# RCHM × Hodge Conjecture — Research Round 12
## Blind Reconstruction from a Published Period Signature: Recovering a Bitangent Conic Pair

**Status:** Internal research note + reproducible exact algebra experiment  
**Version:** Round 12 / v1.2  
**Date:** 2026-09-17  
**Methodological base:** RCHM; Typed Relation & Constraint Algebra; X-Layer Mathematical Completeness; Graph-Native Mathematics  
**Previous rounds:** RCHM × Hodge Conjecture — Rounds 1–11  
**Canonical source:** UTF-8 Markdown

---

# 0. Round 12 Objective

Previous closed-loop experiments used theorem-backed Hodge signatures on carriers whose ground-truth geometry was known to us during construction.

Round 12 performs a stricter blind benchmark.

The input is taken from a published period-reconstruction computation:

1. a smooth quartic surface;
2. a period-derived linear form spanning the degree-$1$ annihilator of a conic class;
3. the minimal polynomial of the algebraic coefficient appearing in that linear form.

The conic equations themselves are **not** used as input.

The goal is to determine whether the RCHM gate pipeline can reconstruct the geometric witnesses from:

$$
\boxed{
\text{published Hodge/period signature}
+
\text{ambient equation}.
}
$$

It can.

---

# 1. Published Quartic Surface

Consider:

$$
\boxed{
X
=
V
\left(
5x^4
-4x^2zw
+8y^4
-5z^4
+4zw^3
\right)
\subset
\mathbb P^3.
}
$$

Movasati--Sertöz prove that this quartic has Picard number $14$, contains $102$ conics and $4$ lines, and that the conic planes split into several Galois orbits.

One orbit consists of three planes:

$$
\boxed{
h_a=z+aw=0,
}
$$

where:

$$
\boxed{
25a^3+4a+20=0.
}
$$

Each such plane contains a pair of bitangent conics.

For the present benchmark, only:

$$
X,
\qquad
h_a,
\qquad
m(a)=25a^3+4a+20
$$

are treated as input.

The conic equations are hidden from the reconstruction pipeline.

---

# 2. Information Mode

The available information is:

$$
\boxed{
\text{Period-Signature Mode}.
}
$$

The period computation has already produced a degree-$1$ annihilator direction:

$$
\widetilde I_{\delta,1}
=
\langle h_a\rangle.
$$

Thus the first carrier-safe geometric envelope is:

$$
\boxed{
E_1
=
X\cap V(h_a).
}
$$

This is a plane quartic curve.

At this stage, the pipeline does **not** assume:

- reducibility;
- number of components;
- conic carrier type;
- equations of the components;
- tangency behavior.

These must be reconstructed.

---

# 3. Restricting the Ambient Quartic to the Period Plane

From:

$$
h_a=z+aw=0,
$$

we set:

$$
z=-aw.
$$

Substituting into the quartic equation gives:

$$
\begin{aligned}
F_a(x,y,w)
={}&
5x^4
+4a x^2w^2
+8y^4\\
&-5a^4w^4
-4aw^4.
\end{aligned}
$$

The coefficient $a$ satisfies:

$$
25a^3+4a+20=0.
$$

Multiplying this relation by $a$ gives:

$$
25a^4+4a^2+20a=0,
$$

so:

$$
-5a^4-4a
=
\frac45a^2.
$$

Hence:

$$
\boxed{
F_a
=
5x^4
+4a x^2w^2
+8y^4
+\frac45a^2w^4.
}
$$

Multiplying by $5$:

$$
\boxed{
5F_a
=
25x^4
+20a x^2w^2
+40y^4
+4a^2w^4.
}
$$

Therefore:

$$
\boxed{
5F_a
=
(5x^2+2aw^2)^2
+
40y^4.
}
$$

This identity is obtained solely from the ambient quartic and the published period-plane signature.

---

# 4. Minimal Splitting Extension

Introduce:

$$
b
$$

satisfying:

$$
\boxed{
b^2=-10.
}
$$

Then:

$$
40y^4
=
-(2by^2)^2.
$$

Therefore:

$$
\boxed{
5F_a
=
C_+C_-,
}
$$

where:

$$
\boxed{
C_\pm
=
5x^2+2aw^2
\pm
2by^2.
}
$$

Thus the hidden carriers are reconstructed as two conics in the period plane:

$$
\boxed{
\mathcal C_\pm
=
V
\left(
z+aw,\;
5x^2+2aw^2\pm2by^2
\right).
}
$$

The carrier equations were not input.

