🔢 mathematics

Modularity of special cycles on Shimura varieties: a survey

This paper surveys recent progress on Kudla's conjecture concerning the modularity of generating series for special cycle classes in orthogonal and unitary Shimura varieties, while also proposing new conjectures for other types of Shimura varieties and period domain quotients.

Original authors: François Greer, Salim Tayou

Published 2026-03-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: François Greer, Salim Tayou

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a master architect trying to build a perfect, infinite city. In the world of mathematics, this city is called a Shimura Variety. It's a complex, multi-dimensional space that acts as a map for understanding deep relationships between numbers, geometry, and symmetry.

Now, imagine you want to place special landmarks (called Special Cycles) throughout this city. These landmarks aren't just random; they represent specific solutions to mathematical puzzles, like finding points where a curve has a certain shape or where a number behaves in a special way.

The big question this paper asks is: If we collect all these landmarks and arrange them in a specific order, do they form a perfect, rhythmic pattern?

In mathematics, a "perfect rhythmic pattern" is called a Modular Form. Think of a modular form like a song that repeats itself in a very specific, beautiful way no matter how you shift the time or the pitch. If your landmarks form a modular form, it means the universe of these numbers is singing a harmonious song.

Here is a breakdown of the paper's journey using everyday analogies:

1. The City and the Landmarks (Shimura Varieties & Cycles)

The authors look at four different "neighborhoods" (Types I, II, III, and IV) in this mathematical city.

The Landmarks: In some neighborhoods, the landmarks are like distinct islands (Type I and IV). In others, they are like bridges connecting different parts of the city (Type II and III).
The Song: Mathematicians have already proven that in the "island" neighborhoods, if you list the landmarks by size, they create a perfect song (a modular form). This is like knowing that if you count the stars in a specific constellation, they follow a perfect rhythm.

2. The Problem: The City Has No Walls (Compactification)

The problem is that these mathematical cities are infinite. They stretch out forever. You can't really write a song about an infinite city without hitting a wall.

The Fix: Mathematicians build "fences" around the city to make it finite. These are called Compactifications.
The Issue: When you build a fence, the landmarks near the edge get messy. They might spill over the fence, or the fence might cut them in half. When you try to write your song including these messy edge-landmarks, the rhythm breaks. The song stops being a perfect modular form; it becomes a "jittery" or "mock" song.

3. The Solution: The "Correction" (Kudla's Conjecture)

The paper focuses on a brilliant idea proposed by a mathematician named Kudla. He suggested that even though the landmarks near the fence are messy, we can fix the song.

The Analogy: Imagine you are recording a song, but there's a loud construction noise near the microphone (the fence). Kudla says, "If we subtract a specific recording of that construction noise, the original song will come back clear."
The Paper's Goal: The authors are surveying recent proofs that show this "noise subtraction" works for the "island" neighborhoods. They are also proposing that this same trick should work for the "bridge" neighborhoods (Types II and III), even though no one has fully proven it yet. They are essentially saying, "We think the song can be saved here too, if we just add the right corrections."

4. The "Virtual" Landmarks (Excess Intersections)

In some neighborhoods, the landmarks overlap in confusing ways (like two bridges crossing at the same spot).

The Metaphor: Imagine trying to count people in a crowd where some people are standing on top of each other. If you just count heads, you get the wrong number.
The Fix: The authors suggest using "Virtual Landmarks." Instead of counting the messy overlaps directly, they assign a "ghost weight" to them. It's like saying, "This overlapping spot counts as 1.5 people." By doing this math trick, the messy overlaps turn back into a perfect song.

5. The "Universal" City (PEL Varieties)

The authors then zoom out. They realize that all these specific neighborhoods are actually just special cases of a giant, universal city called a PEL Shimura Variety.

The Big Picture: They propose a "Master Conjecture." If the "noise subtraction" trick works for the specific neighborhoods, it should work for the entire universal city. They are trying to write a single rulebook that explains how to fix the song for any type of mathematical city, not just the ones we already know.

6. The "Twin Cities" (Isogenies)

Finally, the paper looks at two cities that look different on the outside but are actually the same underneath (like a city built on a hill vs. a city built in a valley, but with the same population).

