🔢 mathematics

On the images of higher signature maps

This paper investigates the image of the quadratic real cycle class map from Chow-Witt groups to the cohomology of the real locus for smooth real varieties, focusing on specific codimensions to formulate a precise conjecture regarding the exponents of its cokernel.

Original authors: Samuel Lerbet

Published 2026-03-25

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Samuel Lerbet

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

The Big Picture: Mapping Shapes to Numbers

Imagine you are an architect designing a building (this is your algebraic variety, or $X$ ). You have two different ways of looking at this building:

The Blueprint (Algebraic View): This is the mathematical formula and the list of parts used to build it. It's precise, abstract, and exists in the world of pure math.
The Physical Structure (Topological View): This is the actual building standing in the real world. It has walls, doors, and holes. It exists in physical space.

Mathematicians love to connect these two views. They want to know: "If I have a specific feature in my blueprint (like a specific wall or a hole), does it show up as a real, physical feature in the building?"

This paper is about a specific tool called the Quadratic Real Cycle Class Map. Think of this tool as a translator that tries to turn "Blueprint Features" into "Physical Features."

The Problem: The Translator is Imperfect

In the world of complex numbers (like a building made of glass and light), this translator works almost perfectly. If you have a feature in the blueprint, you can almost always find it in the building. This is known as the Hodge Conjecture.

However, in the world of real numbers (our everyday world of solid matter), the translator is glitchy.

Sometimes, a feature in the blueprint doesn't show up in the building at all.
Sometimes, the building has a feature that wasn't in the blueprint.
Sometimes, the translator gets the numbers wrong by a factor of 2, 4, or 8.

The author, Samuel Lerbet, is trying to figure out exactly how bad the glitch is. He wants to know: "How close is the translator to being perfect?"

The "Twist": Adding a Line Bundle

To make things even more interesting, the author adds a variable called a Line Bundle.

Analogy: Imagine your building has a special lighting system. In some rooms, the lights are bright (trivial bundle); in others, the lights are dim or flickering (twisted bundle).
This "twist" changes how the blueprint features translate into physical features. The author studies how this lighting system affects the accuracy of the translator.

The Main Discovery: The "Power of Two" Rule

The paper focuses on specific types of features (dimensions $c$ ) and specific building sizes (dimension $d$ ).

Lerbet proposes a Conjecture (a very strong guess based on evidence) that acts like a rule of thumb for the translator's errors.

The Rule:
If you are looking at a feature of size $c$ in a building of size $d$ , the translator might miss the mark, but it will never miss by more than a factor of $2^{(d-c)}$ .

Let's break that down with an analogy:

Imagine you are counting the number of bricks in a wall.
If the wall is huge ( $d$ is big) and you are looking at a tiny detail ( $c$ is small), the translator might be off by a lot (a large power of 2).
If you are looking at the whole wall ( $c$ is close to $d$ ), the translator is very accurate. The error is small (maybe just off by 1 or 2).

The "Exponent" Explained:
The paper talks about "exponents." In math, if a group has an exponent of 4, it means if you add any number in that group to itself 4 times, you get zero (or the "neutral" state).

Lerbet's finding: The "error" in the translation is always a multiple of a specific power of 2.
Example: If the rule says the error is bounded by $2^2$ (which is 4), it means the translator might give you a number that is 4 times too big, or 4 times too small, but it will never be 5 times too big. It's a very specific kind of "fuzziness."

What Did He Prove?

Lerbet didn't just guess; he proved this rule works for several important cases:

The Whole Building (Top Dimension): When looking at the entire building, the translator is perfect. There is no error.
The Walls (One dimension down): When looking at the walls, the translator is off by at most a factor of 2.
The Corners (Zero dimension): When looking at the corners or connected parts of the building, the error follows the rule perfectly.
Surfaces and Curves: He proved this rule works for 2D surfaces (like a sheet of paper) and 1D curves (like a line).

Why Does This Matter?

In the real world, we often have to deal with "approximations."

In Physics: We approximate the motion of planets.
In Engineering: We approximate the stress on a bridge.

