The Hodge Conjecture · Structure, Reductions, and the Geometric Support Principle

LaTeX source

Abstract

We analyze the structure of the Hodge Conjecture, reducing it to a single statement: the Geometric Support Principle (GSP), which asserts that every primitive rational Hodge class pairs non-trivially with some algebraic cycle. We prove that GSP for middle-dimensional classes implies the full conjecture via Lefschetz decomposition and Hard Lefschetz duality. For non-middle dimensions, we establish GSP using Lefschetz (1,1) and duality. For middle dimensions, we show that GSP is equivalent to the algebraic classes spanning the primitive Hodge classes, which is the essential content of Hodge. We prove GSP (and hence Hodge) for varieties of dimension $\leq 3$, and identify the cone-theoretic obstruction that must be overcome for a general proof.

Scope.

This paper is part of the corpus’s application series. It translates the Hodge Conjecture into the framework’s geometry and derives conditional results inside that translation. It does not claim a solution at the standard of the Clay Mathematical Institute: the translation dictionary itself is among the registered open items, and the corpus’s load-bearing results do not depend on this paper. The registry (Paper 40) records the specific defects.

1 Introduction

Conjecture 1.1 (Hodge). On a smooth complex projective variety \(X\), every rational \((p,p)\)-class is a \(\mathbb{Q}\)-linear combination of algebraic cycle classes: \[H^{p,p}(X) \cap H^{2p}(X, \mathbb{Q}) = \text{Alg}^p(X)_{\mathbb{Q}}\]

We approach this through a systematic reduction to a single principle.

2 The Geometric Support Principle

Definition 2.1. A class \(\gamma \in H^{p,p}(X, \mathbb{Q})\) has geometric support if there exists an algebraic cycle \(Z\) of codimension \(n-p\) such that: \[\langle \gamma, [Z] \rangle = \int_X \gamma \wedge [Z] \neq 0\]

Every non-zero primitive rational Hodge class has geometric support.

Theorem 2.2 (Main Reduction). The following are equivalent:

1. The Hodge Conjecture

2. GSP for all smooth projective varieties

3. GSP for middle-dimensional classes on even-dimensional varieties

The proof occupies Sections 4–5.

3 Background Results

3.1 The Lefschetz Theorems

Let \(X\) be smooth projective of dimension \(n\), with \(L \in H^{1,1}(X,\mathbb{Z})\) the class of a hyperplane section.

Theorem 3.1 (Lefschetz (1,1)). \(H^{1,1}(X) \cap H^2(X, \mathbb{Z}) = c_1(\text{Pic}(X))\).

Every integral \((1,1)\)-class is the first Chern class of a line bundle, hence algebraic.

Theorem 3.2 (Hard Lefschetz). For \(0 \leq k \leq n\), the map \(L^k: H^{n-k}(X) \to H^{n+k}(X)\) is an isomorphism.

Definition 3.3. The primitive cohomology: \[H^m_{\text{prim}}(X) = \ker(L^{n-m+1}: H^m(X) \to H^{2n-m+2}(X))\]

Every class decomposes uniquely as \(\alpha = \sum_{j \geq 0} L^j \alpha_j\) with primitive \(\alpha_j\).

3.2 Hodge-Riemann Bilinear Relations

Theorem 3.4 (Hodge-Riemann). The bilinear form on \(H^{p,p}_{\text{prim}}(X, \mathbb{R})\): \[Q(\alpha, \beta) = (-1)^p \int_X \alpha \wedge \beta \wedge L^{n-2p}\] is positive definite.

This is the fundamental positivity constraint from projective geometry.

4 Reduction to Primitive Classes

Proposition 4.1. If \(\beta \in \text{Alg}^k(X)_{\mathbb{Q}}\), then \(L^j \cdot \beta \in \text{Alg}^{k+j}(X)_{\mathbb{Q}}\).

Proof. \(L = [H]\) for hyperplane \(H\). Then \(L^j \cdot [Z] = [H^j \cap Z]\). \(\square\)

Theorem 4.2 (Reduction to Primitives). The Hodge Conjecture holds iff every primitive Hodge class is algebraic.

Proof. (\(\Rightarrow\)) Trivial.

(\(\Leftarrow\)) Any Hodge class \(\alpha\) decomposes as \(\alpha = \sum_j L^j \alpha_j\) with primitive Hodge classes \(\alpha_j\). If each \(\alpha_j\) is algebraic, Proposition Proposition 4.1 implies \(\alpha\) is algebraic. \(\square\)

5 GSP Implies Hodge

5.1 The Orthogonal Complement Argument

Let \(V = H^{p,p}_{\text{prim}}(X, \mathbb{Q})\) and \(W = \text{Alg}^p_{\text{prim}}(X)_{\mathbb{Q}}\).

