Intersections of special cycles on Shimura curves and Siegel Maass forms

1. Problem Statement

The paper addresses the problem of counting and understanding the distribution of intersection points between closed geodesics, Heegner points, and mixed configurations on compact Shimura curves.

• Context: While the distribution of single closed geodesics (by length) and Heegner points (by discriminant) is well-studied (e.g., via the Siegel-Weil formula and theta lifts), the counting of intersections of two geodesics with specific discriminants and intersection angles has been less explored.
• Specific Gap: Previous work by Rickards provided counting formulas for intersection points of geodesics with positive discriminants under specific "admissibility" conditions using local factors. However, a unified framework explaining the factorization of these counts and extending them to non-intersecting cases (common perpendiculars) or mixed cases (geodesic-Heegner point pairs) was lacking.
• Goal: To show that the generating series of these intersection counts (and related geometric quantities) are Fourier coefficients of a non-holomorphic Siegel Maass form of genus 2 and weight $1/2$.

2. Methodology

The authors employ the theta lift method, connecting automorphic forms on the Shimura curve to Siegel modular forms.

• Geometric Setup:

  • Let $B$ be an indefinite quaternion algebra over $\mathbb{Q}$ and $X$ a compact Shimura curve associated with an Eichler order.
  • Geodesics and Heegner points are identified with vectors in a rational quadratic space $V$ of signature $(2,1)$ (trace-zero elements of $B$).
  • Pairs of such objects correspond to pairs of vectors $(x_1, x_2) \in V^2$. The intersection geometry is encoded in the matrix $T = Q(x) = \begin{pmatrix} D_1 & b \\ b & D_2 \end{pmatrix}$, where $D_i$ are discriminants and $b$ encodes the angle or distance.

• Theta Lift Construction:

  • They define a Siegel theta function $\theta_L^{(2)}(\tau, z)$ of genus 2 associated with the lattice $L \subset V$.
  • They construct a theta integral $I(\tau, f) = \int_{\Gamma \backslash \mathbb{H}} f(z) \theta_L^{(2)}(\tau, z) d\mu(z)$, where $f$ is a Maass form on the Shimura curve.
  • When $f=1$ (the constant function), the integral yields a specific Siegel Maass form $G(\tau)$ of weight $1/2$.

• Fourier-Taylor Expansion:

  • A core technical innovation is the analysis of the Fourier coefficients $F_T(v)$ of the Siegel Maass form.
  • Instead of treating $F_T(v)$ as a scalar, the authors expand it in terms of characters of the compact group $SO(T)$ (or its indefinite analogue). This leads to a geodesic Taylor expansion of the input Maass form $f$ at the intersection points (or along geodesic cycles).
  • They construct explicit generalized Whittaker functions $W_{B,n}(v, s)$ that serve as the archimedean factors of these coefficients.

• Siegel-Weil Formula:

  • For the constant function $f=1$, the theta integral is identified with a Siegel Eisenstein series of weight $1/2$ via the Siegel-Weil formula.
  • This allows the computation of the non-archimedean parts of the Fourier coefficients using local Whittaker functions (following Kudla and Yang), thereby recovering Rickards' counting formulas.

3. Key Contributions and Results

A. Main Theorem: Generating Series as Siegel Maass Forms

Theorem 1.1: The generating series of the number of intersection points (and related geometric configurations) for pairs of geodesics/Heegner points with characteristic matrix $T$ are the Fourier coefficients of a Siegel Maass form $G(\tau)$ of genus 2 and weight $1/2$.

• Geometric Interpretation of Coefficients: The $T$-th Fourier coefficient $G_T(v)$ is a product of a local factor (counting the number of geometric configurations) and an explicit archimedean Whittaker function.
• Cases Covered:
  1. Intersecting Geodesics: $T$ positive definite. The count is the number of intersection points with angle $\theta$ determined by $b/\sqrt{D_1 D_2}$.
  2. Non-intersecting Geodesics: $T$ indefinite ($D_1, D_2 > 0$). The count is the number of pairs with a common perpendicular of length $\ell$ determined by $\cosh(\ell) = b/\sqrt{D_1 D_2}$.
  3. Geodesic-Heegner Pair: $T$ indefinite ($D_1 > 0 > D_2$). The count involves the distance from a Heegner point to a geodesic.
  4. Heegner-Heegner Pair: $T$ negative definite. The count involves the distance between two Heegner points.

B. Generalized Whittaker Functions and Taylor Expansions

Theorem 1.2 & 3.14: For a general Maass form $f$ (not just constant), the Fourier coefficients of the theta lift admit a Fourier-Taylor expansion:
$$IT_n(v, f) = W_{B,n}(v, s) \cdot \sum_{x \in R_T} f_{-2n}(z_x)$$

• Here, $f_{-2n}(z_x)$ are the geodesic Taylor coefficients of $f$ at the intersection point $z_x$.
• $W_{B,n}(v, s)$ are explicit generalized Whittaker functions constructed via orbital integrals.
• This result generalizes the work of Katok-Sarnak (which dealt with averages over Heegner points) to averages over intersection cycles and geodesic Taylor coefficients.

C. Recovery and Generalization of Rickards' Results

Theorem 5.2 & 2.6: By applying the Siegel-Weil formula to the constant function case, the authors derive an explicit formula for the number of intersection points $|R_T|$ as a product of local factors.

• This recovers Rickards' formula for the modular curve and generalizes it to arbitrary Shimura curves and level structures.
• The formula explicitly handles the "admissibility" conditions and provides a unified explanation for the local factors via non-archimedean Whittaker functions.

D. Hecke Action

Proposition 4.4: The authors establish the compatibility of the theta lift with Hecke operators. If $f$ is a Hecke eigenform on the Shimura curve, the theta lift $I(\tau, f)$ is a Hecke eigenform for the genus 2 metaplectic group. They provide explicit relations between the eigenvalues, extending the Rallis-Zhuravlev method to the metaplectic case.

4. Significance

1. Unification of Counting Problems: The paper provides a single automorphic object (the Siegel Maass form $G$) that encodes the counting of all types of special cycle intersections (geodesic-geodesic, geodesic-point, point-point) on Shimura curves, regardless of whether they intersect or have a common perpendicular.
2. Geometric Interpretation of Fourier Coefficients: It establishes a deep link between the Fourier coefficients of Siegel Maass forms and the geodesic Taylor expansion of Maass forms on the underlying curve. This offers a new perspective on the spectral theory of Shimura curves.
3. Generalization of Rickards: It removes restrictive "admissibility" assumptions from previous counting results and extends them to general levels and indefinite discriminants, providing a complete arithmetic description of the intersection numbers.
4. Bridge to Arithmetic Geometry: The authors draw parallels between their results and the Gross-Zagier formula. Just as Gross-Zagier relates derivatives of Eisenstein series to heights of Heegner points, this work relates the values of a Siegel Eisenstein series (via the Siegel-Weil formula) to the intersection numbers of geodesic cycles, suggesting a potential real-quadratic analogue of singular moduli theory.
5. Technical Advancement: The construction of generalized Whittaker functions for indefinite matrices and the detailed analysis of the theta lift for non-constant Maass forms represent significant technical progress in the theory of automorphic forms on metaplectic groups.

In summary, this paper successfully lifts the problem of counting geometric intersections on Shimura curves into the realm of Siegel modular forms, providing explicit formulas, generalizations of known results, and a rich structural framework connecting local arithmetic data with global automorphic forms.