Recovering p-adic points from fundamental-group sections

← All digests

Family 019 · Number theory
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-10

1. The question

Can the rational points of an algebraic curve be recognized purely from its groups of covering symmetries? Grothendieck’s section conjecture proposes a striking answer: a point should be equivalent to a certain splitting of the curve’s arithmetic fundamental group.

The distinction between a point and a splitting is substantial. A point is a solution of polynomial equations. A splitting is a continuous homomorphism between infinite profinite groups. The conjecture says that, for suitable curves and fields, the group-theoretic object cannot exist without a geometric point producing it.

Family 019 claims the local version over every finite extension of $\mathbb Q_p$. It also contains a geometric theorem constructing one étale cover with simultaneously prescribed behavior over several disks and their common exterior. The section manuscript gives global consequences under an additional finite-descent hypothesis, including stated modular-curve cases over $\mathbb Q$.

2. Background and essential definitions

Covering spaces in algebraic geometry

A finite étale cover $Y\to X$ is a finite algebraic map with no ramification. It is the algebraic counterpart of a finite covering space. The étale fundamental group $\pi_1(X)$ records how the sheets of every such cover can be permuted compatibly.

Unlike the usual fundamental group from elementary topology, it is profinite: an inverse limit of finite groups. Its topology is part of the data, so homomorphisms in the section conjecture must be continuous.

Let $k$ be a finite extension of $\mathbb Q_p$, with algebraic closure $\bar k$. Its absolute Galois group is $G_k=\operatorname{Gal}(\bar k/k)$. For a smooth proper geometrically connected curve $X/k$, the fundamental group fits into

$$1\longrightarrow\pi_1(X_{\bar k}) \longrightarrow\pi_1(X) \overset{r}{\longrightarrow}G_k \longrightarrow1.$$

The kernel records covers after the constants are algebraically closed. The quotient records arithmetic symmetry of the constants.

A section is a continuous group homomorphism $s:G_k\to\pi_1(X)$ with $r\circ s=\operatorname{id}$. A rational point $a\in X(k)$ supplies a section. Choices of paths or base-point identifications change it by conjugation inside $\pi_1(X_{\bar k})$, so the intrinsic object is a conjugacy class.

Here proper curves have no punctures, and “hyperbolic” means genus at least two. The genus measures the curve’s geometric complexity; this restriction is central to the conjecture’s stated form.

Valuations, disks and birational information

A valuation measures sizes of field elements. A decomposition group consists of Galois automorphisms preserving a chosen extension of a valuation. It can therefore locate group-theoretic information at a point or at a more general valued position.

The function field $K=k(X)$ contains rational functions on the curve. Its absolute Galois group records all finite separable extensions, including covers ramified at points. The étale group records only unramified covers of the whole curve. Consequently, lifting an étale section to a birational section of the function-field Galois group is an additional task.

For geometric constructions the papers use Berkovich analytic curves over $\Omega=\widehat{\bar k}$, the completed algebraic closure. These spaces include generalized points that make nonarchimedean curves connected and allow finite graph-like skeleta to describe their degeneration. An open disk has a single attachment point to its exterior; annuli become edges of a skeleton. This geometry makes it possible to cut, glue and control covers analytically before recovering algebraic covers.

3. Development history and previous partial results

Grothendieck formulated the section conjecture in his 1983 letter to Faltings for arithmetic fields, with the properness condition relevant to the version used here. The local $p$-adic problem keeps the full arithmetic and geometric fundamental groups but brings valuation theory into the foreground.

Koenigsmann established the birational $p$-adic section theorem: a birational section is located at a rational point. Pop later obtained a related criterion from suitable small quotients of the Galois group. These results do not automatically give the étale theorem because a section of a quotient need not lift through the larger group.

Pop and Stix proved that every étale section can be localized at a valuation extending the $p$-adic valuation. Their primary paper turns the question into one about which valuations can support a section. Pajwani developed this perspective using étale homotopy and fixed points in Berkovich spaces. The remaining issue is to force the relevant valuation to come from a rational point.

On the covering side, Tamagawa constructed resolution covers for points of stable marked models, Lepage treated Mumford curves, and Mochizuki–Tsujimura supplied positive-genus resolution for arbitrary hyperbolic curves over $p$-adic local fields. M1 builds on this geometry to impose exterior splitting and divisibility of disk-component degrees simultaneously.

