NRTidalv2 Pv2 aligned-carrier receipt — 27 September 2026

Result: The conditional, pre-rotation BNS carrier assembled from a supplied face-on BBH carrier and the published finite-size terms passed in Torch and JAX CPU FP64. This is not a complete precessing BNS waveform gate.

The frozen specification follows Dietrich et al. (2019), Eqs. (3)–(4), (17)–(29), Table II and Sec. III.4. The signed Eq. (24) amplitude is added to the BBH amplitude; the printed tidal plus 2PN, 3PN and 3.5PN self-spin phases then act on that sum. The merger-to-1.2 f_mrg Planck window is applied before rotation. The scalar reference composes the existing independent 80-digit SymPy/mpmath phase, spin, amplitude and taper references. The native Torch/JAX function accepts a supplied BBH carrier and computes every finite-size term without calling that reference.

Three spinning positive-Lambda systems, including exchanged bodies, span 100 Hz, 800 Hz and eight fractions of the printed merger frequency. Native IMRPhenomD supplies the face-on BBH carrier to both the native composition and scalar reference, so this is conditional on that separately validated carrier. At supplied nonzero angles the complete assembled carrier also passes through the native positive-22 projector against the independent scalar projector. Both complex carriers and polarizations meet the frozen 1e-10 normalized limit without phase/time alignment, refitting or output rescaling; post-taper bins are exactly zero. No LAL source or binary residual was used.

Execution: host len, cwd /home/xangma/tmp/diffgw-nrtidalv2-pv2-carrier-20260927, Linux CPU FP64; Python environment /home/xangma/repos/diffgw-worktrees/xas-whole-waveform-gate/.venv/bin/python with Torch 2.14.0+cpu, JAX 0.11.2, mpmath 1.3.0, SymPy 1.14.0 and pytest 9.1.1. Read-only environment and process diagnostics preceded copying the isolated checkout into owned scratch. Other active jobs were left alone. The scoped command was:

PYTHONPATH=src:tests JAX_ENABLE_X64=1 OMP_NUM_THREADS=1 \
/home/xangma/repos/diffgw-worktrees/xas-whole-waveform-gate/.venv/bin/python \
  -m pytest -q tests/test_nrtidalv2_pv2_aligned_carrier_math_reference.py

Result: 6 passed in 10.15 s. A combined run of the new gate and the existing aligned-additive, tidal-mode, projection, paper-D and Pv2 driver tests returned 52 passed, 1 skipped in 18.20 s. The skip is the existing optional LAL/PyCBC check. Both SSH test processes exited; no background job or PID remains to stop.

The helper is not wired to the public Pv2 BNS dispatch. The published insertion order is source-complete, while BNS precession angles, remnant and spline-time conventions, and reference-frequency anchoring remain to be fixed before claiming whole-waveform parity. The Eq. (28)/Fig. 7 Love–Q and printed merger-fit/pinned-binary conflicts remain open. The historical binary_compatible default was not changed.