They arise from exact factorization of the safely reconstructed envelope.

---

# 5. Field-of-Definition Layer

The period plane is defined over:

$$
\boxed{
K=\mathbb Q(a),
}
$$

where:

$$
25a^3+4a+20=0.
$$

The two individual conics are defined over:

$$
\boxed{
L=K(b),
\qquad
b^2=-10.
}
$$

The involution:

$$
b\mapsto-b
$$

exchanges:

$$
\mathcal C_+
\leftrightarrow
\mathcal C_-.
$$

Therefore the unordered pair:

$$
\boxed{
\{\mathcal C_+,\mathcal C_-\}
}
$$

is defined over the smaller field $K$ even when the individual components require the quadratic extension.

This is an explicit instance of the RCHM field-of-definition layer.

---

# 6. Smoothness of the Reconstructed Conics

Inside the plane with coordinates:

$$
[x:y:w],
$$

the conic equations are:

$$
C_\pm
=
5x^2
\pm2by^2
+2aw^2.
$$

Their gradients are:

$$
\nabla C_\pm
=
(10x,\;\pm4by,\;4aw).
$$

Because:

$$
a\neq0
$$

and:

$$
b\neq0,
$$

the three partial derivatives can vanish simultaneously only at:

$$
x=y=w=0.
$$

This is not a projective point.

Therefore:

$$
\boxed{
\mathcal C_+
\text{ and }
\mathcal C_-
\text{ are smooth conics}.
}
$$

---

# 7. Exact Bitangency Reconstruction

The two conics intersect where:

$$
C_+=C_-=0.
$$

Subtracting gives:

$$
4by^2=0.
$$

Since:

$$
b\neq0,
$$

we get:

$$
y=0.
$$

Then:

$$
5x^2+2aw^2=0.
$$

Over the algebraic closure, this gives two projective intersection points.

At either point:

$$
y=0,
$$

so:

$$
\nabla C_+
=
(10x,0,4aw),
$$

and:

$$
\nabla C_-
=
(10x,0,4aw).
$$

Thus the tangent lines coincide at both intersection points.

Therefore:

$$
\boxed{
\mathcal C_+
\text{ and }
\mathcal C_-
\text{ are bitangent}.
}
$$

This tangency property is not inserted from the literature statement.

It is recovered from the reconstructed equations.

---

# 8. Intersection Multiplicity

Two plane conics have total Bézout intersection number:

$$
2\cdot2=4.
$$

The reconstructed conics have two distinct intersection points.

At each point their tangent lines coincide.

Hence each intersection point contributes multiplicity at least $2$.

Since the total is $4$, both multiplicities are exactly:

$$
2.
$$

Therefore:

$$
\boxed{
I_p(\mathcal C_+,\mathcal C_-)=2
}
$$

at each of the two intersection points.

This gives the exact bitangency structure.

---

# 9. Hodge-Class Numerical Calibration

The published period computation finds classes:

$$
\delta
$$

in the Hodge lattice satisfying:

$$
\delta^2=-2,
$$

and:

$$
h\cdot\delta=2.
$$

These are the numerical invariants expected for smooth conics on a quartic K3 surface:

$$
\deg C=2,
$$

and by adjunction:

$$
C^2=-2.
$$

Thus the reconstructed carrier type passes the full-class numerical gate.

---

# 10. Gate-by-Gate Reconstruction

The blind reconstruction can now be written entirely as RCHM gates.

## G0 — Hodge-lattice candidate

Input class satisfies:

$$
\delta^2=-2,
\qquad
h\cdot\delta=2.
$$

Candidate carrier type:

$$
\text{degree-2 rational curve}.
$$

---

## G1 — Period annihilator

Period computation produces:

$$
\widetilde I_{\delta,1}
=
\langle z+aw\rangle.
$$

This gives the carrier envelope:

$$
E_1=X\cap V(z+aw).
$$

---

## G2 — Exact coefficient reconstruction

The coefficient satisfies:

$$
25a^3+4a+20=0.
$$

So the numerical period output has been algebraically reconstructed.

---

## G3 — Ambient restriction

Compute:

$$
X|_{z=-aw}.
$$

No carrier equations are assumed.

---

## G4 — Factorization

Using the minimal polynomial of $a$:

$$
5F_a
=
(5x^2+2aw^2)^2+40y^4.
$$

---

## G5 — Minimal field extension

Introduce:

$$
b^2=-10.
$$

Factor:

$$
5F_a=C_+C_-.
$$

---

## G6 — Geometric realizability

Each:

$$
\mathcal C_\pm
$$

is automatically contained in:

$$
X
$$

because it is a component of the exact ambient restriction.