This paper is about mathematical approximation. It tells us the limits of our approximation. It says, "You can't get the exact integer answer every time, but you can guarantee that your answer is 'close enough' in a very specific, predictable way."

The "Bottom Line" (The Conclusion)

Samuel Lerbet has drawn a map of the "glitches" in our mathematical translator.

The Conjecture: The translator is never too wrong. The error is always a power of 2, and that power depends on how big the building is versus how small the feature is.
The Proof: He has shown this is true for lines, flat surfaces, and the edges of 3D shapes.
The Future: He believes this rule holds true for all shapes, but proving it for every single case is like trying to solve a massive jigsaw puzzle where some pieces are still missing. He has solved the puzzle for the most common and important pieces.

In short: The paper is a guidebook for understanding how much "noise" exists when we try to translate abstract mathematical shapes into real-world physical shapes, proving that the noise is always a predictable, manageable power of two.

Technical Summary: On the Image of Higher Signature Maps

Author: Samuel Lerbet
Date: March 25, 2026
Subject: Real Algebraic Geometry, Chow-Witt Groups, Cohomology of Real Loci

1. Problem Statement

The paper addresses the fundamental problem of characterizing the image of the quadratic real cycle class map in real algebraic geometry.

Context: In complex algebraic geometry, the cycle class map $\gamma^c_{\mathbb{C}}: CH^c(X) \to H^{2c}(X(\mathbb{C}), \mathbb{Z})$ relates algebraic cycles to singular cohomology. The Hodge conjecture predicts the image of this map.
Real Setting: For a smooth variety $X$ over $\mathbb{R}$ , the real locus $X(\mathbb{R})$ is a smooth manifold that does not necessarily admit a $\mathbb{Z}$ -orientation. Consequently, a direct analog of the complex cycle class map with integer coefficients cannot be constructed via standard pushforwards.
Existing Approaches:
- The Borel–Haefliger map $\gamma^c_{BH}: CH^c(X) \to H^c(X(\mathbb{R}), \mathbb{Z}/2)$ exists but is too coarse (mod 2).
- The equivariant map $\gamma^c_G$ (Krasnov, Benoist-Wittenberg) uses equivariant cohomology of $X(\mathbb{C})$ with Galois action.
- The quadratic real cycle class map $\tilde{\gamma}^c_{\mathbb{R}}$ (Horst, Fasel, etc.) lifts the Borel–Haefliger map to integer coefficients using Chow-Witt groups $\widetilde{CH}^c(X, L)$ and cohomology with coefficients in a local system $\mathbb{Z}(L)$ .
The Core Question: What is the precise image of the map $\tilde{\gamma}^c_{\mathbb{R}}(X, L): \widetilde{CH}^c(X, L) \to H^c(X(\mathbb{R}), \mathbb{Z}(L))$ ? The naive conjecture that this map is surjective is false (e.g., for $c=0$ and disconnected curves). The paper seeks a precise description of the image, specifically the exponent of the cokernel.

2. Methodology

The author employs tools from $\mathbb{A}^1$ -homotopy theory, Milnor-Witt K-theory, and Witt ring theory to analyze the map.

