By @josebrox

613b16a / docs/results_20261117_v70.md

Raw
# Actual-trace arithmetic and quadratic obstructions

The construction counts remain **25** in general, **11** under ordinary abc, and **195**, **229**, and **68** for the three retained quartic classes. Every count includes the input. The explicit 229-variable polynomial remains the primitive, fully expanded **74,007-byte** polynomial of coefficient height **149³⁰**. The arithmetic premises of each defining construction remain those in its manuscript.

## A smaller curve for actual rational realization

The [arithmetic reduction](../papers/pole_parameters/actual_trace_arithmetic_20261117_v70.tex) starts from the six actual detector tests at a fixed rational point tuple. Each reciprocal factor is a quotient of cubic polynomials in the physical weight `u`. The actual trace can therefore be written as `V=N(u)/D(u)` with degrees at most 36. Clearing a square denominator gives

```math
Y_0^2=\Pi(u)=N(u)^2(1+u)^2+uD(u)(2D(u)-N(u)).
```

The polynomial has degree at most 74, so its smooth normalization has genus at most **36**. The source's nonzero tests give a simple branch at `u=0`, proving geometric integrality. That branch and the two rational points above infinity are outside the guarded affine source.

The exact actual tuple `X=1`, `B=3`, `A=P`, `H=U=5P` gives the following genera:

| Curve | Genus |
|:--|--:|
| Physical-weight curve | 24 |
| Squared-slope quotient | 224 |
| Companion quotient | 225 |
| Original slope curve | 449 |

The four integer polynomials defining the trace and the degree-nine physical weight map are independently reconstructed coefficient for coefficient. Degree-preserving reductions modulo 101 and two explicit Bezout identities certify squarefreeness. The degree-450 composition polynomial is specified exactly by homogeneous composition. The two quotient maps split the slope-curve Jacobian up to an isogeny over the rationals; their genera add to 449.

The useful arithmetic work starts on the genus-24 curve. Every candidate must still pass

```math
u\mathsf B(t)-\mathsf A(t)=0,\qquad
\mathsf B(t)\ne0,\qquad t=z^2\text{ with }z\in\mathbb Q,
```

and all original point, trace, balanced-denominator and source guards. Here `t=z²`, distinct from the source contact expression `a²−z²`, whose nonzero guard is retained. A zero terminal ordinate is permitted. Singular affine points, when present in other fixed-tuple models, must be checked directly because they need not lift rationally to the normalization. Prepared witnesses also require all the prescribed local, real and whole-character conditions.

At the same actual tuple, the slope `z=a+12` gives an exact 17-adic square and a 31-adic nonsquare. This checks one split contact and the real condition. Both primes are in the retained exceptional set; the example is not fully prepared and does not disprove the global realization criterion. Likewise, `B=3` is an algebraic fixture, not a sufficient defining shift. The package supplies an exact arithmetic reduction and a smaller curve to study, without producing an admissible global witness.

## Actual-trace families through prepared points

The [vertical-valuation theorem](../negative_results/papers/pole_parameters/vertical_and_fixed_slope_20261117_v70.tex) supplies a restriction on rational maps from abelian varieties into the actual-trace square loci. It complements the [curve and constant-slope theorem](../negative_results/papers/pole_parameters/curves_and_abelian_families_20261117_v69.tex).

At a fixed input, every fixed algebraic curve in either specified locus has finitely many rational points. The new result excludes nonconstant abelian families whose images contain a prepared point. In the general two-detector locus every one of the nine point roles may vary. In the derivative locus this assertion also fixes the root point. The result includes the stated small square-class alternatives and finite point-chart covers.

The proof uses an actual arithmetic prime. At a split good rational prime `p>7` with positive odd detector-weight valuation, pass from the abelian parameter variety to its quotient by the connected kernel of the point map. A putative square root descends because the smaller function field is algebraically closed in the larger one. The quotient is isogenous to an abelian subvariety of a power of the restriction-of-scalars surface, so it has good reduction at the chosen prime. A nonconstant point coordinate reduces surjectively. Its reciprocal abscissa is therefore a unit at the generic point of the special fiber, in both split slots.

The factor valuations then make the actual trace divisible by the odd weight power. In the radicand

```math
\mathcal C=V^2+w(2-V),
```

the first term has strictly larger valuation than the second. Consequently the radicand has positive odd valuation and cannot be a square in that function field. The general first-detector statement also requires the explicitly retained unit condition on `lambda+1`. Prepared source points supply the needed primes and unit conditions.

This concerns nonconstant families, not individual points. It proves neither a failure of global realization nor a smaller defining count. The nine-variable route still needs a guarded rational solution at the actual trace for every sufficiently large positive integer. The general 21-variable route still needs both actual square equations simultaneously.

## Varying derivative roots with an affine slope relation

A separate proposition in the same paper fixes the input and rational constants `m,b`, and assumes

```math
z=ma+b.
```

The root point, its trace, the scale and the slope must then be constant in any rational abelian family. This proposition requires no chosen-prime hypothesis. It includes fixed slopes by taking `m=0`; it does not treat unrestricted simultaneous variation of the root and slope.

