The Archimedean height pairing for differential forms on degeneration of Riemann surfaces

This paper defines and analyzes the asymptotic behavior of the Archimedean height pairing for fiberwise cohomologically trivial differential forms on one-parameter degenerations of Riemann surfaces, utilizing Dai--Yoshikawa's results on small eigenvalues to extend the current-valued pairing of Filip--Tosatti to broader geometric settings.

The Gist

Here is an explanation of Junyu Cao's paper, "The Archimedean Height Pairing for Differential Forms on Degeneration of Riemann Surfaces," translated into everyday language with creative analogies.

The Big Picture: A Fading Movie and a Measuring Tape

Imagine you are watching a movie that is slowly fading out. The screen is a complex, beautiful landscape (a Riemann surface). As the movie progresses toward the end (the degeneration), the landscape starts to crumble. Trees merge, rivers dry up, and the smooth terrain turns into a jagged, broken mess (the singular fiber).

Mathematicians love to measure things on these landscapes. They have a special tool called a pairing. Think of this pairing as a "measuring tape" that tells you how two different shapes or flows (called differential forms) interact with each other.

Usually, this tape works perfectly on the smooth parts of the movie. But what happens when the movie gets to the messy, broken ending? Does the tape break? Does the measurement go to infinity?

Junyu Cao's paper answers this question. He invents a new, super-robust version of this measuring tape (the Archimedean height pairing) that works even when the landscape is falling apart. He proves that while the measurement might wiggle a bit, it doesn't explode; it behaves in a very predictable, calm way.

---

Key Concepts Explained with Analogies

1. The "Smooth" vs. The "Broken"

• The Setup: Imagine a family of donuts (smooth surfaces) that are slowly being squished. As they get squished, they turn into a figure-eight shape (a singular fiber with a pinch point).
• The Problem: If you try to measure the "height" or "energy" of a ripple on the donut as it turns into a figure-eight, the math usually gets messy. The numbers might shoot up to infinity.
• The Solution: Cao defines a specific way to measure these ripples (forms) that accounts for the squishing. He shows that if you subtract a specific "correction factor" (which looks like a logarithm, $\log |s|$), the measurement becomes smooth and continuous, even at the moment the donut breaks.

2. The "Preferred Potential" (The Best Map)

To measure the height, you need a map. But there are infinite ways to draw a map of a landscape (you can shift the sea level up or down).

• The Analogy: Imagine trying to measure the height of a mountain. You could say the base is at sea level, or at the bottom of the valley, or at the center of the Earth.
• The Innovation: Cao picks a "Preferred Potential." This is like a GPS system that automatically adjusts the sea level so that the average height of the mountain is always zero. By forcing this rule, he eliminates the confusion of "where is zero?" and gets a unique, stable measurement.

3. The "Small Eigenvalues" (The Whispering Strings)

This is the most technical part, but here is the metaphor:

• The Analogy: Imagine the donut is a guitar string. When it's smooth, it has a full range of notes (frequencies). As it gets squished into a figure-eight, some of the notes become very, very quiet (low energy), while others stay loud.
• The Discovery: Cao and his collaborators (using recent work by Dai and Yoshikawa) realized that these "whispering" notes (small eigenvalues) are the ones causing the trouble. They behave like a slow, steady hum that grows as the shape breaks.
• The Fix: Cao separates the "loud" notes from the "whispering" notes. He proves that the loud notes are easy to handle, and the whispering notes, while tricky, follow a strict rule: they grow exactly as fast as the logarithm of the time left in the movie. Once you subtract that growth, everything is calm.

4. The Application: The "Parabolic Automorphism" (The Infinite Loop)

The paper ends with a cool application involving K3 surfaces (a type of complex shape) and automorphisms (rules that move points around the shape).

• The Scenario: Imagine a robot (an automorphism) that walks around a K3 surface, repeating a pattern forever. Sometimes this robot moves in a "hyperbolic" way (stretching things apart wildly), and sometimes in a "parabolic" way (shifting things around in a loop).
• The Question: If you watch the robot walk for a very long time, what does the landscape look like? Does it settle into a smooth, continuous shape?
• The Result: Cao proves that for the "parabolic" robot, the landscape does settle into a shape with a continuous, smooth surface (a continuous potential). This is surprising because earlier mathematicians thought it might be jagged or broken.
• The Twist: However, he also proves that while the final shape is smooth, the process of getting there is not smooth. It's like watching a video where the final frame is a perfect picture, but the frames leading up to it are jittery and chaotic. This answers a famous question by mathematician Valentino Tosatti, showing that the convergence isn't as perfect as some hoped.

