🔢 mathematics

Poincaré duality in logarithmic motivic homotopy theory

This paper establishes Poincaré duality for smooth projective morphisms in logarithmic motivic homotopy theory by adapting arguments from Annala-Hoyois-Iwasa, and uses this result to demonstrate that the crystalline cohomology of a log compactification is independent of the choice of compactification.

Original authors: Doosung Park

Published 2026-02-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Doosung Park

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 an architect trying to understand the shape and structure of a building. In the world of mathematics, specifically a field called Motivic Homotopy Theory, mathematicians build "blueprints" (called motives) for geometric shapes (schemes) to study their properties using tools from topology and algebra.

For a long time, these blueprints worked great for "perfect" buildings—smooth, closed shapes with no edges or boundaries. But real-world geometry often involves shapes with edges, corners, or boundaries (like a disk with a rim, or a manifold with a boundary).

This paper, written by Doosung Park, is about upgrading the mathematical toolkit to handle these "edgy" shapes using a concept called Logarithmic Geometry. Here is the story of what the paper achieves, explained simply.

1. The Problem: The "Perfect" vs. The "Real"

In standard math, there is a famous rule called Poincaré Duality. Think of it like a perfect mirror. If you have a smooth, closed shape (like a sphere), the rule says: "What you see on the inside is perfectly reflected on the outside." Mathematically, this allows you to calculate things about the whole shape by looking at just one part of it.

However, this mirror breaks when you have a shape with a boundary (like a flat disk with a rim). The standard rules of the "perfect" world don't apply to the "edgy" world.

2. The Solution: Logarithmic Geometry (The "Boundary" Toolkit)

To fix this, mathematicians use Logarithmic Geometry. You can think of this as adding a special "log" or "label" to the edges of your shapes.

Standard Geometry: A disk is just a disk.
Log Geometry: A disk is a disk plus a label on its rim saying, "I am a boundary."

This allows mathematicians to treat shapes with boundaries as if they were "smooth" in a new, expanded sense. The paper uses a specific framework called Logarithmic Motivic Homotopy Theory (let's call it LogSH).

3. The Main Discovery: The New Mirror (Poincaré Duality)

The author's biggest achievement is proving that the Mirror Rule (Poincaré Duality) works even in this new Log world.

The Old Way: You needed a very specific, rigid set of rules (called $\mathbb{A}^1$ -invariance) to make the mirror work. This excluded many interesting shapes and theories.
The New Way: Park shows that by using Log Gysin Morphisms (a fancy term for a "boundary-aware translation tool"), the mirror works for any smooth shape with a boundary.

The Analogy: Imagine you have a translator who can speak "Standard Math" and "Boundary Math." Park proved that this translator is perfect. If you give them a shape with a boundary, they can translate its properties back and forth perfectly, just like the mirror does for perfect spheres.

4. Why Does This Matter? (The Crystal Ball)

The paper doesn't just prove a rule; it uses this rule to solve a practical problem involving Crystalline Cohomology.

The Problem: Imagine you have a broken vase (a geometric shape with singularities or bad spots). To study it, you try to "patch it up" by gluing it into a bigger, perfect vase (a log compactification).
The Issue: There are many different ways to patch the vase. Does the result depend on how you patched it? If you patch it one way, do you get a different "crystal" (mathematical data) than if you patch it another way?
The Result: Using the new Mirror Rule, Park proves that it doesn't matter how you patch it. The resulting "crystal" is always the same.

The Metaphor: Think of the "crystal" as the true essence of the broken vase. Park proved that no matter which "patch kit" (log compactification) you use to fix the vase, the essence you extract remains identical. This gives mathematicians a reliable, universal way to study broken or complex shapes.

5. The "Gotcha": Not Everything is Perfect

The paper also points out a limitation. In the old "perfect" world, there was a rule called the Localization Property (if you know the whole and one part, you know the other part).
Park shows that in this new "Log" world, this rule fails for certain types of boundaries.

