A 401-sample c=6 biweight filter reduces the fitted depth of Boyajian's 2019 TESS dip below 0.5%

We test whether a 401-sample biweight filter retains broad dimming in public TESS data for Boyajian's star. The test uses an explicit c=6 location estimate and five SPOC sectors at 120-second cadence. For the known 0.875-day event at BTJD 1730.164, a fixed trapezoid fit gives 1.075% depth before filtering. After filtering, the fitted depth is 0.155%, which retains 14.4% of the input coefficient. A cosine-squared fit gives 1.398% before filtering and 0.209% after filtering. A 1601-sample biweight window retains a trapezoid depth of 1.006%. We also run 1920 paired injection trials with two shapes and three depths. For 1% inputs, median transmission at 0.875 day is 25.7% for the cosine and 19.9% for the trapezoid. At five days, the corresponding medians are 1.05% and 1.17%. These finite results show that the tested short filter can hide broad dimming at the percent level. They do not measure search completeness or reproduce code that the source paper does not supply. The event was reported before this study. Its cause remains open. The signed package contains original archive data, hashes, code, all trials, figures, and offline reproduction instructions.

TESS · Boyajian star · KIC 8462852 · TIC 185336364 · biweight filter · detrending · signal retention · injection tests · public data