Semistable Reduction Theorem for Overconvergent $F$-isocrystals over Laurent Series Fields

Abstract: We prove the semistable reduction theorem for $\mathcal{E}<sup>†_K$-valued and $K$-valued overconvergent $F$-isocrystals over $k((t))$-varieties which were introduced by Lazda and Pál. As an application, we prove the finite dimensionality of $\mathcal{E}<sup>†_K$-valued rigid cohomology with compact support.

• The paper introduces a semistable reduction theorem for overconvergent F-isocrystals, demonstrating that unipotent monodromy can be achieved through descent techniques and alteration methods.
• It leverages a compactification strategy and the bounded Robba ring framework to extend classical rigid cohomology results to varieties over Laurent series fields.
• The work confirms the finite dimensionality of compactly supported rigid cohomology, laying the groundwork for advances in p-adic Hodge theory and arithmetic D-module research.

Semistable Reduction for Overconvergent $F$-isocrystals on $k((t))$-Varieties

Introduction and Motivation

The paper "Semistable Reduction Theorem for Overconvergent $F$-isocrystals over Laurent Series Fields" (2604.17799) establishes a semistable reduction theorem for both $\mathcal{E}^\dag_K$-valued and $K$-valued overconvergent $F$-isocrystals on varieties over $k((t))$, where $k$ is a field of characteristic $p>0$ and $K$ is a complete discrete valuation field of mixed characteristic with residue field $k((t))$0. These objects extend the framework for $k((t))$1-adic cohomology and arithmetic D-modules to non-perfect base fields, specifically over Laurent series fields. The paper connects the geometric theory of overconvergent isocrystals with the functional-analytic structure of the bounded Robba ring, and addresses key challenges in proving finiteness and extending previous results to this more general context.

Context and Background

The classical semistable reduction theorem, originated by Kedlaya, proved that for overconvergent $k((t))$2-isocrystals on $k((t))$3-varieties, after suitable alteration, local monodromy is unipotent and the isocrystal extends to a log structure over a smooth pair. These results, crucial for rigid cohomology and arithmetic $k((t))$4-modules, depend on the assumption that coefficients are valued in a field whose residue field is $k((t))$5. However, for $k((t))$6-varieties, geometric rigidity and monodromy cannot be captured by absolute objects over $k((t))$7; rather, the appropriate objects are $k((t))$8-modules over the bounded Robba ring $k((t))$9, as defined by Lazda and Pál [lazda2016rigid].

The main technical innovation is the passage to a compactification into proper $F$0-schemes, rather than $F$1-schemes, allowing the definition of $F$2-valued rigid cohomology that aligns with geometric expectations. This shift introduces difficulties in proving classical results, such as the finiteness of compactly supported rigid cohomology.

Semistable Reduction and Descent of Unipotence

Reduction to Classical Cases

The semistable reduction for $F$3-valued overconvergent $F$4-isocrystals is achieved by reducing to the classical case via the "descent of unipotence" principle. This asserts that, for certain weakly complete algebras $F$5, a $F$6-module $F$7 over the Robba ring $F$8 is unipotent if and only if its completion $F$9 is unipotent. This principle allows geometric arguments to transcend technical analytic difficulties; it generalizes and strengthens previous results in [kedlaya2006finiteness, lazda2016rigid], and is treated in-depth in the paper via explicit geometric descent.

Monodromy and Extension

For strictly semistable pairs $\mathcal{E}^\dag_K$0, the paper proves that an absolute overconvergent isocrystal $\mathcal{E}^\dag_K$1 extends to a log isocrystal along a component $\mathcal{E}^\dag_K$2 if and only if it has unipotent monodromy. Locally, $\mathcal{E}^\dag_K$3 is identified with an annulus, and unipotence implies unique extension to a log $\mathcal{E}^\dag_K$4-module with nilpotent residues. The descent argument shows log-extendability for the $\mathcal{E}^\dag_K$5-valued setting, as soon as it holds for the completed isocrystal, which Kedlaya's alteration theorem covers.