The Mystery: In one city, the landmarks form a simple song. In the other, they form a complex, multi-part song.
The Connection: The authors wonder if there is a "translator" (a mathematical lift) that can turn the simple song into the complex one. They suspect that these different songs are actually just different versions of the same melody, played in different keys.

Summary

In simple terms, this paper is a survey of a mathematical detective story:

The Mystery: Do special geometric shapes in complex spaces form a perfect, repeating pattern (a modular form)?
The Obstacle: The spaces are infinite, and the edges mess up the pattern.
The Theory: If we subtract the "noise" from the edges, the pattern returns.
The Progress: We have proven this for some shapes; we are guessing it works for others.
The Dream: To find a single, universal rule that explains how to fix the pattern for all these mathematical shapes, revealing a hidden harmony in the universe of numbers.

The authors are essentially the conductors of this orchestra, checking the sheet music to see if the rhythm holds up, and suggesting how to fix the notes that are currently out of tune.

1. Problem Statement

The paper addresses the modularity of generating series of special cycle classes on Shimura varieties and their compactifications.

Core Question: Do the generating series of special cycles (algebraic cycles defined by geometric conditions, such as Noether-Lefschetz loci or isogeny factors) form modular forms when valued in cohomology or Chow groups?
Context: While the modularity of these series in the interior of orthogonal (Type IV) and unitary (Type I) Shimura varieties was established by Kudla and Millson, the behavior of these cycles in compactifications (specifically toroidal compactifications) is more complex.
The Challenge: When extending special cycles to the boundary of a compactification, the naive generating series often fails to be modular. It is expected to be a "mixed mock modular form." The central problem is to determine if one can correct these series by subtracting specific boundary cycle classes to restore modularity.
Scope: The paper surveys known results for Types I and IV, proposes conjectures for Types II and III (PEL Shimura varieties), and explores generalizations to non-classical period domains and exceptional isogenies between different Shimura types.

2. Methodology

The authors employ a combination of algebraic geometry, automorphic forms, and Hodge theory:

Classification of Shimura Varieties: The paper focuses on the four infinite families of Hermitian symmetric domains (Types I–IV):
- Type I: Unitary ($SU(a,b)$)
- Type II: Quaternionic ( $SO^*(2n)$ )
- Type III: Symplectic ( $Sp(2n, \mathbb{R})$ )
- Type IV: Orthogonal ($SO(n,2)$)
Cycle Construction: Special cycles $Z(M)$ $Z (M)$ are constructed as loci where specific geometric conditions hold (e.g., existence of maps from elliptic curves, or orthogonality to lattice vectors).
- For Types I and IV, cycles are unions of sub-Shimura varieties.
- For Types II and III, cycles are Hecke translates of products of lower-dimensional domains.
Compactification Analysis: The authors analyze the Satake-Baily-Borel (SBB) compactification ( $X^*$ ) and toroidal compactifications ( $X_\Sigma$ ). They study the stratification of the boundary and the behavior of cycle limits as they approach the boundary divisors.
Correction Mechanism: The methodology involves defining "corrected" cycles $\tilde{Z}(M) = Z(M) + \sum a_i(M)Y_i$ , where $Y_i$ are boundary cycles. The goal is to find coefficients $a_i(M)$ such that the generating series becomes a holomorphic modular form.
Virtual Cycles: For cases involving excess intersections (where cycles intersect improperly), the authors introduce "virtual" cycles using excess intersection terms (products of tautological Chern classes) to formulate precise modularity conjectures.

3. Key Contributions and Results

A. Orthogonal and Unitary Cases (Types IV and I)

Orthogonal (Type IV): The paper reviews the Kudla-Millson theorem, which establishes modularity in cohomology.
- Conjecture 2.2 (Kudla): Proposes that for toroidal compactifications, subtracting boundary cycles restores modularity.
- Recent Progress: The authors highlight that this conjecture has been proven for divisors (codimension 1) in [EGT23, Gar23] and for zero-cycles in [BRZ24]. The corrected series involves mock modular forms (e.g., Zagier's $F^+_N$ ) and Eisenstein series.
Unitary (Type I):
- Conjecture 2.4: Analogous to the orthogonal case, proposing modularity for corrected cycles in toroidal compactifications.
- Progress: Proven for specific cases ( $g=b=1$ and $g=a, b=1$ ). The authors' joint work [GT25] provides a proof for the cohomological version when $b=1$ , utilizing the Hard Lefschetz theorem and harmonic polynomials on the boundary.

