated by the Hermitian symmetric domain $\mathcal{D}$ and Shimura varieties $X_K$.

Special cycles $Z(T,\varphi)_K$ in Chow groups are parameterized by positive semidefinite Hermitian matrices $T$ and Schwartz functions $\varphi$. The generating series

$$Z(\tau,\varphi)_K = \sum_{T\ge 0} Z(T,\varphi)_K q^T$$

is shown to be a holomorphic modular form (in cohomology, or in the arithmetic case, in the arithmetic Chow group). The geometric theta lift of a modular form $\phi$ yields a (cohomology or Chow) class by integrating the modular form against this generating series, with nonvanishing controlled by special values of $L$-functions (via the Rallis inner product and Siegel–Weil formulas) (Li, 2024).

In orthogonal settings (notably Hilbert modular surfaces), the correspondence links intersection numbers of cycles (e.g., Hirzebruch–Zagier cycles) to the Fourier coefficients of modular forms of weight $2$. Boundary corrections, such as linking numbers in the Borel–Serre compactification, are accounted for through exact forms on the boundary, ensuring the modularity of generating series even in the noncompact case (Funke et al., 2011).

4. Geometric Local Theta Correspondence and the Langlands Program

In the context of the geometric Langlands program, the geometric theta correspondence is formulated for local fields, particularly for tamely ramified or Iwahori-level structures (Farang-Hariri, 2015, Farang-Hariri, 2013). The Schwartz space is replaced by derived categories of (perverse) $\ell$-adic sheaves on infinite-dimensional spaces; equivariance under Iwahori subgroups is imposed.

Hecke functors on affine flag varieties induce commuting actions of Iwahori–Hecke algebras on these categories, and the geometric theta module becomes a bimodule for these algebras. In the case $(\mathrm{GL}_1,\mathrm{GL}_m)$, the category of $I_G\times I_H$-equivariant simple perverse sheaves is indexed by $\mathbb{Z}$, with precise descriptions of Hecke operators, matching the representation-theoretic structures at the function-theoretic level (Farang-Hariri, 2015).

The construction extends to a conjectural description for general dual pairs, connecting the $K$-theory of Springer fibers and nilpotent cones to spaces of Hecke bimodules, and it is proved in the toral case. This establishes a local model of Langlands functoriality and provides geometric realization for the Howe correspondence (Farang-Hariri, 2015, Farang-Hariri, 2013).

5. Currents, Singular Theta Lifts, and Maass Forms

The geometric theta correspondence also manifests through singular theta lifts for orthogonal groups, notably for $\mathrm{SO}(2,1)$, yielding locally harmonic Maass forms with singular sets along geodesics (Crawford et al., 2021). The singular theta lift associates to harmonic weak Maass forms a current on the modular curve, representing a distribution-valued automorphic form.

Key properties:

• The theta lift produces real-analytic, automorphic functions on the upper half-plane, with singularities prescribed by divisors of geodesic cycles.
• Wall-crossing phenomena and explicit formulas describe the discontinuity of the lift across geodesics, controlled by negative-index Fourier coefficients of the input Maass form.
• The differential structure interrelates the singular theta lift and the Shimura–Shintani correspondence, and the treatment as currents allows interpretation as Green functions for cycles.

This framework generalizes classical lifts and unifies diverse constructions (e.g., Hövel, Bringmann–Kane–Viazovska), further extending the view of theta correspondences as maps between spaces of automorphic forms and geometric currents (Crawford et al., 2021).

6. Arithmetic Extensions and the Kudla Program

The arithmetic theta correspondence extends the geometric perspective to arithmetic cycles on integral models of Shimura varieties. Here, special cycles are equipped with Green currents, forming generating series in the arithmetic Chow group. The arithmetic Siegel–Weil formula equates arithmetic intersection numbers (or heights) of cycles with the central derivatives of incoherent Eisenstein series (Li, 2024).

A central conjecture (proved in several cases) is that the generating series of arithmetic cycles is an (often nonholomorphic) modular form, with height pairings of arithmetic theta lifts governed by derivatives of automorphic $L$-functions (arithmetic Gan–Gross–Prasad conjectures). Specializations recover Gross–Zagier type formulas and $p$-adic height analogues.

This arithmetic viewpoint is deeply woven with the geometric theory, each stage accompanied by a variant of the Siegel–Weil formula and its associated inner product formula.

7. Significance, Generalizations, and Research Directions

The geometric theta correspondence synthesizes arithmetic, geometry, representation theory, and analysis. Its explicit formalism:

• Provides direct geometric interpretation of automorphic relations and $L$-values.
• Unifies algebraic and analytic constructions, offering geometric proofs of relations between modular forms (e.g., Borisov–Gunnells relations) and illuminating the cycle structure of modular and Shimura varieties (Branchereau, 1 Feb 2026, Li, 2024).
• Generalizes to higher-rank groups, both in compact and noncompact settings, with methods extending beyond $\mathrm{SL}_2$ to orthogonal, unitary, and symplectic groups, and plays a pivotal role in recent progress on the arithmetic Gan–Gross–Prasad and the arithmetic fundamental lemma (Li, 2024).
• At the local level, it informs the geometric Langlands program and the broad mathematical landscape of functoriality.

Recent advances include applications to constructions of Hodge and Tate classes, explicit arithmetic intersection formulas, $p$-adic extensions, and categorical description of the local theta correspondence. The geometric theta correspondence thus forms a cornerstone of modern research at the interface of geometry and automorphic representation theory.