For global applications, Harari–Stix connected sections with finite descent, while Stoll developed criteria under which finite descent cuts out exactly the rational points. Stoll’s paper, together with its corrections cited in M2, is the arithmetic input for several consequences. Betts–Stix also proved restrictions on local images of globally constrained Selmer sections; this concerns sections already arising from points at every completion, not the full local surjectivity claimed here.

4. What the new manuscripts claim

M1: one exterior sheet and nontrivial disk degrees

Let $X/\bar k$ be smooth, proper and connected of genus at least two. Choose finitely many disjoint open disks $V_1,\ldots,V_d$, each with radius in $|\Omega^\times|$ and with its closed disk strictly inside a larger local-coordinate disk. Set

$$U=X^{\mathrm{an}}\setminus\bigcup_{i=1}^{d}V_i.$$

Assume $U$ is nonempty and contains an $\Omega$-rational point. M1 claims the existence of a single connected finite étale surjective cover $g:Y\to X$ such that one component of $g^{-1}(U)$ maps isomorphically to all of $U$, while every connected component above every $V_i$ has degree divisible by $p$.

The disks are the prescribed disks; they are not shrunk in the conclusion. Only one exterior sheet is required. The other sheets need not split, and the final cover need not be Galois.

M2: the full local section conjecture

For every prime $p$, finite extension $k/\mathbb Q_p$, and smooth proper geometrically connected curve $X/k$ of genus at least two, M2 claims a bijection

$$X(k)\longrightarrow \{s:G_k\to\pi_1(X):r\circ s=\operatorname{id},\ s\text{ continuous}\} \big/\pi_1(X_{\bar k}).$$

The quotient is by conjugation by the full geometric étale fundamental group. The theorem both identifies every section with a point and distinguishes different points. No good-reduction assumption is imposed.

Its main intermediate theorem concerns completions. Let $\widetilde K\subset K^{\mathrm{sep}}$ be the compositum of the fields of all connected finite étale covers, retaining all embeddings. For a real-valued rank-one norm extending the normalized $p$-adic norm, choose a prolongation to $K^{\mathrm{sep}}$. Inside the same completed field, M2 claims

$$K^{\mathrm{sep}}\subseteq\widehat{\widetilde K} \subseteq\widehat{K^{\mathrm{sep}}}.$$

These are the specified embedded copies of the fields. The claim is about a completion; it does not say that every ramified cover was already an étale cover in the algebraic tower.

Global consequences retain an arithmetic condition

For a proper genus-at-least-two curve $C/F$ over a number field, a global section now determines local points at finite places and connected components at real places. M2 identifies these modified adelic tuples with the finite-cover descent locus: tuples that survive every finite torsor test after an appropriate global twist.

If that locus equals the diagonal rational points,

$$C(\mathbb A_F)_{\bullet}^{\mathrm{f\text{-}cov}}=C(F),$$

M2 claims the global section conjecture for $C$. The equality is an additional hypothesis, not a consequence of local point-theoreticity alone.

The stated sufficient cases include a nonconstant map to an abelian variety $A/F$ with $A(F)$ finite and no nonzero divisible subgroup in its Tate–Shafarevich group, or the analogous condition on the curve’s Jacobian. The Tate–Shafarevich group records torsors locally solvable everywhere; its divisible subgroup consists of elements divisible by every positive integer.

M2 also claims the conclusion for the smooth projective modular curves $X_0(N)$ and $X_1(N)$ over $\mathbb Q$ whenever their genus is at least two. These parametrize elliptic curves with a cyclic subgroup or a marked point of order $N$, with cusps added in their compactification. A further elliptic-target criterion uses the corank-zero Selmer converse from family 002. The modular-curve consequence is stated separately and does not depend on that Selmer specialization.

5. Why the advance matters

Provisional importance: 5/5. Taken as stated, the local theorem resolves a central problem in anabelian geometry: recovering arithmetic geometry from fundamental groups. It turns every abstract continuous splitting in the specified sequence into a unique $p$-adic point.

The completion theorem addresses the precise gap between étale and birational information. The global consequences then connect this local reconstruction with a different arithmetic problem, finite descent. M1 contributes a flexible geometric construction with simultaneous control over an entire exterior and every deleted disk.

