Hostile internal audit: the balanced pairing

8 September 2026. Internal mathematical audit of an ATTACK research note. No external specialist review or endorsement is asserted.

Contents

Verdict

The balanced-pairing vanishing argument is valid with its explicitly declared formal dependencies, in the native all-plus central sector with a complete strict angular certificate. The decisive steps are the full integral boundary-fiber bijection and the exact scalar/phase conversion. The final inversion identity is a published GHHR result; the checker does not prove that external result.

The stronger no-rescue result exhausts the monomial translates certified by the same complete vector for a one-cusp triangulation. It does not exhaust all possible pairings. This is a supporting research note, not a proof that an actual filled vacuum or trace vanishes, not a conjecture resolution, and not a new Publication.

Attempted failure modes

Challenge	Audit finding	Boundary of the conclusion
The pairing might not annihilate the original meridian relation.	The left-seed formula uses the original character. Its expression as a shifted-character average is an exact Weyl identity, inherited from the prior correct-sided theorem.	The complete-character average alone has meridian eigenvalue $-1$ ; it is the translated functional that has the original eigenvalue $-A^2$ .
The root lattice may have been saturated to match normal surfaces.	Surjectivity reconstructs each $\lambda$ from the integer edge coefficients in $2X+nR_s+rR_l\in\mathbb E+\mathbb T$ .	No real or rational edge solution is substituted for an integer one.
Half-integral raw rows might be illegal index arguments.	Only their differences enter the root coordinates. The parity coset ensures integral differences; subtracting the block maximum gives an integral normal class.	The maximum and minimum translations themselves can be half-integral; they must not be rounded.
The image might cover only one mod-2 homology sector.	Doubling annihilates every mod-2 class. GHHR Theorem 7.1 therefore proves coverage of all integral normal classes with the specified boundary.	The result is the total $\sum_a I^a(\gamma)$ , not generally $I^0(\gamma)$ .
The same normal class might be counted many times because edge rows are dependent.	Summation is over the image $\Lambda$ , and $P(X)=-(\lambda+rb)/2$ is injective.	Summing over every free list of edge coefficients would be wrong.
Angular balancing could silently remove a phase.	$\chi_0(\lambda)=(-1)^nB^{2E-S_R}$ . The residual $(-1)^n$ is retained in the normal-index formula.	Replacing the character by its valuation alone invalidates the proof.
A longitude scalar was chosen to obtain cancellation.	The fixed $c_l=B^\rho$ is used throughout; (29) checks its cancellation against the seed translation exactly.	Other scalars would not satisfy the displayed native bridge.
A Weyl exponent has the wrong sign or factor two.	$\Omega(\lambda,rb)/2=rn$ , and the verifier checks the full bilinear translate identity.	The term $rn$ remains in every summand degree.
Positive angles might be incorrectly used to force every summand positive.	They supply all-input convergence through the retained theorem. They do not remove the term $rn$ , and no coefficient positivity is assumed.	The balanced value is not deduced by comparing only leading degrees.
Boundary inversion could have been assumed because a skein is unoriented.	The proof first establishes the exact full-fiber formula, then invokes GHHR’s published Remark 8.3, printed p. 312. The published and arXiv versions both contain it.	This source theorem, not a generic intuition about unoriented links, is a formal dependency.
Boundary inversion might hold term by term in the root seed.	It does not: $I_\Delta(0,1)$ begins with $1$ , while $I_\Delta(0,-1)$ begins with $-q=-A^4$ .	No $\lambda\mapsto-\lambda$ termwise equality is asserted.
The outer weighted sum may only converge after cancellation.	The inherited all-input theorem supplies unordered summability of the individual lattice terms before grouping.	Fiber grouping and $n\mapsto-n$ are justified without a conditional summation prescription.
Different same-certificate translates could be independent detectors.	In one cusp, $w=v+b$ lies in the saturated Lagrangian span. If its vacuum parity is supported, integral normal homology forces $w\in\Lambda$ , giving (33). Otherwise both terms are zero.	This is not the false claim that the saturation equals $\Lambda$ .
A multicusped boundary gives extra directions.	Yes. The proof covers $w\in\Lambda\otimes\mathbb R$ ; only in one cusp does same-character orthogonality force this automatically.	Other-cusp insertions are not exhausted.
The other-cusp test might have a missing normalization.	The full bridge is $c_t c_l^rF_0(W(w+rb))=\sum_n(-1)^nA^{rn}\mathcal Z(-t-rl-ns)$ , and the vacuum difference has the factor $c_l/c_t$ .	Formula (37) is a convergent correct-sided candidate test, not a proved nonzero value.
Simultaneous inversion might silently be replaced by separate-cusp inversion.	In (37), applying simultaneous inversion changes both cusp components, and reindexing changes the meridian coefficient. The remaining difference reverses only the other-cusp class.	Separate-cusp symmetry or asymmetry is not assumed or proved.
A zero pairing might imply a zero target.	The same functional has $\mathcal G_{-b}(l)=c_l(1+\text{positive order})\ne0$ .	The target remains nonzero. The actual vacuum may survive another pairing, as it does in the two-angle chamber.
The all-plus calculation might cover all sign sectors.	No. General sectors retain exact central sign characters, and need a separately justified phase-refined fiber formula.	The all-plus vanishing conclusion is not promoted to every six-root sector.
A derivative of a character might remain an eigenfunctional.	Differentiation gives an inhomogeneous moment equation unless a further correction is supplied.	No derivative/residue detector is claimed here.

Algebraic audit of the crucial identity

Let $M=(S_R+r\rho)/2-T$ , $\chi_X=E-T$ . Before using any normal-index symmetry, the native scalar is

$B^{r\rho}\,(-1)^nB^{-2E+S_R}\,A^{rn}\,B^{-2M} =(-1)^nA^{rn}B^{-2\chi_X}.$

The checker verifies the $A$ exponent as a rational-polynomial identity. It separately verifies the $i$ exponent modulo four: the raw and normal phases differ by $-4n$ , which is harmless for integer $n$ . Thus the argument does not conflate agreement of valuations with agreement of Laurent series.

The phase identity is then multiplied by the same exact $\prod J_\Delta(X_j)$ ; no series truncation is involved in proving the bridge.

Dependency audit

The normal-homology theorem is used twice: once to prove full-fiber surjectivity, and once to show that a supported same-certificate translate lies in the original image lattice. Those are separate uses of the integral kernel statement. The mod-2-zero property of every peripheral raw row is indispensable in both uses.

GHHR Remark 8.3 is used only after that bridge. The published source labels the equality as immediate from its definition; this packet cites it as supplied and does not claim an independent proof of the underlying boundary-index inversion. If that source identity were unavailable, the verified conclusion would stop at formula (26), not at universal balanced vanishing. The source dependency is therefore visible and removable rather than hidden in an assumed symmetry.

The general holomorphic-block factorization in Duan–Garoufalidis–Gu is conjectural and is not used. No surgery-invariant identification is used. The original Panitch–Romo conjecture is a scope comparator, not a premise proving anything in this note.

Reproducibility boundary

The standard-library verifier checks exact universal exponent identities, normal-coordinate reconstruction, parity elimination, correct-sided Weyl factors, and a local seed-series inequality with a certified finite cutoff. It does not prove the sourced GHHR theorems, recognize triangulations, establish a new actual angle certificate, enumerate topology, or establish first priority.

The proof is the mathematical argument in the main note. The number of checks is not a measure of mathematical significance.