q=10 TP/TPHM precession gate decision (2026-09-27)

Using the frozen measurements at qualification head 5380379, no bounded, source-backed production correction can be selected for the two q=10 precession failures. The case is (m1,m2)=(60,6), spins (0.2,0.1,0.8) and (0.1,0,0.1), f_ref=20 Hz. The matched-option holdout receipt records TP/TPHM MSA relative L2 0.0202333/0.0225165 against the retained 0.01 gate and numerical PV300/FS4 L2 0.308418/0.307103 against the retained 0.03 gate. MSA mismatch is 0.000200/0.000249 versus 0.0001; PV300/FS4 mismatch is 0.04753/0.04700 versus 0.0005. The strict expected failures in tests/test_phenomt_family_lal.py remain the release gates; none is widened or removed.

Path	Earliest demonstrated discrepancy	Source-backed decision	Exact missing dependency
MSA PV223	The scalar-angle receipt finds 0.0162258 rad relative rotation across 5,326 samples strictly before MECO. Production and independent published-equation scalar angles agree below 3e-10. A separately audited compatibility path drops the public residual to 3.32e-6 rad, beginning with initial-reference spin contractions instead of precession-averaged radiation-reaction contractions; a later q² quadratic-nutation phase convention also changes.	Retain the published equations. Chatziioannou et al., Appendix A, Eqs. (90)–(95) explicitly prescribe precession-averaged couplings. Binary agreement alone does not authorize the contrary substitution.	An independent author specification, derivation, or licensed author artifact for the PV223 implementation convention that explicitly resolves both the flux spin contractions and quadratic-nutation phase convention. Then recheck pre-MECO angles and the frozen whole-strain gate.
Numerical PV300/FS4	The trajectory receipt finds native and independent published RHS/transport agreement below 4.78e-9 rad, but public Euler-derived rotation diverges before merger. The clock/chart receipt finds public carrier frequency agreement within 6.53e-7 relative and minimal-rotation gamma residual 1.70e-11 rad over 2 s, while the reference-event orbital-axis speed is 0.516152/s publicly versus 0.446028/s from the published RHS.	Retain the published spin and Kerr quadrupole terms; output agreement does not identify a term to change. Estellés et al., Sec. II.4.1, Eqs. (12)–(14) specifies the carrier and evolution structure but delegates higher-order spin evolution to SpinTaylor infrastructure; Sturani provides the independent PN spin equations.	An independently documented PV300 precession state/time-map and Euler construction, or author-provided raw v(t), spin vectors and orbital-axis series for this fixed input. Those would separate a hidden precession clock, spin/axis RHS, and alpha/beta chart before changing an equation.

These dependencies are model-compatibility provenance, not a request to fit the public binary. The public outputs expose angles and co-precessing modes, but not the intermediate PV300 state. No LAL implementation source or constants were used, and no waveform equation, coefficient, or whole-strain acceptance threshold is changed by this decision receipt. The detailed component checks and limits remain in the validation report. The scalar-angle receipt retains its convergence and onset summaries while excluding its sampled table. The PV300 receipts retain historical method and execution metadata; their exploratory generators are excluded from the distributed repository. Their recorded commands do not promise a maintained replay workflow. The maintained scalar references and strict whole-waveform tests retain the qualification checks and failures described here.

The maintained q=10 scalar gate in tests/test_clean_room_msa.py::test_q10_msa_flux_spin_couplings_obey_published_scalar_averages now evaluates the six contractions in Appendix A (90)–(95) with 50-digit mpmath arithmetic, taking the separately checked MSA roots and c1 as inputs. The published-equation flux path passes this isolated downstream gate; the initial-reference PV223 compatibility path differs already in the first spin projection by more than 5e-4. This establishes which path follows the published averaging prescription, but cannot identify an undocumented PV223 implementation convention.

The adjacent tests/test_clean_room_msa.py::test_q10_msa_appendix_c_quadratic_phase_diagnostic evaluates Appendix C, Eqs. (116)–(118) directly with 50-digit scalar arithmetic and the same upstream c1 and averaged spin magnitude. For this case the printed Eq. (117) gives psi2=9.7078492509; the limiting-root equation path gives 9.4134154957 and the PV223 compatibility path gives 7.3367234407. Eq. (116) agrees with the equation path at 2e-12, and the q=10 case passes the separate small-velocity cubic-root expansion check in tests/test_precession_msa_pn_clean.py. Both evaluations use the same upstream c1 and averaged spin magnitude; the two factors under Eq. (118)'s square root are negative at this input, so the test uses the positive real root of their positive product. Algebraic expansion of the Appendix B cubic with J² = L² + 2c1/v + S_av² isolates the discrepancy to the second factor of Eq. (118)'s Delta: its printed c1² eta² / delta_m⁴ must be c1² q eta / delta_m⁴ for the factorized cubic discriminant. In source notation, the required increment to that factor is c1² q³(q+2)/(1-q)⁴. Writing F and G for the two printed Eq. (118) factors and B_{-2}, C_{-2}, D_{-2} for the v^-2 coefficients of Appendix B, Eqs. (97)–(99), the cubic gives C_{-2}²-4B_{-2}D_{-2}=(4 delta_m⁴/eta)² F [G+c1² q³(q+2)/(1-q)⁴]. All other terms in Eq. (117) agree identically with the cubic expansion. With this inferred correction, the 50-digit scalar calculation recovers psi2=9.4134154957; it does not recover the PV223 compatibility value. This is an algebraic reconciliation, not a published author erratum or evidence for changing a production coefficient. The independent PV223 phase convention and the frozen whole-waveform gate remain unresolved.

A defensible next experiment needs an author specification of the PV223 convention, or, for PV300, author-provided intermediate state at the same frozen q=10 input.