Analogy: In the perfect world, if you know the whole cake and the frosting, you know the sponge. In the Log world, if you know the cake and the "boundary label," you might not be able to perfectly deduce the sponge if the boundary is tricky. This is actually a good thing! It means the new toolkit is sensitive enough to detect subtle differences that the old, blunt tools missed.

Summary

The Goal: Create a universal mirror rule for geometric shapes that have edges or boundaries.
The Method: Use "Logarithmic" labels to treat edges as smooth, and use a special "translation tool" (Log Gysin morphisms).
The Win: Proved the mirror rule works, allowing mathematicians to extract consistent, reliable data (Crystalline Cohomology) from complex shapes, regardless of how they are "patched up."
The Takeaway: We now have a more robust, flexible mathematical framework that can handle the messy, "edgy" reality of geometry, not just the perfect, idealized versions.

1. Problem Statement and Context

The Core Problem:
In classical motivic homotopy theory (specifically the $\mathbb{A}^1$ -homotopy category $SH$), Poincaré duality is a fundamental result stating that for a smooth projective morphism $f: X \to S$ , the motive $M_S(X)$ is dualizable with a specific dual involving the Thom space of the relative cotangent bundle. However, the category $SH$ is $\mathbb{A}^1$ -invariant, meaning it cannot represent cohomology theories that are not invariant under the affine line (e.g., Hodge cohomology, topological cyclic homology, prismatic cohomology).

The Logarithmic Setting:
Logarithmic motivic homotopy theory ( $\logSH$ ) was developed to handle cohomology theories that are not $\mathbb{A}^1$ -invariant by incorporating log structures (viewing schemes as "schemes with boundary"). While Poincaré duality was recently established for a specific $\infty$ -category of motivic spectra ($MS$) by Annala, Hoyois, and Iwasa, their proof relied on the existence of "Tang's Gysin morphisms" with specific naturality properties, which were not yet available at the time of their writing.

The Gap:
There was no established Poincaré duality theorem for the logarithmic motivic homotopy category $\logSH$ that could accommodate non- $\mathbb{A}^1$ -invariant theories. Furthermore, the independence of crystalline cohomology from the choice of log compactification (a crucial property for arithmetic geometry) needed to be proven within this robust framework without assuming resolution of singularities.

2. Methodology and Framework

The author works within a general axiomatic setting for a sheaf of presentably symmetric monoidal stable $\infty$ -categories, denoted as $\mathcal{T}$ , defined on the category of fine and saturated (fs) log schemes ( $\mathbf{lSch}$ ). This framework includes $\logSH_{sat}$ (the saturated version of logarithmic motivic homotopy theory).

Key Technical Tools:

Log Gysin Morphisms: Unlike the $MS$ setting, $\logSH$ possesses log Gysin morphisms (constructed by Binda, Park, and Østvær). These are essential for defining the duality maps.
Log Cyclotomic Trace: The paper utilizes the log cyclotomic trace (constructed in [21, 22]) to relate K-theory to crystalline cohomology in the log setting.
Deformation to the Normal Bundle: The author adapts the deformation space technique ( $D_{ZX}$ ) to the log setting, proving that the log Gysin sequence holds for strict closed immersions in log smooth schemes.
Naturality Issues: A major technical hurdle is that the log Gysin morphisms do not satisfy the same strong naturality conditions as Tang's Gysin morphisms in the $MS$ setting. The author circumvents this by proving a "weaker" naturality (Theorem 5.5) sufficient for establishing duality, following an argument style similar to Déglise's work in the classical setting.