Chow-Witt Groups and Rost-Schmid Complexes: The source of the map is the Chow-Witt group $\widetilde{CH}^c(X, L) = H^c(X, \mathbf{KMW}_c(L))$ . The author utilizes the Rost-Schmid complex to compute these cohomology groups and relate them to Milnor K-theory and powers of the fundamental ideal $I^n$ of the Witt ring.
Twisted Sheaves: The analysis is conducted for a line bundle $L$ on $X$ , introducing twisted sheaves $\mathbf{Z}(L)$ and $\mathbf{I}^n(L)$ .
Signature Maps: The core mechanism is the signature morphism $\text{sign}_n: \mathbf{I}^n \to \iota_* \mathbb{Z}$ (where $\iota: X(\mathbb{R}) \hookrightarrow X$ ), which relates Witt-theoretic data to integer cohomology on the real locus.
Reduction to Algebraic Problems:
- The author uses Jacobson's Theorem (Theorem 2.30), which states that the signature map $\text{sign}_\infty: \mathbf{I}^\infty(L) \to \iota_* \mathbb{Z}(L)$ is an isomorphism.
- This reduces the study of the cycle class map to the study of the transition maps between powers of the fundamental ideal, specifically the multiplication by $\llcorner -1 \lrcorner$ (which corresponds to multiplication by 2 in the Witt ring).
- The problem is reduced to analyzing the kernel and cokernel of the map $\otimes \llcorner -1 \lrcorner^{\otimes k}: \mathbf{I}^n(L) \to \mathbf{I}^{n+k}(L)$ in cohomology.
Exponent Analysis: The paper focuses on determining the exponent of the cokernel (the smallest integer $e$ such that $e \cdot \text{Coker} = 0$ ). This is achieved by constructing exact sequences involving the kernels and cokernels of the transition maps between $\mathbf{I}^n$ and $\mathbf{I}^{n+1}$ .

3. Key Contributions and Results

A. The Refined Conjecture (Conjecture 6.5)

The paper proposes a precise conjecture regarding the image of the quadratic real cycle class map:

Conjecture: For a smooth $\mathbb{R}$ -variety $X$ of dimension $d$ and a line bundle $L$ , the cokernel of the map $\tilde{\gamma}^c_{\mathbb{R}}(X, L)$ has exponent $2^{d-c}$ .
Furthermore, this exponent is optimal: for any $c < d$ , there exist examples where the cokernel does not have exponent $2^{d-c-1}$ .

This implies that the image contains the subgroup $2^{d-c} H^c(X(\mathbb{R}), \mathbb{Z}(L))$ .

B. Proven Cases (Theorem 6.6)

The author proves the conjecture for specific codimensions $c \in \{0, d-2, d-1, d\}$ :

Codimension $d$ (Zero-cycles): The map is surjective (exponent $2^0 = 1$ ). This holds generally (Theorem 3.5).
Codimension $d-1$ : The cokernel has exponent 2 (Theorem 4.4). The image is the inverse image of the algebraic mod 2 classes under reduction mod 2.
Codimension $d-2$ : The cokernel has exponent 4, provided a specific étale cohomology vanishing condition holds ( $H^{2d-1}_{\text{ét}}(X \times \mathbb{C}, \mathbb{Z}/2) = 0$ ) (Corollary 4.8).
Codimension 0 (Connected components): The cokernel has exponent $2^d$ (Proposition 5.3).

C. Quantitative Results

Surjectivity after inverting 2: The paper confirms (Corollary 6.3) that the map becomes an isomorphism after tensoring with $\mathbb{Z}[1/2]$ , contrasting with the rational Hodge conjecture which remains open.
Optimality: The author provides explicit examples (e.g., $X = \mathbb{A}^d \setminus \{0\}$ , rational surfaces with multiple real components) demonstrating that the predicted exponents $2^{d-c}$ cannot be improved in general.

4. Significance

Refinement of Real Cycle Theory: This work moves beyond the mod 2 limitations of the Borel–Haefliger map and the complexity of equivariant cohomology, providing a concrete, integral description of the image of the cycle class map in terms of explicit powers of 2.
Unification of Tools: It successfully bridges $\mathbb{A}^1$ -homotopy theory (Chow-Witt groups) with classical real algebraic geometry (signature maps, Witt rings) and étale cohomology.
New Framework for Obstructions: The paper suggests that the "defect" of surjectivity is controlled by the dimension of the variety and the codimension of the cycle, quantified by the exponent $2^{d-c}$ .
Future Directions: The author hints at a deeper connection between these exponents and the differentials in the Pardon spectral sequence (associated with the filtration of the Witt sheaf) and the Bockstein spectral sequence, suggesting a potential path to a full characterization of the image using Steenrod squares and vanishing conditions.

In summary, Lerbet establishes a robust quantitative framework for understanding the integral cycle class map in real algebraic geometry, proving that the image is "close" to being surjective (containing a large power of 2) and identifying the precise bounds of this closeness for low and high codimensions.