# RCHM × Hodge Conjecture — Research Round 16
## Generalization Test on Cubic Fourfolds: Canonical Fano Bridges, Denominator-Aware Closure, and Conditional K3 Escalation

**Status:** Internal research note + executable bridge-selection prototype  
**Version:** Round 16 / v1.6  
**Date:** 2026-09-17  
**Methodological base:** RCHM; Typed Relation & Constraint Algebra; X-Layer Mathematical Completeness; Graph-Native Mathematics  
**Previous rounds:** RCHM × Hodge Conjecture — Rounds 1–15  
**Canonical source:** UTF-8 Markdown

---

# 0. Round 16 Objective

Round 15 built a bridge-search compiler and replayed it on a Weil abelian fourfold.

Round 16 tests whether the compiler generalizes to a structurally different target:

$$
\boxed{
\text{smooth cubic fourfolds}.
}
$$

This test is useful because:

1. the target is not itself an abelian variety;
2. the generic target has no associated K3 surface in Hassett's special sense;
3. a canonical hyperkähler bridge exists for every smooth cubic;
4. special cubic fourfolds introduce a conditional K3 bridge;
5. a generic Kuga--Satake bridge is formally available at the Hodge level but is vastly more expensive.

The main result is that the compiler should select the Fano variety of lines as the first geometric bridge **before** trying associated K3 surfaces.

---

# 1. Cubic Fourfold Hodge Signature

Let:

$$
X\subset\mathbb P^5
$$

be a smooth cubic fourfold.

Its middle Hodge numbers are:

$$
h^{3,1}=1,
\qquad
h^{2,2}=21,
\qquad
h^{1,3}=1.
$$

The square of the hyperplane class:

$$
h^2
$$

is algebraic.

Removing it gives primitive cohomology:

$$
H^4_{\mathrm{prim}}(X,\mathbb Z)
$$

of rank:

$$
\boxed{
22.
}
$$

After the Tate twist:

$$
H^4_{\mathrm{prim}}(X,\mathbb Z)(1),
$$

the Hodge numbers are:

$$
\boxed{
(1,20,1).
}
$$

Thus the primitive middle cohomology is a weight-two Hodge structure of K3 type.

---

# 2. Bridge Compiler Input

For a generic smooth cubic, use the typed signature:

```text
target:
  type: smooth_cubic_fourfold
  middle_weight: 4
  primitive_rank: 22
  tate_shifted_weight: 2
  hodge_numbers_shifted: [1,20,1]
  primitive_discriminant_abs: 3
  canonical_algebraic_class: h^2
  special_lattice_known: false
```

The compiler is asked to search for a bridge that reduces the primitive Hodge-witness problem.

---

# 3. Candidate A — Fano Variety of Lines

Let:

$$
F(X)
$$

be the Fano variety of lines contained in $X$.

Beauville--Donagi prove:

$$
\boxed{
F(X)
\text{ is a hyperkähler fourfold of }
K3^{[2]}
\text{ type}.
}
$$

Let:

$$
P
=
\{(x,\ell)\in X\times F(X):x\in\ell\}
$$

be the universal line, with projections:

$$
p:P\to X,
\qquad
q:P\to F(X).
$$

The Abel--Jacobi map:

$$
\boxed{
\operatorname{AJ}
=
q_*p^*
:
H^4(X,\mathbb Z)
\to
H^2(F(X),\mathbb Z)
}
$$

restricts to an isomorphism:

$$
\boxed{
H^4_{\mathrm{prim}}(X,\mathbb Z)
\simeq
H^2_{\mathrm{prim}}(F(X),\mathbb Z).
}
$$

With the standard sign convention:

$$
q_{BB}(\operatorname{AJ}(x),\operatorname{AJ}(y))
=
-x\cdot y.
$$

Thus the bridge matches the entire primitive rank-$22$ K3-type structure.

---

# 4. Degree-Lowering Property

Under the Hodge shift:

$$
H^{2,2}_{\mathrm{prim}}(X)
\longrightarrow
H^{1,1}_{\mathrm{prim}}(F(X)).
$$

This is an exceptionally favorable bridge because Hodge classes on the bridge land in cohomological degree two.

By Lefschetz $(1,1)$:

$$
\boxed{
H^{1,1}(F(X))
\cap
H^2(F(X),\mathbb Q)
=
\operatorname{NS}(F(X))_{\mathbb Q}.
}
$$

Therefore every rational $(1,1)$ class on the bridge is a rational divisor class.

