Monsky–Washnitzer cohomology

A p-adic cohomology theory for smooth affine varieties in characteristic p, built from weakly completed lifts and de Rham forms.

Version

v1 · 2026-09-28 · History

Domain-specific #

10803

Domain group

Formal Sciences

Origin domain

Mathematics

Subdomain

Arithmetic Geometry → Mathematics

Aliases

Monsky-Washnitzer cohomology

Core Idea

Monsky–Washnitzer cohomology is a p-adic cohomology theory for smooth affine varieties over positive-characteristic fields. It relates a variety defined in characteristic p to characteristic-zero p-adic differential forms by choosing a suitable lift of its coordinate algebra and passing to a weak or dagger completion. The cohomology classes come from a de Rham-type complex and are designed to be independent of admissible presentation choices. The lift is part of a construction, not an assertion that the original variety literally changed characteristic.

For varieties over finite fields, Frobenius can act on these p-adic groups, enabling arithmetic uses. Kedlaya's hyperelliptic-curve work uses this action to approximate a Frobenius characteristic polynomial for point counting. That purpose is an application of the theory, not its definition. Ordinary characteristic-zero algebraic de Rham cohomology lacks the positive-characteristic source and weak p-adic completion; rigid cohomology extends related ideas to a wider geometric domain. A precise claim names the input variety, lift/completion framework, group, and any additional Frobenius assumptions.

Structural Signature

Sig role-phrases:

• Smooth affine characteristic-p source — Fixes the variety and positive-characteristic base to which the theory applies. It is constitutive. Counterfactual: A smooth complex projective variety in ordinary de Rham theory is not the same input category.
• Characteristic-zero lift — Supplies an algebra over a p-adic base reducing to the source algebra. It is constitutive. Counterfactual: Merely retaining characteristic-p polynomial forms misses the lift-based construction.
• Weak completion and differential complex — Uses overconvergent or dagger functions and de Rham forms rather than unrestricted formal completion alone. It is constitutive. Counterfactual: Changing the completion can destroy the finite-dimensional, lift-independent theory being named.
• Cohomology groups and independence — Takes differential classes with appropriate well-definedness across admissible lift choices. It is constitutive. Counterfactual: A single chosen presentation with no invariance claim is not yet the cohomology theory of the variety.
• Frobenius action and scope — In finite-field uses, relates p-adic groups to Frobenius and point-counting while keeping this an application. It is boundary. Counterfactual: Point-counting is not required to define the groups for every smooth affine source.

What It Is Not

• Not ordinary de Rham cohomology alone. The source variety has positive characteristic and a p-adic weak lift.
• Not one polynomial lift. The theory's groups must be well-defined beyond an incidental presentation.
• Not every rigid-cohomology case. Rigid theory broadens the domain and formal setting.
• Not the point-counting algorithm. Frobenius arithmetic is one use of the cohomology groups.
• Closest near-miss. Rigid cohomology is the closest miss: it generalizes related p-adic methods to broader varieties, but the domain and formal construction are not simply interchangeable by name.

Scope of Application

• Arithmetic geometry. Study smooth affine positive-characteristic varieties with p-adic invariants.
• Finite-field curves. Analyze Frobenius actions on appropriate cohomology groups.
• Point counting. Use characteristic-polynomial information in bounded curve algorithms.
• Theory comparison. Distinguish original smooth-affine MW scope from rigid extensions.

Clarity

The input must be smooth and affine in characteristic p, and the construction must use an admissible p-adic lift with weak completion and de Rham forms. Rigid cohomology is the nearest miss because it extends related p-adic methods beyond this original scope. Kedlaya's point-counting work is a use of the theory, not the definition; a bare polynomial lift without the invariant cohomology groups is incomplete.

Manages Complexity

A weakly completed lift converts finite-characteristic geometric questions into p-adic differential invariants, reducing arithmetic problems to structured group and operator questions. That compression hides delicate choices of lift, completion, and base; well-definedness ensures those choices do not become arbitrary answers. Frobenius matrices then add a second layer for point counting, whose precision and curve assumptions should not be read back into every instance of the cohomology theory.

Abstract Reasoning

1. Identify a smooth affine variety and its characteristic-p base.
2. State the admissible characteristic-zero p-adic lifting context.
3. Use the weak/dagger completion and differential complex defining the theory.
4. Take cohomology with the required independence of lift or presentation.
5. Introduce Frobenius separately when an arithmetic application needs it.

Knowledge Transfer

The same lift–weak completion–de Rham pattern moves from one smooth affine finite-field curve to another only after its coordinate algebra and admissible lift are rebuilt. Kedlaya's hyperelliptic Frobenius calculation does not automatically solve all affine varieties. Moving to singular or non-affine sources generally invokes rigid-cohomology extensions rather than pretending the original MW construction has unchanged domain.

Examples

Canonical

Take a nonsingular affine curve over a finite field of characteristic p. Choose an admissible lift of its coordinate algebra to a p-adic characteristic-zero ring, pass to a weak dagger completion, and consider differential forms modulo exact forms. The resulting cohomology belongs to the curve rather than to one accidental polynomial presentation, while any Frobenius use requires its own specified lift and precision. This conceptual case states roles, not a computational recipe.

Mapped back: Smooth affine characteristic-p source → nonsingular affine finite-field curve; Characteristic-zero lift → admissible p-adic lift of its coordinate algebra; Weak completion and differential complex → dagger algebra and de Rham forms; Cohomology groups and independence → classes independent of admissible presentation; Frobenius action and scope → optional finite-field endomorphism, not definition.