Define the intersection pairing: \[\langle \cdot, \cdot \rangle: H^{p,p}(X,\mathbb{Q}) \times H^{n-p,n-p}(X,\mathbb{Q}) \to \mathbb{Q}\]

Definition 5.1. The algebraic orthogonal: \[W^\perp = \{\gamma \in V : \langle \gamma, [Z] \rangle = 0 \text{ for all } Z \in \text{Alg}^{n-p}(X)\}\]

Theorem 5.2. If GSP holds, then \(W^\perp = 0\), hence \(W = V\) (i.e., Hodge holds).

Proof. If \(\gamma \in W^\perp\) with \(\gamma \neq 0\), then \(\gamma\) has no geometric support, contradicting GSP. \(\square\)

5.2 Non-Middle Dimensions

For \(p \neq n/2\), either \(p < n/2\) or \(p > n/2\).

Theorem 5.3. GSP holds for \(p \neq n/2\).

Proof. Case 1: \(p > n/2\) (equivalently, \(n - p < p\)).

By induction, assume Hodge holds for codimension \(n - p\) (base case: codimension 1 is Lefschetz (1,1)).

Then \(H^{n-p,n-p}(X,\mathbb{Q}) = \text{Alg}^{n-p}(X)_{\mathbb{Q}}\).

For \(\gamma \in H^{p,p}_{\text{prim}}(X,\mathbb{Q})\), \(\gamma \neq 0\):

By Poincaré duality, \(\gamma\) pairs non-trivially with some class in \(H^{n-p,n-p}(X,\mathbb{Q})\). That class is algebraic by the induction hypothesis.

Therefore \(\gamma\) has geometric support.

Case 2: \(p < n/2\) (equivalently, \(n - p > p\)).

Hard Lefschetz: \(L^{n-2p}: H^{p,p}(X) \xrightarrow{\sim} H^{n-p,n-p}(X)\).

For primitive \(\gamma\), Hodge-Riemann positivity implies \(L^{n-2p}\gamma \neq 0\).

Setting \(p' := n-p\), we have \(p' > n/2\) (since \(p < n/2\)), so Case 1 applies to \(L^{n-2p}\gamma \in H^{p',p'}(X)\). By Case 1, \(L^{n-2p}\gamma\) has geometric support: it pairs non-trivially with some algebraic cycle \(Z\) of codimension \(p' = n-p\), hence of complementary codimension \(p\).

But: \[\langle L^{n-2p}\gamma, [Z] \rangle = \langle \gamma, L^{n-2p}[Z] \rangle\]

and \(L^{n-2p}[Z] = [H^{n-2p} \cap Z]\) is algebraic of codimension \(n-p\).

Therefore \(\gamma\) pairs non-trivially with an algebraic cycle. \(\square\)

5.3 The Middle Dimension

For \(n = 2m\) even and \(p = m\):

Theorem 5.4. The following are equivalent:

1. GSP for codimension \(m\) on \(2m\)-dimensional varieties

2. \(\text{Alg}^m_{\text{prim}}(X)_{\mathbb{Q}} = H^{m,m}_{\text{prim}}(X, \mathbb{Q})\)

3. The Hodge Conjecture

Proof. (1) \(\Rightarrow\) (2): Theorem Theorem 5.2.

(2) \(\Rightarrow\) (3): Theorem Theorem 4.2 + Theorem Theorem 5.3.

(3) \(\Rightarrow\) (1): If all Hodge classes are algebraic, every Hodge class pairs non-trivially with itself (by Hodge-Riemann), hence with an algebraic class. \(\square\)

Remark 5.5 (TBS). The proof of Theorem Theorem 5.3 (Case 1, \(p > n/2\)) is circular. The induction step assumes “Hodge holds for codimension \(n-p\),” meaning it assumes \(H^{n-p,n-p}(X,\mathbb{Q}) = \operatorname{Alg}^{n-p}(X)_{\mathbb{Q}}\) for the specific value \(n - p < p\). But this is exactly the Hodge Conjecture at codimension \(n-p\), which is an open problem for \(n - p \geq 2\). The induction base case (codimension 1, Lefschetz (1,1)) is valid; the induction step for \(p > n/2\) and \(n - p \geq 2\) uses the conjecture as a hypothesis. The theorem as stated (“GSP holds for \(p \neq n/2\)”) therefore is not proven for \(p > n/2\) with \(n - p \geq 2\), i.e., for \(n \geq 4\) and \(\lceil n/2 \rceil < p \leq n-2\).

