Hodge conjecture

An equation can pick a smaller shape out of a larger one. For example, $y=0$ meets the curve $x^2+y^2=1$ at $(-1,0)$ and $(1,0)$. These pieces also make sense with complex coordinates, which allow a square root of −1, inside a compact plane that includes points at infinity. Hodge concerns smooth shapes defined by polynomials inside such compact complex projective spaces. A topological record describes coarse placement and intersections, rather than every point. Hodge asks whether every rational topological record passing a particular test from the complex coordinates comes from polynomial-defined pieces, allowing fractions as weights. These tested records are called Hodge classes.[1] [4]

First keep the geometric pieces separate from their cohomological records. In the complex projective plane, coordinates $[x:y:z]$ are nonzero triples identified under multiplication by a common nonzero scalar. The conic $x^2+y^2=z^2$ and line $y=0$ meet at $[1:0:1]$ and $[-1:0:1]$; there is no extra intersection with $z=0$. This is an illustrative calculation inside the conjecture’s domain. For a smooth projective complex variety $X$, a subvariety $Z$ of complex codimension $p$—meaning $p$ complex dimensions smaller than $X$—has a class $[Z]$ in degree $2p$ cohomology. The Hodge decomposition groups classes using differential forms built from complex coordinate changes and their conjugates. Type $(p,p)$ uses $p$ of each. Algebraic-cycle classes have this balanced type. The conjecture says any rational class of that type has the form $\alpha=\sum_i q_i[Z_i]$, with $q_i\in\mathbb{Q}$. The weights may be negative or fractional; it does not require one actual subvariety whose class is exactly $\alpha$.[3] [4]

The example shows how equations give pieces and intersection data; it does not settle the conjecture. The open direction starts with a permitted record and asks for algebraic pieces realizing it. This is proved in some dimensions, but not for all smooth projective complex shapes.[1]

The general gap is constructing algebraic cycles from the rational Hodge condition. The divisor case follows from the Lefschetz (1,1) theorem; it does not settle all codimensions. Replacing rational coefficients by integers fails in general, and dropping projectivity to allow all compact Kähler manifolds also changes the claim to a false statement. The conjecture does not say that topology alone reconstructs the original equations.[1] [2]

Polynomial shapes can be too complicated to study by solving all their equations. Topology gives coarser information, while complex coordinates reveal additional structure. Understanding how those descriptions constrain the algebraic pieces inside a shape is a way to organize geometry and compare shapes that otherwise look unrelated.[1] [3]

Consider the diagonal in $X\times X$, the algebraic piece made of pairs $(x,x)$. Its cohomology class splits into components associated with different cohomological degrees, and these components are Hodge classes. The conjecture would realize each by algebraic cycles, giving geometric correspondences that separate the degree components of cohomology. This is part of the motivation for motives: seeking a common geometric structure behind different cohomology theories, rather than treating their matching properties as coincidences.[2] [3]