---

## G7 — Carrier verification

Verify:

- degree $2$;
- smoothness;
- rationality;
- pairwise bitangency;
- expected K3 numerical invariants.

The witness is reconstructed.

---

# 11. Ground-Truth Separation

This round uses the literature statement that the plane contains a pair of bitangent conics only as an external validation target.

The reconstruction algorithm itself is given only:

$$
\boxed{
X,
\quad
h_a,
\quad
m(a).
}
$$

It derives:

$$
\boxed{
C_\pm
}
$$

by exact algebra.

Therefore this is a genuine **blind carrier reconstruction benchmark** relative to the published period signature.

---

# 12. Period-to-Carrier Compression

The published computational route is:

$$
X
\to
\text{period matrix}
\to
\text{Hodge lattice}
\to
\delta
\to
\widetilde I_{\delta,1}.
$$

Round 12 begins at the last object:

$$
\widetilde I_{\delta,1}.
$$

From there the RCHM gate pipeline gives:

$$
\boxed{
\widetilde I_{\delta,1}
\to
\text{plane envelope}
\to
\text{ambient restriction}
\to
\text{factorization}
\to
\text{conic witnesses}.
}
$$

Thus the expensive transcendental step and the cheap exact geometric step are cleanly separated.

---

# 13. Evidence-Typed Edge Chain

The proof graph contains:

$$
X
\xrightarrow{\text{published period computation}}
h_a,
$$

$$
h_a
\xrightarrow{\text{exact substitution}}
F_a,
$$

$$
(F_a,m(a))
\xrightarrow{\text{exact reduction}}
(5x^2+2aw^2)^2+40y^4,
$$

$$
b^2=-10
\xrightarrow{\text{exact factorization}}
C_+C_-,
$$

$$
C_\pm
\xrightarrow{\text{exact gradient test}}
\text{smooth conics},
$$

$$
(C_+,C_-)
\xrightarrow{\text{exact tangent test}}
\text{bitangent pair}.
$$

Different evidence classes remain explicitly typed.

---

# 14. What Was Not Recomputed

Round 12 did not independently recompute:

- the 5000-digit period matrix;
- the Hodge lattice;
- the Picard rank $14$;
- the LLL reconstruction of the period-plane coefficient;
- the complete set of $102$ conics.

Those are published results.

The new computation begins from the published period-derived linear form and reconstructs the carrier pair exactly.

---

# 15. A New Distinction: Signature Field vs. Carrier Field

The period signature plane is defined over:

$$
K=\mathbb Q(a).
$$

But the individual carriers require:

$$
L=K(\sqrt{-10}).
$$

Therefore:

$$
\boxed{
\text{field of the signature}
\neq
\text{field of an individual witness}.
}
$$

The signature may define a Galois-stable witness ensemble rather than a single witness.

This suggests a refined witness type:

$$
\boxed{
\operatorname{WitnessEnsemble}
(
K,
\{W_i\},
\operatorname{Gal}(L/K)
).
}
$$

---

# 16. Witness Ensemble

Define:

$$
\mathcal E
=
\{\mathcal C_+,\mathcal C_-\}.
$$

The Galois involution:

$$
b\mapsto-b
$$

permutes the elements of:

$$
\mathcal E.
$$

So:

$$
\mathcal E
$$

is defined over $K$ even when its individual members are not.

This is more faithful than forcing the Hodge signature to reconstruct one carrier over the signature field.

Thus:

$$
\boxed{
\text{one Hodge signature}
\to
\text{one witness ensemble}
}
$$

can be the correct inverse output.

---

# 17. Ensemble-Level Identifiability

Round 6 defined carrier identifiability depth.

Round 12 shows that identifiability must be relative to the desired object.

The degree-$1$ signature may uniquely identify the **unordered conic pair** over $K$ even if it does not define either conic individually over $K$.

Therefore define:

$$
\boxed{
\iota_{\mathrm{ens}}
}
$$

for witness-ensemble identifiability.

One may have:

$$
\boxed{
\iota_{\mathrm{ens}}
<
\iota_{\mathrm{individual}}.
}
$$

This is a field-sensitive form of identifiability.

---

# 18. Relation to Witness-Carrier Transmutation

The conic pair is not a carrier transmutation in the Round 2 sense.

Instead it exhibits **carrier splitting**:

$$
\boxed{
\text{one period-plane signature}
\to
\text{two conjugate carrier components}.
}
$$

Thus inverse reconstruction can branch.

