Bergman kernels and Poincaré series

This paper establishes that the Bergman kernel of a finite-volume quotient of a Hermitian manifold with bounded geometry is the average of the kernel over the discrete isometry group, and applies this result to Hermitian symmetric spaces to prove the non-vanishing of a broad class of relative Poincaré series, thereby extending previous findings to general locally symmetric spaces of finite volume.

The Gist

Here is an explanation of the paper "Bergman Kernels and Poincaré Series" by Ioos, Lu, Ma, and Marinescu, translated into everyday language with creative analogies.

The Big Picture: Stitching Together a Global Puzzle

Imagine you have a giant, infinite, perfectly patterned wallpaper (this is the covering manifold, $\tilde{X}$). It's so big it goes on forever in every direction. Now, imagine you have a pair of scissors and a rule: you can cut out a specific shape and glue the edges together to make a finite, closed object, like a donut or a soccer ball (this is the quotient space, $X$).

The problem mathematicians face is this: We know exactly how the pattern looks on the infinite wallpaper. But what does the pattern look like on the finished donut?

This paper solves that problem. It proves that to figure out the pattern on the donut, you don't need to start from scratch. You just need to take the pattern from the infinite wallpaper, copy it, and stack all the copies on top of each other according to the rules of how you glued the edges.

The Key Characters

1. The Bergman Kernel (The "Perfect Blueprint"):
   Think of this as the ultimate "perfect copy" of a function or a shape that fits perfectly on your surface. On the infinite wallpaper, we have a perfect blueprint ($\tilde{P}_p$). On the donut, we want a perfect blueprint ($P_p$).

   • The Discovery: The paper proves that the blueprint for the donut is simply the sum of all the blueprints from the infinite wallpaper, shifted around by the gluing rules. It's like taking a photo of a single flower in a field and then creating a kaleidoscope image by reflecting that flower across every possible angle. The final image is the sum of all those reflections.

2. The Poincaré Series (The "Summation Machine"):
   This is the mathematical formula that does the "stacking and summing." It's a machine that takes a single point (or a small shape) and generates a complex, repeating pattern that fits the whole donut.

   • The Old Problem: Mathematicians have used these machines for a long time, but they weren't 100% sure the machine would actually produce a non-zero result (i.e., that it wouldn't just output a blank page).
   • The New Result: This paper proves that as long as you crank the machine's "weight" (a parameter $p$) high enough, the output will never be zero. It will always produce a valid, non-empty pattern.

The "Bohr-Sommerfeld" Condition: The Magic Key

To make the machine work for specific, tricky shapes (not just single points, but loops or rings), the paper introduces a concept called the Bohr-Sommerfeld condition.

• The Analogy: Imagine you are trying to tune a guitar string. If you pluck it at the wrong spot, it makes a dull thud (zero result). But if you pluck it at a "magic spot" where the physics aligns perfectly, it rings out with a clear, loud note (non-zero result).
• In the Paper: The "magic spots" are specific geometric shapes (like a closed geodesic, which is the shortest path that loops back on itself, like a great circle on a globe). The paper shows that if your shape satisfies this "tuning" condition, the Poincaré Series machine will definitely ring out a loud, clear note.

The Real-World Examples (The "Main Course")

The authors test their theory on three famous types of geometric spaces, showing that their "stacking" method works everywhere:

1. The Hyperbolic Plane ($SL_2(\mathbb{R})$):

   • Analogy: Imagine a surface that curves away from you like a Pringles chip, stretching infinitely.
   • Result: They show that if you take a loop on this surface and run the Poincaré Series machine, you get a valid pattern. This helps mathematicians understand "modular forms," which are like the DNA of number theory.

2. The Siegel Upper Half Space ($Sp_{2n}(\mathbb{R})$):

   • Analogy: A higher-dimensional version of the Pringles chip, used in complex physics and number theory.
   • Result: The same "stacking" rule works here too.

3. The Complex Ball ($SU(n, 1)$):

   • Analogy: Imagine a solid ball in a higher-dimensional space where the rules of geometry are slightly different (hyperbolic geometry).
   • Result: They prove that even for complex loops inside this ball, the machine works. This is a big deal because it extends previous results that only worked for simple, flat surfaces.

Why Does This Matter?