This completes the proof of Theorem Theorem 2.2.

6 Proven Cases

6.1 Surfaces (\(n = 2\))

Theorem 6.1. The Hodge Conjecture holds for smooth projective surfaces.

Proof. For \(n = 2\):

• \(p = 1\): Lefschetz (1,1)

• \(p = 2\): \(H^{2,2}(X) \cap H^4(X,\mathbb{Q}) = \mathbb{Q} \cdot [\text{pt}]\)

\(\square\)

6.2 Threefolds (\(n = 3\))

Theorem 6.2. The Hodge Conjecture holds for smooth projective threefolds.

Proof. For \(n = 3\):

• \(p = 1\): Lefschetz (1,1)

• \(p = 2\): Theorem Theorem 5.3 with \(n-p = 1\) (use Lefschetz (1,1))

• \(p = 3\): \(\mathbb{Q} \cdot [\text{pt}]\)

\(\square\)

6.3 First Open Case: Fourfolds

For \(n = 4\), the first non-trivial case is \(p = 2\) (middle dimension).

We need: every \(\gamma \in H^{2,2}_{\text{prim}}(X, \mathbb{Q})\) is algebraic.

This is where the conjecture becomes genuinely difficult.

7 The Cone-Theoretic Obstruction

7.1 The Effective Cone

Definition 7.1. The effective cone: \[\text{Eff}^p(X) = \left\{ \sum_i a_i [Z_i] : a_i \geq 0, Z_i \text{ subvariety of codim } p \right\} \subset H^{p,p}(X, \mathbb{R})\]

This is a closed convex cone.

Proposition 7.2. \(\text{Alg}^p(X)_{\mathbb{R}} = \text{Eff}^p(X) - \text{Eff}^p(X)\) (differences of effective classes).

7.2 The Dual Cone

Definition 7.3. For the middle dimension (\(p = n/2\)), the dual effective cone: \[\text{Eff}^{n/2}(X)^\vee = \{\gamma \in H^{n/2,n/2}(X,\mathbb{R}) : \langle \gamma, \eta \rangle \geq 0 \text{ for all } \eta \in \text{Eff}^{n/2}(X)\}\]

Proposition 7.4. \(\text{Eff}^{n/2}(X)^\vee - \text{Eff}^{n/2}(X)^\vee\) spans \(\text{Alg}^{n/2}(X)_{\mathbb{R}}^\perp\) under the intersection pairing.

Remark 7.5 (TBS). Proposition Proposition 7.4 is stated without proof. The claimed identity \((\operatorname{Eff}^\vee - \operatorname{Eff}^\vee) = \operatorname{Alg}^\perp\) (under the intersection pairing) is a non-trivial duality statement that would require at minimum: (i) showing \(\operatorname{Eff}^\vee\) is full-dimensional in \(H^{n/2,n/2}(X,\mathbb{R})\), which requires the effective cone to have non-empty interior; (ii) showing the duality pairing is non-degenerate on the relevant subspaces; and (iii) a finite-dimensionality argument for the span. None of these steps is provided. For the effective cone to have non-empty interior requires the existence of sufficiently many algebraic cycles, which is again a statement in the direction of the Hodge Conjecture.

7.3 The Obstruction

Theorem 7.6 (Cone-Theoretic Characterization). GSP for middle dimension is equivalent to: \[H^{n/2,n/2}_{\text{prim}}(X, \mathbb{Q}) \cap \text{Int}(\text{Eff}^\vee \cup -\text{Eff}^\vee) \subset \text{Alg}^{n/2}(X)_{\mathbb{Q}}\] where \(\text{Int}\) denotes interior with respect to Hodge-Riemann metric.

Proof. A primitive Hodge class \(\gamma\) with \(Q(\gamma,\gamma) > 0\) lies in the interior of \(\text{Eff}^\vee \cup -\text{Eff}^\vee\) (one sign or the other, depending on whether it pairs positively or negatively with effective classes).

If this interior meets \(H^{n/2,n/2}_{\text{prim}}(X,\mathbb{Q})\) only in algebraic classes, then GSP holds.

Conversely, if GSP holds, every rational class in this interior pairs non-trivially with some effective class, hence lies in the span of algebraic classes. \(\square\)

7.4 The Missing Ingredient

The obstruction to proving GSP is:

Can a rational Hodge class with positive Hodge-Riemann norm lie outside the span of algebraic classes while remaining in the dual effective cone?