The graph must permit:

$$
\text{signature node}
\to
\text{witness ensemble node}
\to
\text{individual carrier nodes}.
$$

---

# 19. AMRAL Representation

Suggested signature object:

```text
period_signature:
  ambient_variety_ref
  hodge_class_ref
  degree: 1
  linear_form: z + a*w
  coefficient_field:
    generator: a
    minimal_polynomial: 25*a^3 + 4*a + 20
```

Suggested ensemble:

```text
witness_ensemble:
  base_field: Q(a)
  splitting_field: Q(a,b)
  extension_relation: b^2 + 10 = 0
  members:
    - 5*x^2 + 2*a*w^2 + 2*b*y^2
    - 5*x^2 + 2*a*w^2 - 2*b*y^2
  galois_action:
    b -> -b
  ensemble_verified: true
```

Suggested relation:

```text
carrier_relation:
  type: bitangent_pair
  intersection_points: 2
  local_intersection_multiplicity: 2
```

---

# 20. Reproducible Exact Verification

The accompanying Python script checks symbolically:

1. the minimal-polynomial reduction:
   $$-5a^4-4a=\frac45a^2;$$
2. the restricted quartic identity;
3. the factorization after adjoining $b^2=-10$;
4. nonvanishing of $a$ and $b$ from their defining equations;
5. smoothness of the reconstructed conics;
6. the intersection equations;
7. equality of tangent gradients at both intersection points;
8. degree and K3 numerical calibration.

The script uses polynomial reduction modulo:

$$
25a^3+4a+20
$$

and:

$$
b^2+10.
$$

No floating-point arithmetic is required.

---

# 21. New Principles

## Principle A — Published-Signature Blind Reconstruction

A period/Hodge output can be treated as a typed external edge, after which carrier reconstruction may proceed independently and exactly.

---

## Principle B — Signature-Field / Carrier-Field Separation

$$
\boxed{
K_{\mathrm{signature}}
\subsetneq
K_{\mathrm{carrier}}
}
$$

may occur naturally.

---

## Principle C — Witness-Ensemble Principle

The correct inverse image of a Hodge signature may be a Galois-stable ensemble rather than a single carrier.

---

## Principle D — Carrier Splitting

A safely reconstructed envelope may factor into multiple valid witness components.

The inverse graph should branch rather than forcing a unique output.

---

## Principle E — Ensemble Identifiability

A witness ensemble can be uniquely reconstructed over the signature field even when individual members require a field extension.

---

# 22. Round 12 Conclusion

Round 12 completes the first genuinely blind carrier reconstruction in the RCHM Hodge series using a published period-derived signature.

Input:

$$
\boxed{
X:
5x^4-4x^2zw+8y^4-5z^4+4zw^3=0,
}
$$

$$
\boxed{
h_a=z+aw,
}
$$

$$
\boxed{
25a^3+4a+20=0.
}
$$

No conic equations are used as input.

Exact reconstruction gives:

$$
\boxed{
5X|_{h_a}
=
\left(
5x^2+2aw^2+2by^2
\right)
\left(
5x^2+2aw^2-2by^2
\right),
}
$$

where:

$$
\boxed{
b^2=-10.
}
$$

Therefore the two hidden witnesses are:

$$
\boxed{
\mathcal C_\pm:
\begin{cases}
z+aw=0,\\
5x^2+2aw^2\pm2by^2=0.
\end{cases}
}
$$

They are smooth conics.

They meet at two points.

At both points they have the same tangent line.

Hence they form a bitangent pair, exactly matching the published geometric conclusion.

The reconstruction chain is:

$$
\boxed{
\text{period signature}
\to
\text{safe plane envelope}
\to
\text{exact ambient restriction}
\to
\text{field extension}
\to
\text{witness ensemble}
\to
\text{bitangency verification}.
}
$$

This is the main closure of Round 12.

---

# References

1. Hossein Movasati, Emre Can Sertöz, **On reconstructing subvarieties from their periods**, Rendiconti del Circolo Matematico di Palermo Series 2, 70 (2021), 1441--1457.  
   DOI: https://doi.org/10.1007/s12215-020-00562-x  
   Relevant material: Section 3.1, especially Proposition 3.2 and the conic-plane reconstruction.

2. Emre Can Sertöz, **PeriodSuite**, public software implementation for period computations of hypersurfaces and Hodge-lattice reconstruction.  
   Repository: https://github.com/emresertoz/PeriodSuite