Why Does This Matter?

1. Stability in Chaos: It gives mathematicians a reliable way to measure things even when the geometry of the universe (or the shape they are studying) is breaking down.
2. Connecting Fields: It bridges the gap between Arakelov theory (a field dealing with number theory and heights) and complex geometry (shapes and surfaces). It's like finding a common language between two different tribes of mathematicians.
3. Solving Mysteries: It settles a debate about how these complex shapes behave under infinite repetition, proving that while the end result is smooth, the journey there is messy.

The Takeaway

Junyu Cao built a new kind of ruler. This ruler can measure the "height" of shapes even as they crumble into dust. He showed that if you know exactly how to adjust for the dust (the logarithmic correction), the measurement remains steady. This allows us to understand the behavior of complex mathematical objects in extreme situations, proving that even in a degenerating world, there is still a hidden order and continuity.

Technical Summary

Here is a detailed technical summary of the paper "The Archimedean Height Pairing for Differential Forms on Degeneration of Riemann Surfaces" by Junyu Cao.

1. Problem Statement

The paper addresses the asymptotic behavior of the Archimedean height pairing for smooth real $(1,1)$-forms on a one-parameter degeneration of Riemann surfaces (a complex surface $\pi: X \to S$ where $S$ is a disk and $X_0 = \pi^{-1}(0)$ is a singular fiber).

Specifically, given two smooth real $(1,1)$-forms $\alpha$ and $\beta$ on $X$ that are cohomologically trivial on smooth fibers (i.e., $\int_{X_s} \alpha = \int_{X_s} \beta = 0$ for $s \neq 0$), the author defines a pairing:
$$\langle \alpha, \beta \rangle(s) = \int_{X_s} \phi_s \beta$$
where $-\phi_s$ is a potential for $\alpha$ on the fiber $X_s$ (satisfying $dd^c \phi_s + \alpha|_{X_s} = 0$).

The central problem is to understand the behavior of this pairing as $s \to 0$ (approaching the singular fiber). While similar asymptotic results exist for algebraic cycles (divisors) in Arakelov theory (e.g., Holmes–de Jong) and for Quillen metrics (Yoshikawa, Eriksson–Freixas–Mourougane), this paper extends these results to differential forms and handles general singular fibers (reduced or non-reduced, reducible components), which is crucial for applications in complex dynamics.

2. Methodology

The proof relies on a sophisticated combination of spectral geometry, potential theory, and fiber integral asymptotics.

• Spectral Geometry of Degeneration: The author heavily utilizes recent work by Dai and Yoshikawa [DY25] regarding the small eigenvalues of the Laplacian on degenerating Riemann surfaces.

  • As $s \to 0$, the first $N-1$ eigenvalues (where $N$ is the number of irreducible components of $X_0$) tend to zero like $O(1/\log|s|^{-1})$.
  • The remaining eigenvalues are bounded away from zero.
  • This spectral gap allows for a decomposition of functions on fibers into low-frequency (spanned by small eigenfunctions) and high-frequency parts.

• Preferred Potentials: The author introduces "preferred potentials" $\phi_s$ for the forms $\alpha$, defined by the condition $\int_{X_s} \phi_s \omega_s = 0$. These are expanded using the eigenfunctions of the Laplacian.

  • High-frequency part: Shown to be uniformly bounded ($L^\infty$) using Cheng-Li heat kernel estimates and Sobolev inequalities.
  • Low-frequency part: Analyzed using "model functions" (approximations of characteristic functions of the thick components of the fiber) constructed by Dai-Yoshikawa. The growth of this part is controlled by the integrals of $\alpha$ over the irreducible components of the singular fiber.

• Fiber Integral Continuity: The paper establishes technical lemmas (extending Barlet's work) regarding the continuity of fiber integrals $\int_{X_s} f_s \alpha$ as $s \to 0$. This involves proving that if a sequence of functions converges locally on the regular part of the singular fiber and satisfies certain $L^p$ bounds, the integrals converge, provided the contribution near the singular locus vanishes.