The difficult codimension-two witness problem on $X$ is reduced to a divisor problem on $F(X)$.

---

# 5. Rational Reverse Bridge

The bridge must also transport the algebraic divisor class back to $X$.

The standard Fano correspondence has a dual direction.

For primitive degree-two classes on $F(X)$, the correspondence involving:

$$
\boxed{
\frac16g^2,
}
$$

where:

$$
g
$$

is the Plücker polarization, identifies:

$$
H^2_{\mathrm{prim}}(F(X),\mathbb Z)(-1)
$$

with the appropriate primitive degree-six cohomology, and the dual Fano correspondence maps back to:

$$
H^4_{\mathrm{prim}}(X,\mathbb Z).
$$

Consequently, for every primitive integral Hodge class:

$$
\alpha\in H^{2,2}_{\mathrm{prim}}(X,\mathbb Z),
$$

the canonical bridge immediately gives:

$$
\boxed{
6\alpha
\text{ algebraic}.
}
$$

Therefore:

$$
\boxed{
\alpha
\text{ is algebraic over }\mathbb Q.
}
$$

This closes the rational Hodge problem for primitive degree-four classes.

The nonprimitive direction:

$$
\mathbb Q h^2
$$

is already algebraic.

Hence the bridge closes the rational Hodge conjecture for codimension-two classes on a smooth cubic fourfold.

This is a reconstruction of a known consequence, not a new proof claim.

---

# 6. Bridge Denominator

Round 16 introduces a new typed bridge quantity.

Define the **bridge denominator**:

$$
\boxed{
\operatorname{den}(\mathcal B)
=
\min
\left\{
N>0:
N\cdot D_{\mathrm{target}}
\text{ is algebraically reached by the bridge}
\right\},
}
$$

when such a uniform integer exists in the declared lattice scope.

For the simple Fano-line reverse bridge:

$$
\boxed{
\operatorname{den}(\mathcal B_F)\mid6.
}
$$

Using the standard correspondence described above, the explicit closure uses:

$$
\boxed{
N=6.
}
$$

Thus the compiler records:

```text
rational_hodge_closure: PASS
integral_lattice_closure: NOT_PROVED_BY_THIS_BRIDGE
bridge_denominator: 6
```

---

# 7. Rational Closure Is Not Integral Closure

The factor:

$$
\frac16g^2
$$

is rational but not integral as a correspondence class in the required sense.

Therefore the simple bridge does not by itself prove that every integral Hodge class is represented integrally.

It proves:

$$
6\alpha
$$

is algebraic.

The integral Hodge conjecture for cubic fourfolds is known by deeper results, but that is an additional closure theorem.

RCHM should therefore keep separate:

$$
\boxed{
\text{rational witness completeness}
}
$$

and:

$$
\boxed{
\text{integral lattice saturation}.
}
$$

---

# 8. Saturation Defect

Let:

$$
H_{\mathbb Z}
=
H^{2,2}(X,\mathbb Z)
$$

and let:

$$
A_{\mathbb Z}
$$

be the subgroup generated by integral algebraic cycle classes.

Define the saturation defect:

$$
\boxed{
\operatorname{SatDef}(X)
=
H_{\mathbb Z}/A_{\mathbb Z}.
}
$$

The rational Hodge conjecture states:

$$
\operatorname{SatDef}(X)
$$

is torsion.

The integral Hodge conjecture states:

$$
\boxed{
\operatorname{SatDef}(X)=0.
}
$$

A denominator-$N$ bridge shows:

$$
\boxed{
N\cdot\operatorname{SatDef}(X)=0.
}
$$

For the elementary Fano bridge:

$$
N=6.
$$

This is a precise way to record what the bridge has and has not solved.

---

# 9. Candidate B — Associated K3 Surface

The K3-type appearance of:

$$
H^4_{\mathrm{prim}}(X)(1)
$$

may tempt the compiler to immediately search for a K3 surface.

For a generic cubic this is too early.

A polarized K3 surface $(S,L)$ has primitive cohomology:

$$
H^2(S,\mathbb Z)_{L\text{-pr}}
$$

of rank:

$$
21.
$$

Thus it does not match the entire rank-$22$ primitive cubic structure directly.

The correct K3 bridge arises only after a cubic becomes **special**.