B. PEL Shimura Varieties (Types II and III)

The paper formulates new conjectures for Siegel modular varieties (Type III) and Quaternionic Shimura varieties (Type II).

Siegel Modular Varieties (Type III):
- Special cycles are defined via isogenies from products of elliptic curves (or abelian varieties).
- Conjecture 3.1: The generating series of special divisors is a modular form of weight $2n$ .
- Virtual Series (Conjecture 3.2 & 3.3): Due to improper intersections, the authors propose "virtual" generating series involving excess intersection terms (Chern classes $\lambda_i$ ). They conjecture these virtual series are Siegel modular forms.
Quaternionic Shimura Varieties (Type II):
- Cycles are defined via orthogonal splittings of quaternionic skew-Hermitian spaces.
- Question 3.5: Poses the existence of toroidal corrections for these varieties, analogous to Kudla's conjecture.

C. General PEL Framework

Conjecture 3.8: A unified conjecture for general PEL Shimura varieties. It suggests that for any PEL datum, there exists a smooth compactification and boundary corrections such that the generating series of special cycles is a modular form valued in the Chow group.
Heuristic: The modularity is expected to arise from pulling back Kudla-Millson series from a larger period domain, though improper intersections complicate the direct pushforward.

D. Non-Classical Period Domains

The authors extend the discussion to variations of Hodge structure (VHS) over arbitrary bases $S$ that are not necessarily Shimura varieties.
Question 5.1 & 5.2: They ask if Noether-Lefschetz generating series on the base $S$ are modular forms in the Chow group, and if corrections exist for projective compactifications of $S$ .

E. Isogenies and Lifting Relations

The paper investigates "curious coincidences" where different Shimura varieties (from different families) share the same underlying domain due to exceptional isogenies (e.g., $SU(2,2) \cong SO(4,2)$ ).
Conjecture 6.1: Proposes that the generating series for these isogenous pairs are related by classical lifting maps between spaces of modular forms:
- Shimura Lift: Relates half-integral weight to integral weight.
- Saito-Kurokawa Lift: Relates $SL_2$ forms to $Sp_4$ forms.
- Ikeda Lift: Relates forms of different ranks.
This suggests a deep structural link between the modularity of special cycles across different geometric realizations.

4. Significance

Unification: The paper provides a unified framework for understanding special cycles across all four infinite families of Shimura varieties, distinguishing between those defined by sub-varieties (Types I, IV) and Hecke translates (Types II, III).
Advancement of Kudla's Program: It moves the field beyond the classical Kudla-Millson theorem (interior cohomology) to the modern frontier of arithmetic intersection theory on compactifications. The resolution of Kudla's conjecture for divisors is a major milestone highlighted here.
New Conjectures: By formulating precise conjectures for Types II and III (including virtual cycles to handle excess intersections), the paper sets the agenda for future research in arithmetic geometry.
Bridge to Arithmetic Chow Groups: While the paper focuses on rational coefficients, it emphasizes that these results have integral analogues, connecting the geometry of special cycles to arithmetic Chow groups and the arithmetic of Shimura varieties.
Interplay of Geometry and Automorphic Forms: The discussion on isogenies and lifting relations suggests that the modularity of geometric cycles is not isolated but part of a larger web of automorphic phenomena, potentially linking different families of Shimura varieties through lifting operators.

In summary, this survey consolidates recent breakthroughs in the modularity of special cycles, identifies the remaining gaps (particularly for higher codimensions in Types II/III and general PEL cases), and proposes a comprehensive conjectural framework that connects algebraic geometry, Hodge theory, and the theory of automorphic forms.