By @josebrox

613b16a / docs/results_20261117_v76.md

Raw
# Recovery and actual realization: 17 November 2026

The recovered 193-variable quartic and the complete v75 constructions
are retained without modification. The complete general count is 25;
the complete ordinary-abc count is 11, under their stated arithmetic
premises. Neither a complete 21-variable definition nor a complete
9-variable definition has been obtained.

The recovered quartic is specified in
[quartic193_20261117_v74.tex](../papers/quartic/quartic193_20261117_v74.tex)
and [seven_units193_v74](../computation/quartic/seven_units193_v74).
It has 193 total variables, 167 primitive integral quadratic squares,
exact degree four and content one. Both coefficient bases, every
coordinate map, the reconstructed source, and the improved exact height
bounds have independent certificates. The fixed coefficient is
`C=149^N`; the defining assertion requires sufficiently large fixed N
under the retained premises. Neither crossover in the coefficient
height formula is a numerical defining threshold.

The lost prior modulo-53 nonlift fixture has not been recovered or
reproduced. It is not evidence for these results. The available
modulo-53 certificates for the actual halfpoint curves certify
squarefreeness, a different assertion.

## Positive results retained from v75

| Construction | Total variables, including input | Degree upper bound |
|:--|--:|--:|
| Recovered quartic | 193 | 4, exact |
| General source, even abscissa tests | 25 | 4,863,007,848,923,136 |
| Ordinary rational abc, even abscissa tests | 11 | 230,932,097 |

The seven-contact derivative construction is also complete with 11
variables. Its physical equation has degree seven in the squared slope,
compared with degree nine for the original contact set. Its compiled
degree is larger than the best degree in the table. Details, source
circuits and all parameter qualifications are in the
[v75 guide](results_20261117_v75.md) and
[construction records](../data/results_20261117_v75.json).

The actual halfpoint weight curve has genus 12. Its seven-contact
physical pullback has genus 90 in the squared slope and genus 181 in
the slope. These are computations on an actual tuple, with all physical
maps retained; they are not rational-witness existence results.

## Contact alignment and the physical normalizers

For the general source, `a=1+rho*T^2` is determined by the root traces.
The relation `lambda=4/(rho*T^2)` makes `a*lambda=lambda+4`, so the two
contact numerators agree. With `s=(lambda+1)/(lambda+6)`, one has
`u1=s^10*u0` and
`w1/w0=(s^5*(1+u0)/(1+s^10*u0))^2`. Thus both weight pairs, and both
actual pole inputs when their traces are nonzero, have the same
rational square class.

The physical denominators remain different. Their unnormalized ratio
is `s^8`; the normalized ratio is
`(s^4*(1+u0)/(1+s^10*u0))^2`. Both are strictly less than one on the
stated domain. The exact normalized equality threshold is greater
than `2/3`, while `u0<rho<2/3`. The
[complete proof](../papers/pole_parameters/contact_alignment_20261117_v76.tex)
and [independent convolution certificate](../computation/pole_parameters/general21_alignment_v75)
distinguish this coefficient obstruction from accidental equality of
full detector products.

Fixing the aligned slope also changes the order of prime preparation.
The inherited theorem does not supply that new arithmetic realization.
Matching an input square class does not match the complement square
class or furnish the actual trace-square conditions.

## Actual physical counterexamples to automatic replacements

The following examples use actual elliptic multiplication points and
actual physical slopes. Each has all pre-pole guards at every point
character. They are not asserted to have the selected-prime
factorization or complete fixed finite-place preparation, and their
fixture B=3 is not a certified defining shift.

| Proposed automatic step | Actual tuple or family | Prime | Pole-input valuation |
|:--|:--|--:|:--|
| Remove the first four balancing denominators, with either raw exponent one or two | X=2, A=2P, H=10336P, U=20672P, z=670434583579295 | 76777 | -1 and -5 |
| Repair the raw squared variant by taking an odd input and odd multiples of P/2 | X=1, A=P, H=U=2665(P/2), every sufficiently large integer z congruent to 39827 modulo p squared | 85817 | -11 |
| Use the detector's built-in quadratic norm as a native pole witness | X=1, A=P, H=U=P/2, seven-contact z=(10621537/256)^8+8+100003 | 100003 | -7 |

The first example is checked by projective arithmetic, independent
affine arithmetic and a separate division-polynomial computation
modulo p cubed. The odd-coset example has independent affine and
division-polynomial computations modulo p cubed. In both examples
the actual trace-square radicand has odd valuation one.

The norm example evaluates all 30 native pre-pole guards at all 64
point characters. Independent DAG evaluators certify 1,920 nonzero
values at two different split primes. An exact real calculation
also checks 768 balanced denominators. The automatic norm identity
still holds, but the negative odd pole excludes every intrinsic
native witness.

Complete proofs:

- [Raw abscissa preparation](../negative_results/papers/pole_parameters/raw_abscissa_preparation_20261117_v75.tex).
- [Odd halfpoint cosets](../negative_results/papers/pole_parameters/odd_coset_raw_20261117_v76.tex).
- [Native pole norms and ramification](../negative_results/papers/pole_parameters/native_pole_ramification_20261117_v75.tex).