The partial derivative defining the detector is taken before restricting to the affine line. The restricted contact product has at least nine distinct roots. At a collision, the two logarithmic-derivative residues add to a nonzero value, so the detector weight still has a simple zero. The parallel cases `m=1` and `m=-1` either violate a guard identically or retain nine simple contacts. A square identity would make every corresponding zero divisor even. Since the scale is a fourth power, the pole divisor is also even. A finite etale cover of the abelian variety would then map nonconstantly to a hyperelliptic curve of genus at least four, which is impossible.

The [exact algebra certificate](../negative_results/computation/actual_trace_vertical_v70/certificate.json) covers all 35 fixed-slope collision values and four symbolic affine contact types. The [independent review](../negative_results/computation/actual_trace_vertical_v70/independent_review.md) separately checks the geometric descent, good reduction, surjectivity and valuation arguments against the primary references.

## Sequential six-unit bases for the quartic detector

The [quadratic scaling theorem](../negative_results/papers/quartic/sequential_unit_detector_obstruction_20261117_v70.tex) excludes a specified family of attempts to compress the 195-variable construction. Start with unit coordinates `u`, `u^-1`, `v`, `u^s/v` for `s=-2,-1,0,1,2`. Adjoin two further monomial units in sequence, each defined by a quadratic equation using the existing coordinates. These choices give **4,664 ordered extension choices**, **3,166 pairs consisting of an initial value of s and a final basis**, and **2,834 distinct final exponent sets**.

Each of the two detectors uses four comparison fourth powers and four reciprocal squares, assigned independently among four two-slot factors. The comparison slot exponents are 9 and 11; the reciprocal slot exponents are 1 and 3. The theorem allows arbitrary integer diagonal Laurent weights, independent component row division, and every bijective assignment of those quadratic tests. It requires literal monomial rewriting of the factor equations, with no cancellation or substitution using other equations.

The two constant terms and just one shared comparison leave only two possible unit sets. Both contain `v^-3`, with their last unit equal to `u^-1*v^-6` or `u*v^-6`. In either survivor, all four comparisons occupy the first slots of the second detector. Its four reciprocal tests are consequently forced to have second unit-weight component 3. The first detector forces that same component to be 0. This contradiction excludes both survivors for every permitted quadratic-test assignment, without reference to point-chart equations or linear-test allocations.

The proof uses an exact finite enumeration and a short algebraic contradiction. It does not depend on the supplementary SMT runs. The [portable certificate package](../negative_results/computation/sequential_units_v70) reproduces the enumeration with standard-library integer arithmetic. An independent calculation constructs the same bases through unordered pair sums and independently enumerates both constant exponents.

This is a theorem about the stated factor design and unit family. It does not establish a lower bound for arbitrary quartics or rule out other point coordinates, other factor powers, combinations of equations, or nonmonomial auxiliary coordinates. The established general quartic count remains **195**.

## A scalar helper cannot be halved in the specified chart

The [half-helper theorem](../negative_results/papers/elliptic_charts/half_scalar_helper_20261117_v70.tex) treats a different proposal: retain only the rational component of the field product `Zd` in a translated-secant chart. The resulting 21-coordinate interface has exactly **17** quadratic identities, and none forces the scalar quotient.

A 236 by 236 evaluation minor has determinant **514 modulo 1009**. The proof uses a torsion-free integral parameter ring and a saturated coefficient lattice to transfer this rank bound to characteristic zero. The independent checker verifies both split elliptic slots, all denominators, the exact minor, the seventeen coefficient relations and the false scalar residual.

At the nontorsion pair `U=R0`, `V=2R0` with the stated reference coordinate, the correct scalar is `Z=-2`; nevertheless `Z=1` satisfies every quadratic identity and every stated nonzero guard. Finitely many additional nonvanishing guards cannot remove all such incorrect values. Extra equality conditions involving the other global coordinates of the 195-variable source, different helpers and different charts are outside the result.

## Verification and scope

The current [construction records](../data/results_20261117_v68.json) and [research criteria](../data/criteria_20261117_v70.json) distinguish actual defining bounds from unresolved realization targets. The [Sun v2 audit](audits/sun_v2_20261117_v69/audit.md) and [power-shift theorem](../papers/number_fields/power_shifts_and_finite_exceptions_20261117_v69.tex) remain separate results. No all-number-field detector or counterexample has been obtained.

From the repository root:

```sh
python -B scripts/verify_v70.py
python -B scripts/verify_v69.py
python -B scripts/verify_v68.py
python -B scripts/verify.py --full
```

The first gate reproduces the exact certificates in temporary files and checks their hashes, independently reviewed manuscript versions, explicit hypotheses, result records and build receipts. Algebraic certificates do not stand in for the geometric proofs or the unresolved arithmetic foundations.