Coherence Metric Validation on Swift-BAT Light Curve Data: Negative Control Benchmark for Noise-Dominated Time Series

Coherence Metric Validation on Swift-BAT Light Curve Data V1.1 This publication has been withdrawn following retrospective methodological review. The original work was presented as a negative-control validation of the EOS coherence framework using the Swift-BAT light curve of QSO B0003-066. Subsequent reanalysis identified several issues that materially undermine that validation claim: • some headline statistics were calculated on the full 6,392-observation dataset despite the methodology stating that only the 6,201 DATA_FLAG = 0 observations were analyzed; • the reported normalized dispersion used the mean measurement error as its denominator rather than a quadrature/RMS treatment appropriate to heterogeneous per-observation uncertainties, materially changing the central value; • rolling windows described as 30-day windows were implemented as 30-observation windows on an irregularly sampled series, and several reported durations in “days” were observation counts rather than calendar durations; • most importantly, a single low-signal astrophysical time series does not constitute a known-null control sufficient to establish false-positive performance, specificity, robustness, or framework validation. These issues undermine the central claim of this publication as a negative-control validation study. The withdrawal does not assert that the underlying Swift-BAT dataset is invalid, nor that all descriptive observations from the analysis are invalid. Rather, the analysis design does not support the validation conclusions originally drawn from it. The author therefore withdraws this publication rather than attempting to preserve a validation claim that the available evidence cannot support.