These examples exclude the stated automatic implications. They do not
exclude another prepared tuple for the same input, a new compatible
congruence construction, or a different complete definition.

## Native witnesses force ramification

The intrinsic and general native pole equations exclude negative odd
orders over the geometric discrete valued fields used in the proof.
Consequently, if a finite degree-d curve cover of the physical slope
line carries native witnesses, every ramification index over an odd
contact pole is even. With r distinct such poles,
`g >= 1+d*(r-4)/4`; in particular d is even.

| Fixed-point source | Distinct contact poles | Minimum genus at degree two |
|:--|--:|--:|
| First general detector | 10 | 4 |
| Seven-contact derivative | 14 | 6 |
| Nine-contact derivative | 18 | 8 |

This applies directly to native pole witnesses, rather than only to
one selected trace-square formula. It excludes genus-zero and
genus-one finite slope covers in the stated setting. The same
codimension-two reference-test argument forces constant slopes on
the abelian families specified in the paper, including the revised
even-abscissa factors. The inherited simplicity and point-map facts
are explicit premises. Constant-slope families, individual rational
points, and nonabelian families of higher dimension remain outside
this exclusion.

## Prepared abstract products and adaptive scalar criteria

The two classes `Delta(q,V)=(q^2-2q-1)^2+4*q*(q^2-1)*V`, with
`q=2+w*c^2`, and `Lambda(c)=1+w^2*V^4*c^4` preserve the desired
local pole implication. Neither is complete on all prepared positive
native products, even with the added rational scalar c unrestricted.

Two exact products at p=100003 have square reduced pole denominators,
the intrinsic-pole preparation, the older conditions at 2, 3 and 17,
and both real slots close to one. In the first, V is congruent to
one and the trace radicand is a nonsquare unit. The factor Delta is
a local square for every p-adic q. In the second, the trace radicand
has valuation -1 while every Lambda has even valuation. The universal
quantifiers follow from proofs over the local field, not a finite
search. Preparation can be strengthened at any fixed finite set of
primes away from the obstruction prime.

These are abstract native products. Their h and u values have not
been realized by the actual elliptic test tuple and physical contact
map. The
[paper](../negative_results/papers/pole_parameters/adaptive_prepared_pairs_20261117_v76.tex)
and [independent exact package](../negative_results/computation/pole_parameters/adaptive_prepared_pairs_v76)
retain that distinction. The second square equation is a genus-one
quartic with a prescribed abscissa class on an explicit isogenous
congruent-number curve; a rank statement alone does not recover that
class.

## The remaining 21/9 assertion

The defining coefficient parameter must be fixed first at one
sufficient value. For every required integer input, a proof must
then supply actual elliptic source points and a rational physical
slope, retain the source scales and every original and whole-character
guard, and make the actual trace-square radicand rationally square.
The general source requires the two square conditions simultaneously
at the same slope. The ordinary-abc source requires its single actual
square condition. The chosen completeness route must also justify
its real, fixed-place and selected-contact conditions.

A rational point on a free trace conic or on a weight cover does not
prove this assertion. The physical slope equations must be solved,
and an intermediate rational squared slope must itself be a rational
square. The exact coupled-weight lifting theorem in v74 checks this
issue without replacing the actual traces by free variables.

The complete coordinate counts therefore remain 25 and 11. The
[machine-readable realization checkpoint](../data/criteria_20261117_v76.json)
keeps the 21/9 assertions false as proved-result flags, and separately
records that none of the new obstructions disproves global realization.

## Reproduction and recovery

From the repository root:

```sh
python3 -B scripts/verify_v74.py --report /tmp/v74-verification.json
python3 -B scripts/verify_v75.py --report /tmp/v75-verification.json
python3 -B scripts/verify_v76.py --report /tmp/v76-verification.json
python3 -B scripts/verify.py --full --receipt /tmp/full-verification.json
```

The v76 gate rebuilds all five new exact packages in an isolated tree,
compares the rebuilt JSON bytes with the published certificates, checks
the independent reviews and manuscript hashes, and verifies the precise
scope of the result records. Each of the five new manuscripts was
compiled twice and every rendered page was inspected. Package and
publication manifests bind the deliverables and dependencies.

Recovery anchors already published before this supplement:

- `e7918368de24774458cb39c6d0d3052ac5f50849`: reconstructed 193-variable quartic.
- `8589039c1a0985e777bed63274cabae194c29b1a`: exact publication-manifest repair; full remote verification passed.
- `070645a564cf9d18d30eaa573289395ffb605a2e`: complete v75 improvements and exact physical lifting; full remote verification passed.

The saved proof and arithmetic certificates, rather than the interrupted
conversational claim, are the recovery baseline. The new checks audit
the stated transformations and examples; they do not constitute a
complete independent reproof of every inherited arithmetic foundation.