• Semi-stable Reduction: For non-reduced singular fibers, the author employs semi-stable reduction techniques to relate the pairing on the original degeneration to a pairing on a model with a reduced singular fiber.

3. Key Contributions and Results

A. Asymptotics of the Archimedean Height Pairing (Theorem 0.2, 4.5)

The main result characterizes the singularity of the pairing $\langle \alpha, \beta \rangle(s)$ near $s=0$:
$$\langle \alpha, \beta \rangle(s) - c_{\alpha, \beta} \log |s|^2$$
extends continuously to $s=0$.

• The Constant $c_{\alpha, \beta}$: It is explicitly determined by the intersection matrix $M$ of the irreducible components of the singular fiber $X_0$ and the integrals of $\alpha$ and $\beta$ over these components.
  $$c_{\alpha, \beta} = v_\alpha^T M^+ v_\beta$$
  where $v_\alpha = (\int_{C_1} \alpha, \dots, \int_{C_N} \alpha)^T$, $M^+$ is the Moore-Penrose pseudoinverse of the intersection matrix, and $C_i$ are the irreducible components.
• Vanishing Condition: If $\alpha$ (or $\beta$) satisfies the stronger condition that its integral vanishes on every irreducible component of $X_0$, then $c_{\alpha, \beta} = 0$, and the pairing itself extends continuously (without the logarithmic term).

B. Continuity of Current-Valued Pairings (Proposition 0.5, 5.4)

The paper applies these results to the study of parabolic automorphisms on elliptic K3 surfaces.

• Filip and Tosatti [FT23] introduced a "current-valued pairing" $\eta_B(T, [\alpha])$ associated with an automorphism $T$. They proved the continuity of its potential under the assumption that singular fibers are reduced and irreducible.
• Cao's Contribution: By removing the irreducibility assumption and using the continuity of the Archimedean height pairing, Cao proves that the potential of the current-valued pairing is continuous for general singular fibers (Proposition 5.4).

C. Counterexample to Tosatti's Question (Corollary 0.6, 5.5)

The paper addresses a question by Tosatti [Tos25] regarding the regularity of limit currents.

• Result: For a parabolic automorphism $T$ on a K3 surface, the limit current $\lim_{n \to \infty} \frac{(T^n)^* \omega}{n^2}$ has a continuous potential.
• Significance: However, the convergence of the sequence of forms $\frac{(T^n)^* \omega}{n^2}$ to this limit cannot be in the $C^0_{loc}$ topology on the complement of the singular fibers ($X \setminus X_0$).
• This provides a counterexample to the conjecture that such convergence might be $C^0_{loc}$ when $\omega$ is a Ricci-flat metric. The proof relies on comparing the Gauss curvature of the flat metric on fibers (which is 0) with the curvature of the degenerating Kähler metric (which tends to $-\infty$).

4. Significance

• Generalization of Arakelov Theory: The work extends the classical theory of Archimedean height pairings from algebraic cycles (divisors) to general differential forms, bridging a gap between number-theoretic height pairings and complex geometric analysis.
• Dynamics on K3 Surfaces: It resolves technical obstructions in the study of parabolic dynamics on elliptic K3 surfaces, allowing for the analysis of systems with arbitrary singular fibers (including non-reduced ones). This is essential for a complete classification of automorphism dynamics.
• Regularity of Limits: The results clarify the regularity of limit currents in complex dynamics. While the limit current itself is well-behaved (continuous potential), the convergence of the approximating sequence is strictly weaker than $C^0$, highlighting subtle geometric phenomena in the degeneration of Ricci-flat metrics.
• Technical Framework: The paper provides a robust framework for analyzing fiber integrals and spectral projections in degenerating families, leveraging the Dai-Yoshikawa spectral gap estimates to control the behavior of potentials near singularities.

In summary, Junyu Cao establishes the precise logarithmic asymptotics of height pairings for differential forms, uses this to prove the continuity of potentials for current-valued pairings in general settings, and demonstrates that while these limits are continuous, their convergence is not uniform, providing a nuanced counterexample to questions in complex geometry.