Grothendieck’s section conjecture

A curve given by equations can have solutions using the numbers of a chosen number field, such as ordinary fractions. Each such rational point produces a compatible way to lift the field’s symmetries into the curve’s covering symmetries. Does every allowed lift come from exactly one point? Such a lift is called a section. The conjecture concerns smooth projective curves of genus at least two; lifts that differ only by the specified geometric conjugation count as the same.[1] [2]

A rational point on a curve gives more than a solution of its equations: it provides a coherent lift of the coefficient field’s symmetries into the curve’s arithmetic covering group. The unresolved direction asks whether every such lift is produced by a point. A section must respect group multiplication, not merely pick an unrelated preimage of each symmetry.

For a smooth projective geometrically connected $X/K$, with $K$ a number field and genus at least two, consider $1\to\pi_1(X_{\overline K})\to\pi_1(X)\xrightarrow{p}G_K\to1$. A continuous section $s:G_K\to\pi_1(X)$ satisfies $p\circ s=\mathrm{id}$. Is the point-to-section map $X(K)\to\operatorname{Sec}(X/K)$ bijective, where sections are taken modulo $\pi_1(X_{\overline K})$-conjugacy?[1] [2]

Different rational points give different splittings; whether every splitting comes from a point is unknown.[1]

Finiteness of local images of Selmer sections under a no-CM-subfield hypothesis is partial progress, not general surjectivity.[2]

A rational solution must satisfy both the curve’s equations and restrictions on its coordinates. Covering symmetries encode these arithmetic restrictions together with the curve’s geometry. Recovering points from compatible symmetry data would connect the search for arithmetic solutions to a description that also records how finite covers fit together, without promising an algorithm for finding the points.[1] [2]

The section conjecture asks how much arithmetic is retained by the full, generally noncommutative covering group. In conjunction with the known finiteness of rational points, it predicts a matching finiteness and geometric origin for section classes. Betts–Stix obtain a restricted local-image finiteness result, linking methods for rational-point obstructions to constraints on sections. That link is substantive, but it neither reconstructs every global point nor proves that all sections are geometric.[1] [2]