$\mathcal{E}^\dag_K$6-valued Case and Closed Fibers

The $\mathcal{E}^\dag_K$7-valued case requires finer control over monodromy along components of closed fibers. Alterations may introduce vertical divisors not dominating boundary components, necessitating extra care. The paper relies on Kedlaya's recent monodromy theorem [kedlaya2022monodromy], which provides a structural understanding of monodromy representations for $\mathcal{E}^\dag_K$8-adic differential equations in families. Using these tools, the paper establishes semistable reduction for $\mathcal{E}^\dag_K$9-valued overconvergent $K$0-isocrystals, ensuring that after suitable alteration and base change, unipotent monodromy is achieved along all relevant divisors, including those in the closed fiber.

Finiteness of Compactly Supported Rigid Cohomology

As an application, the paper proves the finite dimensionality of $K$1-valued rigid cohomology with compact support, extending previous results from [lazda2016rigid] and complementing Caro-Tsuzuki's overholonomicity results [caro2012overholonomicity]. The proof avoids delicate functional analytic issues (such as Hahn-Banach extension for continuous $K$2-linear maps) by leveraging semistable reduction, isocrystal extension, and comparison with log-convergent cohomology. The argument is structurally parallel to the classical setting, but relies on the new semistable reduction framework developed in this paper.

The strong comparison isomorphisms established between analytic and rigid cohomology, as well as between $K$3 and $K$4-valued versions, imply the finite-dimensionality and perfectness of Poincaré pairings for cohomology with arbitrary overconvergent $K$5-isocrystal coefficients on $K$6-schemes.

Technical Innovations

• Partial dagger algebra and spectra: The paper generalizes Lazda-Pál algebras to partial dagger algebras, establishing noetherianity and faithful flatness, which are crucial for rigidity and descent arguments.
• Adic geometry and tubular neighborhoods: The framework of adic spaces is used to interpret tubes, rigidification, and specialization maps, enabling precise control over neighborhoods and unipotence.
• Log connections and equivalences: Overconvergent log isocrystals are defined and shown to be equivalent to log $K$7-modules with nilpotent residues, both in classical and "absolute" settings over $K$8.
• Spectral and monodromy analysis: The interplay between analytic, algebraic, and cohomological data leverages spectral sequences associated to tube covers, ensuring comparison theorems for cohomology.
• Alteration and descent: A comprehensive use of alteration techniques (de Jong, Kedlaya), combined with descent of unipotence, enables extension and reduction of isocrystals beyond classical limitations.

Implications and Future Directions

The results fill a longstanding gap in the theory of $K$9-adic cohomology with coefficients, particularly over non-perfect base fields and when coefficients are valued in the bounded Robba ring. The semistable reduction theorem for $F$0-valued and $F$1-valued overconvergent $F$2-isocrystals provides a foundation for further developments in arithmetic $F$3-module theory and geometric $F$4-adic Hodge theory over Laurent series fields.

Practically, these results ensure that rigid cohomology with coefficients is finite dimensional, enabling applications to $F$5-adic representations, cohomological correspondences, and geometric class field theory in imperfect settings. The methods pave the way for extensions to arithmetic families, stacks, and potentially to noncommutative and logarithmic variants.

Theoretically, unipotence descent arguments and adic geometric tools may inspire analogous reduction techniques in other cohomology theories (e.g., $F$6-adic, motivic, crystalline), and inform the construction of Riemann-Hilbert correspondences in $F$7-adic frameworks, as indicated by Lazda's recent work [lazda2023rigidification].

Conclusion

This paper rigorously extends semistable reduction theorems to overconvergent $F$8-isocrystals with values in the bounded Robba ring over Laurent series fields. By employing advanced descent techniques and alteration theory, it overcomes analytic and arithmetic obstacles presented by imperfect residue fields and vertical divisors, and proves finiteness of compactly supported rigid cohomology. The results solidify the geometric foundations of $F$9-adic cohomology in the absolute setting, and lay the groundwork for further research in arithmetic geometry, $k((t))$0-adic Hodge theory, and arithmetic $k((t))$1-modules.