---

# 10. Special Cubic Extraction

A special cubic carries a primitive rank-two algebraic lattice:

$$
K_d
\subset
H^{2,2}(X,\mathbb Z)
$$

containing:

$$
h^2,
$$

with discriminant:

$$
d.
$$

Its orthogonal complement:

$$
\boxed{
K_d^\perp
}
$$

has rank:

$$
21.
$$

After Tate twist it has Hodge numbers:

$$
\boxed{
(1,19,1),
}
$$

matching the primitive cohomology of a polarized K3 surface.

Thus only after the special-lattice extraction does the K3 carrier type pass the rank/Hodge gate.

---

# 11. Hassett Admissibility Gate

Hassett's criterion says that a special cubic of discriminant $d$ has an associated K3 surface exactly when:

$$
\boxed{
d
\text{ is not divisible by }
4,\;9,
\text{ or any odd prime }
p\equiv2\pmod3.
}
$$

For smooth special cubic divisors one also has the usual nonempty condition:

$$
d>6,
\qquad
d\equiv0,2\pmod6.
$$

Therefore the compiler implements:

```text
if special_lattice is absent:
    associated_K3 = FAIL

elif d is not admissible:
    associated_K3 = FAIL

else:
    associated_K3 = PASS_COHOMOLOGICAL
```

This gate certifies only the Hodge/lattice bridge. It does **not** by itself certify that the Hodge isometry is induced by an algebraic correspondence. Algebraic witness transfer therefore remains a separate obligation.

---

# 12. Examples of the K3 Gate

### $d=14$

$$
14
$$

is not divisible by:

$$
4,\quad9,
$$

and its odd prime divisor:

$$
7\equiv1\pmod3.
$$

Therefore:

$$
\boxed{
d=14
\text{ is admissible}.
}
$$

An associated K3 surface exists at the Hodge/lattice level.

This does not by itself imply that the Hodge isometry is realized by an algebraic correspondence; that remains a separate bridge gate.

---

### $d=8$

$$
4\mid8.
$$

Therefore:

$$
\boxed{
d=8
\text{ is not admissible for an untwisted associated K3}.
}
$$

Even though cubic fourfolds containing a plane form a special divisor, the untwisted K3 gate fails.

Other twisted-K3 or categorical constructions require a distinct template.

---

# 13. Candidate C — Generic Kuga--Satake

Since:

$$
H^4_{\mathrm{prim}}(X)(1)
$$

is K3-type with:

$$
h^{2,0}=1,
$$

the Kuga--Satake construction is formally available at the Hodge level.

For a rank-$n$ K3-type Hodge structure:

$$
\dim_{\mathbb Q}\mathrm{Cl}^+(V)
=
2^{n-1},
$$

and:

$$
\dim\mathrm{KS}(V)
=
2^{n-2}.
$$

For the cubic primitive rank:

$$
n=22,
$$

this gives:

$$
\boxed{
\dim\mathrm{KS}(V)
=
2^{20}
=
1,048,576.
}
$$

Thus the generic Kuga--Satake candidate is structurally valid but enormous.

---

# 14. Bridge-Cost Comparison

The three candidate families behave very differently.

## Fano lines

$$
\dim F(X)=4.
$$

- universal for every smooth cubic;
- canonical algebraic correspondence;
- exact primitive Hodge match;
- Hodge classes reduced to divisors;
- rational closure immediate;
- denominator $6$ remains for elementary integral closure.

---

## Associated K3

$$
\dim S=2.
$$

- very small bridge;
- only available after special-lattice extraction;
- discriminant arithmetic gate required;
- not a generic cubic bridge.

---

## Kuga--Satake

$$
\dim\mathrm{KS}=1,048,576
$$

for rank $22$.

- formally generic at Hodge level;
- huge;
- algebraicity of the required correspondence is an additional theorem obligation;
- inferior to the canonical Fano bridge when the latter exists.

---

# 15. Bridge Priority Functional

This round uses a lexicographic search priority rather than pretending a universal numeric score.

Prefer, in order:

1. no hard gate failures;
2. canonical geometric realization;
3. algebraic correspondence already available;
4. exact Hodge-rank match;
5. reduction to a solved witness layer;
6. family universality;
7. lower bridge complexity/dimension;
8. smaller denominator;
9. stronger integral closure.

This ordering selects:

$$
\boxed{
F(X)
}
$$

for a generic cubic fourfold before associated K3 or generic Kuga--Satake.

