# Quartics in 163 and 179 total variables This package is based on notes of **Gabe Conant with Astra 6.2**. It contains the complete coefficient tables for the scalar-point reduction proved in `papers/quartic/scalar_point163_179_20261120_v82.tex`. The defining assertions are conditional on the retained arithmetic interfaces stated in that manuscript and in the v78 shared-product audit. They are not consequences of a finite coefficient calculation alone. In particular, the native pole realization and the quantified prime-pattern/approximation inputs remain hypotheses of the transfer. | Construction | Total variables | Quadratic rows | Monomials | Degree | Content | |---|---:|---:|---:|---:|---:| | General family | 163 | 137 | 2789 | 4 | 1 | | Fully specified member | 179 | 151 | 3036 | 4 | 1 | Coordinate one is the rational input. Each polynomial is the sum of the squares of its listed residuals. The family uses $C=149^N$ and defines the integers for every sufficiently large fixed $N$; the sufficient threshold remains ineffective. Its exact coefficient height, for every $N\ge1$, is ```math \max\{126616122944180358034867798936,149^{4N+2}\}. ``` The fixed member uses $R=A_{\rm a}r/(1+25r)$, where ```math A_0=645904574641049503754777843805872291231896003979. ``` The additional factor and normalizer are ```math B_{\rm a}=263\cdot2161\cdot2437\cdot2837\cdot3793 =14904184670611231,\qquad A_{\rm a}=A_0B_{\rm a}^4. ``` Its height is exactly $A_{\rm a}^2$, of 748 bits. This is a variable-count improvement with a larger coefficient height than the preceding 180-variable member. ## What changes For $K=\mathbb Q(\sqrt2)$ and the retained reference reciprocal $f=1/x(H)$, the new scalar relations are ```math J=\frac{Z^2}{R^5}+\frac{f^2}{R^4},\qquad \tau\operatorname{Tr}(Z^2/R^5)=N(J),\qquad b_s=R(R+9792)\tau^2. ``` They force $Z\ne0$ and a nonzero trace, and retain the common-slot inequalities $v(Z)>9\mu/4$ and $v(f)>7\mu/4$ at every consumed odd scalar contact. Scaling the scalar inverse-Vieta chart by $Z$ then gives the identity $M=-2S-W$, so its two rational $M$ components can be removed. The general family weights only the scalar root by $R$ in its aggregate, so its contribution is the stored $P^*=R x(Q_2)$ and no inverse of $R$ is needed. The fixed member adds one inverse of $R$ and retains the effective unweighted aggregate. The square register $R^2$ remains in both. The general soundness proof uses the already proved positive weighted trace theorem with fixed weights $(1,r,1,1)$ at $r=R$. Its constants are uniform in $N$ and the input. The additional term satisfies ```math h(R)\le4h(\beta)+(2N+1)\log149+\log1801500. ``` The scalar lower bound absorbs this term uniformly before $N$ is chosen large. The weighted theorem has an ineffective additive constant, so it is not used to claim a fixed 178-variable member. The changed comparison is ```math E_2^*=A\overline{(V+M)}+2\overline{T_H}\,\overline Z =-\overline{Zt_2}\,C_2. ``` Its factor uses the pair $c^5E_3^4-c^{11}(E_2^*)^4$. The local conclusion is $M_j^3\mid a\Delta$. Since $|a\Delta|\le14B^5$, the retained general height bound has exponent $35/18<2$ and still proves soundness. The manuscript proves the complete ordered converse, including every scalar denominator, the native zero-coordinate branches, the original synchronized depth $d\ge1$, and the explicit endpoint $|a|\ge2^{960}$. ## Files and exact maps `source163.json` and `source179.json` are the complete physical Laurent row definitions. A term lists its powers of $C,c,R$, its ordinary source-labelled coordinate monomial, and its rational coefficient. Source label 112 stores $R^2$. The fixed source has no parameter. Each construction directory contains: - `quadrics.json`: all ordinary quadratic residuals; - `polynomial.json`: the complete expanded integer coefficient table; - `maps.json`: the integer Laurent weights, row weights, coordinate permutation, and definitions of the six unit coordinates; - `certificate.json`: the exact census and conditional theorem statement. For an ordinary coordinate the map is $x_i^{\rm source}=C^{k_i}c^{e_i}R^{r_i}x_i^{\rm stored}$, with $C$ omitted in the fixed member. The guards are $c c^{-1}=1$, $c^{-2}=(c^{-1})^2$, and $R_2=R^2$; the fixed member also has $R R^{-1}=1$. In the general family the ordinary input rows themselves force $R>0$ before any inverse map is used: the rescaled register $\rho$ satisfies $\rho=r(X)>0$ and $R(1+149^2\rho)=149C^2\rho$. Every map is invertible on every rational solution. No point chart is restricted to a generic fiber. ## Independent verification From the repository root, run: ```sh python -B computation/variable_reductions_v82/general/check.py ``` The checker uses Python 3 and SymPy. Its default invocation is read-only. It imports no construction producer. It independently reconstructs every retained physical row from the published enhanced v78 quadratic and weight tables, directly reconstructs the new field equations, reverses all compiled maps, and expands both sums of squares with SymPy. It also checks the universal point/native/comparison identities, all 26,778 anchor residues, and the exact endpoint inequalities. The immutable `independent.json` records the result and binds the source, tables, manuscript, and checker hashes. `arithmetic.json` records the precise arithmetic deductions; the manuscript supplies their proofs. To reproduce the tables into a separate directory, run: ```sh python -B computation/variable_reductions_v82/general/build.py --output /tmp/v82-general-reproduction ``` The producer uses only the Python standard library and the supplied physical rows and weight maps. It needs no integer solver and does not modify the source files. Its output matches the published tables exactly. The historical dependencies used by the independent check are pinned in the receipt and remain unmodified.