Computational artifacts

All scripts locate companions relative to their own path. two_edge_full.py additionally reads ../XVI-K-artifacts/hessian_full_out.txt (XVI-K is a frozen dependency; that folder is not modified by XVI-L). Every check raises RuntimeError on failure (fail-on-error); a run that exits with status $0$ and prints ALL CHECKS PASSED has passed all of its checks. The runs used for this note exited with status $0$.

roles. source_full.py: implements XII-C eq:finite-curvature, eq:ell3, eq:J3, eq:Ccoefficients, eq:M2 atom by atom on independent jet coordinates (sympy sparse ring over $\mathbb{Q}$); refuses a third-order jet; no IBP; no typed monomial table of $J_3$. two_edge_full.py: loads $P_A$ and ${L^{pq}}_{ij}$; builds ${T_i}^r:=(-\mathcal{P}{B}^i{)}^r$, $\tilde{P}$, and ${a^A}_{ab,pq}$; groups by numerical label by accumulation; all-sector coefficient assembly exact in the preparation jets and $w^1$ on $h_{\mu \nu }=(w^1{)}^2{\eta }_{\mu \nu }$ (not a general-$h$ assembly); cancellation-aware ${\Lambda }_{\mathrm{graph}}$ support on the stated background with explicit set checks against ${\Lambda }_{\mathrm{adm}}$ and the XVI-K labels; full source-sector K regression (all $25$ labels $\times$ ten components vs signed polar_s); exports nonzero coefficients to two_edge_coeff_table.txt; writes two_edge_support.txt; fail-on-error. kernel_inventory.py: enumerates placements from L.1 budgets and $|p|+|q|\le 2$; asserts ${\mathcal{V}}_{\mathrm{K}}\subseteq {\mathcal{V}}_{\mathrm{adm}}$ with $|{\Lambda }_{\mathrm{adm}}|=255$; records that $\{0,0\}$ arises in $C_2$ only; prints the inventory table. counterterm_map.py: coherence constraints on contact coefficients (selected wave-operator relations exhibited); jet-bound checks $(2-v)+|\gamma |\le 2+s\le 4$; arithmetic identity $(2-v)+N=2+s$ (not attainment); $W_{\rho }$ orders through $4$; raw contact counts (not dimensions); four overlapping interpretive attributes; fail-on-error. regression_L.py: exact $C_2$-only truncation to XVI-K and constant-slot truncation to XVI-J; scale-family illustrative floating-point recovery at $\mu ={\mu }_0$ (not a compactly supported bump proxy; not a distributional scale proof); net-sign check for all $|\gamma |\le 4$; fail-on-error.