3. Key Contributions and Results

The paper establishes three main theorems (A, B, and C):

Theorem A: Poincaré Duality for Log Smooth Projective Morphisms

Statement: For a projective morphism of schemes $f: X \to S$ , the functor $f^\sharp: \logSH(X) \to \logSH(S)$ is equivalent to $f_* \Sigma^{\Omega^1_f}$ , where $\Sigma^{\Omega^1_f}$ is the twisted suspension by the relative cotangent bundle.
Significance: This proves that motives of smooth projective log schemes are dualizable. Crucially, this duality applies to non- $\mathbb{A}^1$ -invariant cohomology theories (e.g., Hodge, prismatic, syntomic) that admit suitable logarithmic extensions.
Methodology: The proof involves constructing an evaluation morphism and a comparison morphism, showing they are isomorphisms. It relies heavily on the compatibility of log Gysin morphisms with compositions (Theorem 5.5) and the behavior of Thom spaces under log blow-ups.

Theorem B: Motivic Lefschetz Duality

Statement: For a closed immersion $Z \hookrightarrow X$ of smooth projective schemes, the motive $\Sigma^\infty X/Z$ (representing the pair) is dualizable. Its dual is the twisted inverse Thom space of the log cotangent bundle of the associated log scheme $(X, Z)$ .
Significance: This is the motivic analogue of Lefschetz duality for manifolds with boundary. It treats the log scheme $(X, Z)$ as a "smooth scheme with boundary."
Application: This result is a critical ingredient for Theorem 9.3, which establishes that Poincaré duality is preserved under smooth blow-ups (if it holds for three corners of a blow-up square, it holds for the fourth).

Theorem C: Independence of Crystalline Cohomology

Statement: Let $k$ be a perfect field of characteristic $p > 0$ . There exists a $cdh$ sheaf of complexes $R\Gamma^{lc}_{crys}$ on the category of schemes over $k$ such that for any projective log smooth fs log scheme $X$ :
$R\Gamma^{lc}_{crys}(X - \partial X) \simeq R\Gamma_{crys}(X)$
Significance: This proves that the crystalline cohomology of the open part of a log scheme is independent of the choice of log compactification.
Methodology: The author adapts the argument of Annala-Hoyois-Iwasa. They show that the crystalline cohomology spectrum is a module over the $\mathbb{A}^1$ -localized sphere spectrum. This is achieved using the log cyclotomic trace and the log K-theory, avoiding the need for resolution of singularities or weak factorization assumptions required by previous works (Ertl-Shiho-Sprang, Merici).

4. Additional Technical Results

Failure of Localization: The paper demonstrates that the localization property (where $j_! j^* \to \text{id} \to i_* i^*$ is a fiber sequence) fails for closed non-open immersions in $\logSH$ . This is a direct consequence of the existence of non- $\mathbb{A}^1$ -invariant theories (like Hodge cohomology) which are not nil-invariant.
Blow-up Squares: Theorem 9.3 establishes a "four corners" principle for Poincaré duality in smooth blow-up squares, allowing the extension of duality results from projective morphisms to proper smooth morphisms via resolution techniques (if available) or specific geometric constructions.

5. Significance and Impact

Unification of Cohomology Theories: By establishing Poincaré duality in $\logSH$ , the paper provides a unified framework where diverse cohomology theories (classical, Hodge, p-adic, prismatic) satisfy duality theorems simultaneously.
Arithmetic Geometry: The independence of crystalline cohomology from log compactification (Theorem C) is a major result for arithmetic geometry, particularly in characteristic $p$ . It provides a robust definition of cohomology for singular varieties via log compactifications without relying on strong resolution of singularities conjectures.
Technical Advancement: The paper resolves the technical gap regarding the naturality of Gysin morphisms in the log setting, proving that duality can be established even without the strong naturality available in the $MS$ setting.
Foundational for Future Work: The results lay the groundwork for further developments in logarithmic motivic homotopy theory, including the study of duality for non-proper morphisms and the application of these duality theorems to the study of motives of singular varieties.

In summary, Doosung Park successfully extends the powerful machinery of Poincaré duality to the logarithmic setting, enabling the study of a broader class of cohomology theories and solving a long-standing problem regarding the invariance of crystalline cohomology under log compactification.