Before this paper, mathematicians knew how to build these patterns for simple cases or compact (finite) shapes. But many interesting shapes in nature and math are infinite but have a finite volume (like a funnel that gets infinitely thin but holds a finite amount of water).

This paper provides a universal toolkit. It says:

"If you have a shape with bounded geometry and finite volume, and you have a 'magic' shape inside it (satisfying the Bohr-Sommerfeld condition), you can build a non-zero, perfect pattern on it just by summing up the infinite copies."

The Takeaway

Think of the authors as master architects. They found a universal rule for how to build a complex, finite structure (the quotient) by simply stacking infinite copies of a blueprint (the Bergman kernel). They proved that this method never fails to produce a building, as long as you use enough "bricks" (high weight $p$) and the foundation is laid on the right "magic spots" (Bohr-Sommerfeld submanifolds).

This connects the abstract world of infinite geometry with the concrete world of finite shapes, ensuring that the patterns we see in nature and number theory are robust and real.

Technical Summary

Here is a detailed technical summary of the paper "Bergman Kernels and Poincaré Series" by Louis Ioos, Wen Lu, Xiaonan Ma, and George Marinescu.

1. Problem Statement

The paper addresses the relationship between the Bergman kernel on a complete Hermitian manifold $\tilde{X}$ and its quotient $X = \tilde{X}/\Gamma$ by a discrete group $\Gamma$ of isometries. Specifically, it investigates whether the Bergman kernel on the quotient space can be recovered by averaging the Bergman kernel of the covering space over the group $\Gamma$.

Furthermore, the authors aim to apply this relationship to the theory of automorphic forms. The goal is to prove the non-vanishing of a large class of relative Poincaré series. These series are constructed by averaging holomorphic sections (specifically "isotropic states" associated with submanifolds) over cosets of a subgroup $\Gamma_0 \subset \Gamma$. While non-vanishing results were known for compact smooth manifolds (e.g., Borthwick-Paul-Uribe, Barron), this paper extends these results to the more general setting of finite-volume locally symmetric spaces, which may be non-compact and possess orbifold singularities.

2. Methodology

The methodology combines techniques from complex geometry, spectral theory, and geometric quantization.

A. Geometric Setting

• Manifolds: $\tilde{X}$ is a complete Kähler manifold of complex dimension $n$ with bounded geometry. $\Gamma$ acts properly discontinuously and effectively.
• Line Bundles: A holomorphic Hermitian line bundle $(\tilde{L}, h_{\tilde{L}})$ is defined on $\tilde{X}$ such that its Chern curvature satisfies $\frac{\sqrt{-1}}{2\pi} R^{\tilde{L}}(\cdot, J\cdot) = g_{T\tilde{X}}$. This ensures the bundle is "uniformly positive."
• Quotient: The quotient $X = \tilde{X}/\Gamma$ is a Kähler orbifold with an induced line bundle $L$.

B. Core Technical Tool: Bergman Kernel Averaging

The authors establish a fundamental identity (Theorem 0.1 / Theorem 2.2) relating the Bergman kernel $P_p$ on the quotient to the Bergman kernel $\tilde{P}_p$ on the cover:
$$\sum_{g \in \Gamma} (g^{-1}, 1) \cdot \tilde{P}_p(g \cdot \tilde{x}, \tilde{y}) = P_p(\pi(\tilde{x}), \pi(\tilde{y}))$$

• Convergence: The proof relies on the exponential decay of the Bergman kernel $\tilde{P}_p$ away from the diagonal (established in prior work by Ma and Marinescu).
• Heat Kernel Comparison: The proof utilizes the asymptotic behavior of the heat kernel associated with the Kodaira Laplacian as the power $p \to \infty$.
• Bounded Geometry: The assumption of bounded geometry (uniform bounds on curvature and injectivity radius) is crucial for controlling the summation over $\Gamma$ and ensuring the convergence is uniform on compact sets.