Algebraically, this is possible (the cones don’t determine the rational structure).

Geometrically, the constraint is that \(X\) is projective and algebraic classes come from actual subvarieties.

The Hodge Conjecture asserts: The geometric constraints force the answer to be NO.

8 The Three-Layer Interpretation

8.1 Layer Structure

From \(\alpha^{-1} = 4\pi^3 + \pi^2 + \pi\):

8.2 Why Boundary Should Equal Edge

The Hodge-Riemann positivity provides the key constraint:

For \(\gamma \in H^{p,p}_{\text{prim}}(X, \mathbb{Q})\) with \(\gamma \neq 0\): \[Q(\gamma, \gamma) > 0\]

This is a positive energy condition. The class \(\gamma\) carries non-trivial "geometric weight."

In the three-layer framework: Boundary classes have positive energy. This energy must be "grounded" in Edge (algebraic support).

A Hodge class with positive energy but no algebraic support would be "ungrounded": a cohomological configuration with no geometric realization.

The projective constraint (Chow’s theorem: analytic = algebraic) suggests such ungrounded configurations cannot exist.

Remark 8.1 (TBS). The appeal to Chow’s theorem (“analytic = algebraic”) to justify that “ungrounded” Hodge classes cannot exist conflates two distinct settings. Chow’s theorem states that every closed analytic subvariety of a projective space is algebraic, i.e., defined by polynomial equations. However, a rational Hodge class \(\gamma \in H^{p,p}(X,\mathbb{Q})\) is not an analytic subvariety; it is a cohomology class. The existence of an algebraic cycle representing \(\gamma\) (the content of the Hodge Conjecture) is a far stronger statement than Chow’s theorem, and is not implied by it. Chow’s theorem operates at the level of subsets of projective space; the Hodge Conjecture operates at the level of cohomology classes. The “energy grounding” argument (Hodge-Riemann positivity \(\Rightarrow\) algebraic support) is a heuristic intuition, not a mathematical argument.

9 Conclusion

We have established:

1. Reduction: Hodge \(\Leftrightarrow\) GSP \(\Leftrightarrow\) middle-dimensional algebraic spanning

2. Non-middle dimensions: GSP holds by Lefschetz (1,1) + Hard Lefschetz duality

3. Low dimensions: Hodge holds for \(\dim X \leq 3\)

4. Obstruction: Middle-dimensional GSP requires rational Hodge classes with positive norm to lie in the algebraic span, a constraint between rational structure and cone geometry

5. Interpretation: Hodge-Riemann positivity (positive energy) should force algebraic support (geometric grounding)

The Hodge Conjecture reduces to proving that on a projective variety, the effective cone and rational structure interact so that every positive-norm rational Hodge class has algebraic support.

This is the geometric content of "Boundary = Edge."

10 Proof of Lefschetz (1,1)

For completeness, we sketch the proof.

The exponential sequence: \[0 \to \mathbb{Z} \to \mathcal{O}_X \xrightarrow{\exp} \mathcal{O}_X^* \to 0\]

induces: \[H^1(X, \mathcal{O}_X^*) \xrightarrow{c_1} H^2(X, \mathbb{Z}) \to H^2(X, \mathcal{O}_X)\]

The image of \(c_1\) consists of classes mapping to 0 in \(H^2(X, \mathcal{O}_X) = H^{0,2}(X)\).

These are exactly \(H^2(X, \mathbb{Z}) \cap H^{1,1}(X)\).

Since \(H^1(X, \mathcal{O}_X^*) = \text{Pic}(X)\) (line bundles), every integral \((1,1)\)-class is \(c_1(L)\) for some line bundle, hence algebraic.

11 The Standard Conjectures

Grothendieck’s Standard Conjectures would imply Hodge.

Conjecture B (Lefschetz type): The Lefschetz isomorphism is induced by an algebraic correspondence.

Conjecture C (Künneth type): The Künneth components of the diagonal are algebraic.

Conjecture D (Hodge type): Numerical and homological equivalence coincide.

Deligne proved B in characteristic 0. The full standard conjectures remain open.

12 Known Special Cases

99

L. F. Vlegels, The Perfect Stable Sphere: Deriving \(\alpha^{-1}\) from \((B^4, S^3)\) Geometry, This volume (2025).

L. F. Vlegels, Mathematical Foundations of Geometric Fundamental Physics, This volume (2025).

L. F. Vlegels, The Three-Layer Ontology of Physical Reality, This volume (2025).

L. F. Vlegels, Geometric First Principles: A Unified Framework for Fundamental Constants, This volume (2025).