A cross-power track test finds no excess 2 to 10 seconds after GW170817 and gives waveform-dependent recovery

We compare a cross-power phase test with minimum detector power for four-second tracks 2 to 10 seconds after GW170817. No signal exceeds an empirical noise threshold. Cross-power ranks are 0.239 and 0.598 with early and late noise estimates. Thresholds use 116 calibration intervals. Recovery uses 117 separate intervals. At equal detector hrss of 1e-22 strain sqrt(s), a falling-frequency signal is recovered in 35 cases with cross power and 28 with minimum power. The paired 95 percent gain interval is -2.6 to +14.5 percentage points. Amplitude imbalance and frame phase changes alter recovery. Held-out false alarms differ between methods, so equal false-alarm superiority is not shown. These are restricted method tests, not astrophysical limits or a new source. Code, source hashes, figures, aggregate results and retrieval steps are included. Raw strain remains external.

astronomy · GW170817 · cross power · phase consistency · injection recovery · held-out noise · null result