C. Relative Poincaré Series and Bohr-Sommerfeld Submanifolds

To construct non-vanishing sections, the authors employ the method of relative Poincaré series:

1. Isotropic States: They consider a compact isotropic submanifold $\Lambda \subset X$ (where the pullback of the Kähler form vanishes).
2. Bohr-Sommerfeld Condition: The submanifold must satisfy the Bohr-Sommerfeld condition, meaning the pullback of the line bundle admits a non-vanishing flat section. This is a condition from geometric quantization.
3. Construction:
   • Start with an "isotropic state" $s_{\Lambda, p}$ on the quotient $X$ (constructed by integrating the Bergman kernel against a section on $\Lambda$).
   • Lift this to the cover $\tilde{X}$ to get a $\Gamma_0$-invariant section.
   • Average over the cosets $\Gamma_0 \backslash \Gamma$ to produce a $\Gamma$-invariant section (the relative Poincaré series).
4. Non-Vanishing Proof: Using the asymptotic expansion of the norm of isotropic states (Theorem 3.5), they show that for sufficiently large $p$, the $L^2$-norm of the constructed section is non-zero, implying the section itself does not vanish identically.

3. Key Contributions

1. Generalization of Bergman Kernel Averaging:
   The paper proves that the averaging formula for Bergman kernels holds for finite-volume quotients of Hermitian manifolds with bounded geometry, extending previous results that were limited to compact smooth manifolds. This applies even when the quotient is an orbifold.

2. Non-Vanishing of Relative Poincaré Series:
   The authors prove that relative Poincaré series associated with Bohr-Sommerfeld submanifolds (including closed geodesics and Lagrangian tori) do not vanish identically for sufficiently large weights $p$. This is achieved without assuming the quotient is compact.

3. Explicit Computations in Symmetric Spaces:
   The theory is applied to three major classes of Hermitian symmetric spaces, yielding explicit formulas for non-vanishing series:

   • $SL_2(\mathbb{R})^n$ (Hilbert Modular Varieties): Generalizing results on Hilbert modular forms.
   • $Sp_{2n}(\mathbb{R})$ (Siegel Upper Half Space): Generalizing results on Siegel modular forms.
   • $SU(n, 1)$ (Complex Hyperbolic Space): This is a significant extension. The paper handles cases involving:
     • Closed geodesics generated by loxodromic elements with real eigenvalues.
     • Lagrangian tori (specifically in the $n=2$ case).

4. Key Results

• Theorem 0.1 (Averaging Formula): Establishes the equality between the averaged Bergman kernel on the cover and the Bergman kernel on the finite-volume quotient for large $p$.
• Theorem 4.1 (Existence and Non-Vanishing): Proves that the relative Poincaré series constructed via averaging over $\Gamma_0 \backslash \Gamma$ yields a non-zero holomorphic section for large $p$, provided the underlying submanifold satisfies the Bohr-Sommerfeld condition.
• Theorem 0.2 & Theorem 6.5: Specific application to $SU(n, 1)$. If $\Gamma \subset SU(n, 1)$ acts on the unit ball $B_n$ and $\Gamma_0$ is generated by a loxodromic element with real eigenvalues, the relative Poincaré series associated with the closed geodesic (and its endpoints in $\mathbb{R}^n$) does not vanish for large $p$.
• Theorem 6.6: Extension to $SU(2, 1)$ involving relative Poincaré series associated with specific Lagrangian tori, proving their non-vanishing for large weights.

5. Significance

• Unification of Theory: The paper unifies the theory of Bergman kernels on covering spaces with the classical theory of automorphic forms (Poincaré series). It provides a rigorous geometric framework for constructing automorphic forms on non-compact, finite-volume spaces.
• Extension to Orbifolds and Non-Compactness: By handling orbifolds and finite-volume non-compact manifolds, the results generalize classical theorems (like those of Borthwick, Paul, Uribe, and Barron) which were restricted to compact smooth manifolds.
• Geometric Quantization Connection: The explicit use of the Bohr-Sommerfeld condition bridges the gap between symplectic geometry/quantization and the analytic theory of automorphic forms. It demonstrates that the geometric properties of submanifolds (like closed geodesics) directly dictate the existence of non-trivial automorphic forms.
• Explicit Applications: The paper provides concrete, computable formulas for non-vanishing series in the context of Hilbert modular forms, Siegel modular forms, and complex hyperbolic geometry, offering new tools for arithmetic geometry and the study of cusp forms.

In summary, this work provides a robust analytical machinery to generate and prove the non-triviality of automorphic forms on a wide class of locally symmetric spaces, leveraging the asymptotic behavior of Bergman kernels and the geometric quantization of isotropic submanifolds.