---

# 16. Generic Cubic Compiler Output

Input:

```text
special_lattice_known: false
primitive_rank: 22
hodge_numbers: [1,20,1]
```

Output frontier:

```text
1. Fano_lines:
   status: ACTIVE
   correspondence: canonical
   witness_reduction: divisors
   rational_closure: PASS
   denominator: 6

2. Kuga_Satake:
   status: ACTIVE_BUT_EXPENSIVE
   dimension: 1048576
   correspondence_algebraicity: UNKNOWN

3. Associated_K3:
   status: FAIL
   reason: no special rank-2 algebraic lattice supplied
```

---

# 17. Special Cubic Compiler Output

For:

$$
X\in\mathcal C_{14},
$$

the compiler retains both:

$$
F(X)
$$

and an associated K3 surface.

The associated-K3 branch now has:

```text
special_lattice: PASS
d: 14
Hassett_admissible: PASS
rank_match: 21
Hodge_match: [1,19,1]
```

The Fano bridge remains universal.

The K3 bridge is a more specialized geometric compression.

---

# 18. Fano Perspective on the K3 Branch

For special cubic fourfolds satisfying stronger numerical conditions, the Fano variety of lines itself may be birational to:

$$
S^{[2]}
$$

or to a moduli space of stable sheaves on a K3 surface.

Thus the bridge graph can contain:

$$
\boxed{
X
\to
F(X)
\dashrightarrow
S^{[2]},
}
$$

or:

$$
\boxed{
X
\to
F(X)
\dashrightarrow
M_S(v).
}
$$

This is useful because the universal Fano bridge can be discovered first, and the K3 bridge can then appear as a **compression/refinement of the bridge**, rather than as an unrelated guess.

---

# 19. Bridge-of-Bridge Architecture

Round 16 therefore introduces:

$$
\boxed{
\textbf{Bridge-of-Bridge}.
}
$$

A first bridge:

$$
X\to Y
$$

solves or simplifies the Hodge problem.

A second structure:

$$
Y\dashrightarrow Z
$$

may reveal a smaller or more canonical geometry.

For cubic fourfolds:

$$
\boxed{
X
\to
F(X)
\dashrightarrow
K3\text{-moduli geometry}.
}
$$

This layered architecture is preferable to requiring the compiler to guess the terminal bridge immediately.

---

# 20. Canonical-Incidence Heuristic

A general lesson from cubic fourfolds is:

> Before invoking an abstract Kuga--Satake construction, inspect canonical parameter spaces of low-dimensional subvarieties on the target.

Examples include:

- Fano schemes of lines;
- Hilbert schemes of conics;
- moduli of stable objects;
- incidence varieties.

If such a parameter space carries a cohomological degree-lowering Abel--Jacobi map, it may dominate more abstract bridge templates.

Call this:

$$
\boxed{
\textbf{Canonical-Incidence Heuristic}.
}
$$

---

# 21. Degree-Lowering Bridge Index

Let:

$$
\Gamma_*:
H^{2p}(X)
\to
H^{2q}(Y)
$$

be a bridge map.

Define:

$$
\boxed{
\Delta_{\deg}
=
p-q.
}
$$

If:

$$
q=1,
$$

then Hodge witness search lands in divisor theory, which is particularly favorable.

For the cubic Fano bridge:

$$
p=2,
\qquad
q=1,
$$

so:

$$
\boxed{
\Delta_{\deg}=1.
}
$$

The bridge lands exactly in Lefschetz $(1,1)$ territory.

---

# 22. Solved-Layer Targeting

A bridge candidate should receive high priority if it lands in a Hodge level where witness completeness is already known.

Examples:

$$
H^{1,1}
\Rightarrow
\text{divisors}
$$

by Lefschetz $(1,1)$.

Thus define:

$$
\boxed{
\operatorname{SolvedLayer}(Y,q)
}
$$

as a boolean or typed theorem reference.

For:

$$
F(X),q=1,
$$

the value is:

$$
\boxed{
\operatorname{PASS}.
}
$$

This is stronger than mere Hodge-number matching.

---

# 23. Bridge-Denominator-Aware Closure

A bridge certificate should now include:

```text
bridge_closure:
  rational_closed: true
  integral_closed: false_or_external
  denominator_bound: 6
  saturation_defect_status: unresolved_by_this_bridge
```

This prevents the AI from converting a rational proof into an integral proof by type erasure.

---

# 24. Historical Calibration

The Hodge conjecture for cubic fourfolds is classical and known.

The purpose of Round 16 is not to re-prove it as a new theorem.