Applied / In Practice

Kedlaya's published hyperelliptic-curve point-counting work computes a p-adic approximation to a Frobenius characteristic polynomial using Monsky–Washnitzer cohomology of an appropriate smooth affine curve. It is an attested finite-field research application of the groups and their Frobenius action, not a claim that every rigid-cohomology problem or singular curve fits the same algorithm without additional choices.

Mapped back: Smooth affine characteristic-p source → smooth affine hyperelliptic-curve setting in Kedlaya's study; Characteristic-zero lift → p-adic lift used for the curve's cohomology; Weak completion and differential complex → Monsky–Washnitzer dagger/de Rham framework; Cohomology groups and independence → well-defined p-adic groups on which the study operates; Frobenius action and scope → approximated characteristic polynomial for point counting.

Structural Tensions

T1 — Finite-Characteristic Geometry versus Characteristic-Zero Calculation. Lifting enables p-adic methods but requires an invariant result rather than dependence on one lift.

Diagnostic: Which choices disappear in the cohomology group?

T2 — Weak Completion versus Overly Broad Completion. Overconvergence is not decorative; changing function space changes the analytic/cohomological behavior.

Diagnostic: What dagger or weak completion is actually used?

Structural–Framed Character

Monsky–Washnitzer cohomology is structural-leaning but domain-specific. Its evaluative weight is negligible: groups and comparison theorems are not a value judgment. The geometric source exists apart from observers, while the chosen p-adic lift and weak completion are formal mathematical constructions. Its origin in arithmetic geometry affects notation and construction but not institutional membership. Terms such as cohomology, differential form, and lift have analogues elsewhere; the exact dagger algebra over a characteristic-p affine source does not simply travel. A topological cohomology theory can be compared by analogy, not identified with MW without the p-adic construction.

The portable skeleton is an invariant assignment of algebraic objects to geometric sources, with operations that reveal selected structure; this is a future-prime candidate for general cohomology/functor reasoning, not an accepted parent asserted without exact signature proof. Its character: highly formal and structurally testable within arithmetic geometry, yet fixed by positive-characteristic and weak-lift machinery.

Structural Core vs. Domain Accent

The assignment of invariants is thinly portable; the Monsky–Washnitzer construction is not.

What is skeletal. A mathematical source is mapped to groups whose classes and transformations expose properties that are stable under admissible presentation changes. This assignment pattern appears in many cohomology theories. The broader idea is worth extracting, but saying only “assign invariants” would omit the choices and theorems that distinguish one theory from another.

What is domain-bound. The source is a smooth affine variety in characteristic p; a characteristic-zero p-adic lift is weakly completed; de Rham-type differential classes provide groups. In finite-field cases a Frobenius action can be used for arithmetic counts. Kedlaya's curve example depends on exactly this machinery. Remove the weak overconvergent completion or allow arbitrary singular sources without extension, and the named theory's contract is broken.

Why this does not clear the prime bar. Ordinary de Rham, singular, and rigid cohomology all exhibit invariant-assignment reasoning, but they differ in source categories, coefficients, constructions, and valid theorems. The general pattern may be recognized cross-domain; the specific MW recipe cannot. Point counting by a Frobenius matrix is an in-domain use, not proof that the named cohomology travels to non-arithmetic substrates. A cross-domain analogy that imports the MW name without p-adic lifts would misclassify a different theory.

Instantiates / Related Primes

This entry is a kind of Theory.

• Related — de Rham cohomology. Differential forms are used after a p-adic weak lift, not merely on a characteristic-zero variety.

• Related — rigid cohomology. The latter extends p-adic cohomological scope beyond the original smooth-affine case.

Relationships to Other Abstractions

Current abstraction Monsky–Washnitzer cohomology Domain-specific

Parents (1) — more general patterns this builds on

• Monsky–Washnitzer cohomology is a kind of Theory Prime

  Monsky–Washnitzer cohomology is a strict kind of Theory: its frozen identity entails the parent's defining structure while adding domain-specific restrictions.

Hierarchy paths (2) — routes to 2 parentless roots

• Monsky–Washnitzer cohomology → Theory → Formalization → Representation → Abstraction

Show alternative path (1)

Neighborhood in Abstraction Space

Monsky–Washnitzer cohomology sits in a sparse region of the domain-specific corpus (65^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Algebraic Varieties & Arithmetic Cohomology (7 abstractions)

Nearest neighbors

• Motive (algebraic geometry) — 0.86
• K-theory — 0.85
• Quasi-Finite Field — 0.84
• Cubic Fourfold — 0.84
• Weyl Algebra — 0.83

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With

• Rigid cohomology. Tell: Is the source within original smooth-affine MW scope or a broader extension?
• Algebraic de Rham cohomology. Tell: Is a positive-characteristic source lifted and weakly completed?
• Frobenius point-counting algorithm. Tell: Are the cohomology groups being defined or used for arithmetic?
• Arbitrary p-adic completion. Tell: Is the required weak or overconvergent completion specified?

References

• K. Kedlaya, Weil Cohomology lecture notes, Frobenius and Monsky–Washnitzer: https://kskedlaya.org/weil-cohom/chapter-14.html
• C. Davis and D. Zureick-Brown, Integral Monsky–Washnitzer cohomology and the overconvergent de Rham–Witt complex: https://arxiv.org/abs/1304.7307
• K. Kedlaya, Counting Points on Hyperelliptic Curves using Monsky–Washnitzer Cohomology: https://arxiv.org/abs/math/0105031
• Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Monsky%E2%80%93Washnitzer_cohomology (revision 1331751936).
• Preserved source candidate: http://www.numdam.org/item?id=PMIHES_1966__29__95_0
• Preserved source candidate: http://www.numdam.org/item?id=MSMF_1986_2_23__33_0