Tate conjecture over finite fields

Equations can be studied with arithmetic modulo a prime: modulo 5, 2+4 equals1, and squaring gives 0,1,4,4,1 for the five possible inputs. The geometry is richer than this small table, because points over larger fields also belong to it. Do all its geometric records that remain unchanged under arithmetic symmetry come from actual equation-defined pieces such as points, curves or higher-dimensional subshapes?

Tate’s conjecture asks this for smooth projective varieties over finite fields: shapes without singularities, in a setting that includes directions at infinity. The records are cohomology classes, combined with rational ℓ-adic coefficients for a prime ℓ different from the field’s characteristic. The claim is that the symmetry-fixed records are spanned by the records of algebraic subvarieties, not that every record is one subvariety.[1] [2]

A geometric record can survive arithmetic symmetry even when no equation-defined subshape producing it is known. Tate’s conjecture asks whether that gap is always bridgeable for smooth projective varieties over finite fields. A finite-field calculation, such as $y=x^2$ modulo 5, illustrates the arithmetic but does not compute these cohomology records; one must also consider geometry over extensions of the field.

An algebraic cycle is a weighted combination of subvarieties. A cycle of codimension $r$ produces a class in degree-$2r$ ℓ-adic cohomology. Frobenius raises coordinates to the field-size power; the twist $r$ gives the normalization in which the classes of cycles defined over the base field are fixed. Let $X$ be smooth projective over finite $k$, and take $\ell\ne\operatorname{char}(k)$. The map starts at $\mathrm{CH}^r(X)\otimes\mathbb Q_\ell$, cycles modulo rational equivalence with rational ℓ-adic coefficients. Its target is the Galois-fixed part of $H_{\mathrm{et}}^{2r}(X_{\bar k},\mathbb Q_\ell(r))$. Must this cycle map be surjective for every $r$? Surjectivity means every target class is a linear combination of cycle classes, not necessarily the class of a single subvariety.[1] [2]

The cycle-to-record map and the arithmetic symmetry are well defined, and the conjecture is known for some classes of varieties. What is missing is a general proof that every fixed record lies in the span of algebraic pieces. A preprint reporting a result for abelian fourfolds has a narrower target than all smooth projective varieties.[1] [2] [3]

Integral coefficients and Frobenius semisimplicity are separate statements. Restricted new preprints do not prove the full target.[1] [2] [3]

The two descriptions reveal different information: equations give concrete geometric pieces, while cohomology organizes linear records of the whole shape. The conjecture would let arithmetic symmetry identify which records have geometric origins. It would provide a bridge for studying the same geometry through equations and through arithmetic, without claiming an algorithm that explicitly finds every piece.[1] [2]

Cohomology permits linear-algebraic calculations with geometric records; algebraic cycles are actual pieces defined by equations and can be used to relate varieties geometrically. Rational surjectivity would ensure that the invariant records need no additional, non-algebraic generators. That is why the conjecture matters beyond locating one missing subvariety. It also sits beside the Hodge problem over complex fields: both ask when the appropriate structured cohomology classes have algebraic origins. Connections between the conjectures require further hypotheses in precise results, and rational spanning does not itself supply a constructive recipe or uniqueness.[1] [2]