The point is that the bridge-search compiler, when given only the Hodge and incidence structure of a cubic, naturally rediscovers a powerful known geometry:

$$
\boxed{
\text{the Fano variety of lines}.
}
$$

It also explains why this bridge is computationally preferable to generic Kuga--Satake and logically prior to associated-K3 searches.

---

# 25. Prototype Compiler

The accompanying Python prototype evaluates three bridge templates:

1. Fano variety of lines;
2. associated K3 surface;
3. generic Kuga--Satake.

It checks:

- primitive ranks;
- Hodge-number compatibility;
- special-cubic rank reduction;
- Hassett admissibility;
- Kuga--Satake dimension;
- bridge denominator;
- rational/integral closure status.

It runs three test targets:

- generic cubic;
- special cubic $d=14$;
- special cubic $d=8$.

---

# 26. Exact Arithmetic Tests

The verifier confirms:

$$
\boxed{
2^{22-2}=1,048,576.
}
$$

It confirms:

$$
d=14
$$

is Hassett-admissible.

It confirms:

$$
d=8
$$

fails because:

$$
4\mid8.
$$

It confirms the rank transition:

$$
22
\to
21
$$

after removing a rank-two special algebraic lattice from total rank $23$.

---

# 27. New Principles

## Principle A — Canonical Incidence Before Abstract Bridge

Search natural parameter spaces of subvarieties before invoking huge abstract constructions.

---

## Principle B — Solved-Layer Preference

A bridge that lands in a witness-complete Hodge layer receives strong priority.

---

## Principle C — Denominator-Aware Closure

Rational and integral witness completeness are distinct typed outputs.

---

## Principle D — Conditional Bridge Activation

A bridge template may only become legal after another structure is detected.

The associated K3 branch activates only after special-lattice extraction and discriminant validation.

---

## Principle E — Bridge-of-Bridge

A universal first bridge may later compress into a smaller specialized bridge.

---

# 28. Round 16 Conclusion

The generalization test succeeds.

For a generic smooth cubic fourfold, the Bridge Search Compiler should not first search for an associated K3 surface.

The primitive Hodge signature is:

$$
\boxed{
H^4_{\mathrm{prim}}(X)(1)
:
(1,20,1),
\quad
\operatorname{rank}=22.
}
$$

The canonical Fano variety of lines:

$$
F(X)
$$

provides:

$$
\boxed{
H^4_{\mathrm{prim}}(X)(1)
\simeq
H^2_{\mathrm{prim}}(F(X)).
}
$$

The bridge lowers primitive $(2,2)$ classes to $(1,1)$ divisor classes.

By Lefschetz $(1,1)$, the bridge-side witness problem is solved.

The standard dual correspondence returns the class rationally and gives the explicit denominator:

$$
\boxed{
6.
}
$$

Thus:

$$
\boxed{
\text{rational Hodge closure}
=
PASS,
}
$$

while:

$$
\boxed{
\text{integral saturation}
}
$$

remains a separately typed obligation.

For special cubics, the compiler may subsequently activate an associated-K3 bridge if and only if the discriminant passes Hassett's arithmetic gate.

For a generic rank-$22$ K3-type Hodge structure, Kuga--Satake remains a formally available but extremely large fallback:

$$
\boxed{
\dim\mathrm{KS}=1,048,576.
}
$$

The resulting search ordering is:

$$
\boxed{
\text{canonical Fano incidence}
\to
\text{special K3 refinement}
\to
\text{abstract Kuga--Satake fallback}.
}
$$

This is the main closure of Round 16.

---

# References

1. Daniel Huybrechts, **Hodge theory of cubic fourfolds, their Fano varieties, and associated K3 categories**, arXiv:1811.02876.

2. Olivier Debarre, **The geometry of cubic hypersurfaces**, lecture notes/book material; section on cubic fourfolds and the Fano variety of lines.

3. Arnaud Beauville, Ron Donagi, **La variété des droites d'une hypersurface cubique de dimension 4**, C. R. Acad. Sci. Paris / related 1985 work.

4. Brendan Hassett, **Special cubic fourfolds**, Compositio Mathematica 120 (2000), 1--23.

5. Nicolas Addington, Richard Thomas, **Hodge theory and derived categories of cubic fourfolds**, Duke Mathematical Journal 163 (2014), 1885--1927.

6. The integral Hodge conjecture for codimension-two cycles on smooth cubic fourfolds is known by deeper work; the elementary denominator-$6$ bridge in this round is only used to certify rational closure.