The rating reflects this foundational local classification and its global applications. It does not imply that the unrestricted number-field section conjecture follows without further input.

6. The main idea

Make cycles inside the disks without disturbing the exterior

M1 first uses resolution covers to obtain positive-genus components above chosen positions inside each disk. An auxiliary prime $\ell\ne p$ supplies residue-component torsors. Along suitable skeletal paths, edges multiply faster than endpoint vertices, forcing a cycle above every disk.

A cyclic graph cover of degree $p$ can then change sheets around one of these cycles while remaining split above the whole exterior. Analytic covering theory and proper algebraization turn this into a finite étale algebraic cover. Taking all conjugate characters makes the construction compatible with the Galois action.

Finally, quotient by the stabilizer of a chosen exterior component. Its quotient sheet has degree one over $U$. Disk stabilizers retain a nontrivial normal $p$-subgroup, so every disk-component degree remains divisible by $p$. This explains why the two desired properties can coexist in one connected cover.

Replace branched disks by trivial pieces

M2’s completion argument uses analytic patching in two stages. Near a branch point a finite cover has a local form involving $t=z^e$. If another cover supplies a sufficiently divisible root of $t$ on a surrounding annulus, the pulled-back cover splits there. The part over the inner disk can then be replaced by a trivial cover and glued to the exterior.

The roots must exist on every inverse-image sheet of the annulus. One favorable sheet would not trivialize the full covering algebra needed for this replacement.

First, stable reduction and an odd prime $\ell\ne p$ supply prime-to-$p$ collar roots, after moving suitable collars into stable skeleta. Second, an auxiliary curve with parallel skeletal edges gives a Jacobian torus character of slope one. Pullback by multiplication by an arbitrary integer $m$ on the Jacobian supplies an $m$-th root of that character. On a sufficiently long collar, logarithms and exponentials handle the remaining analytic unit.

The auxiliary covers in the second stage retain horizontal ramification indices covered by the first stage. Repeating the replacement therefore places arbitrary finite separable extensions into the same completion. Horizontal ramification here is at closed points over $\bar k$, whose residue characteristic is zero; it is distinct from wild behavior in characteristic-$p$ reduction models.

Convert localization into a rational point

The completion inclusion identifies the relevant absolute and étale decomposition groups. A section localized at a rank-one valuation can consequently be lifted birationally. Koenigsmann’s theorem places that lift at a rational point, producing incompatible henselian valuations unless the localization has the required point-type form.

After excluding the other valuative cases, surjectivity onto $G_k$ forces the remaining closed point to have residue field exactly $k$. A Kummer-theoretic argument on the Jacobian proves uniqueness. The global step then uses finite descent and Stix’s finite-support theorem, rather than assuming that matching local points automatically gives a global point.

7. Scope, limitations, and open questions

The local theorem concerns proper smooth curves of genus at least two over finite extensions of $\mathbb Q_p$, using the full profinite groups. Open curves have cuspidal phenomena and are not covered by this statement. Geometrically abelian or other smaller quotients are different problems.

The global conclusion remains conditional on the stated finite-cover equality except in the cases where the manuscript invokes an established criterion. Local point existence at every place alone does not supply that equality.

M1 and M2 have distinct geometric statements. M1’s prescribed exterior-sheet theorem should not be substituted for M2’s all-sheet collar-root construction; the latter is the route explicitly used in the current section manuscript. Neither theorem supplies an effective algorithm to list all sections or rational points from a finite group presentation.

8. Sources and further reading

M1, Theorem 1.1, specifies the disk radii, exterior and simultaneous cover properties. M2, Theorems 1.1–1.2, states the section and completion claims. Its Theorem 8.2 and Corollaries 8.3–8.5 give the finite-descent, abelian-target and modular-curve consequences.

The Pop–Stix and Stoll papers linked in section 3 explain the valuative and descent inputs. M2’s references include the corrections to Stoll’s covering arguments and the current Selmer-converse source. Optional background on the étale fundamental group and Berkovich spaces can supplement the definitions here.

Source manuscripts

• M1. OpenAI, Étale covers with a prescribed exterior sheet (September 24, 2026).
• M2. OpenAI, The p-adic section conjecture (October 6, 2026).

Sources consulted on 10 October 2026 at repository revision fd4aeeb. Download the manuscript citations.