A 30-day period limit removes three apparent 5% TESS matches to the GJ 9103 A RV signal with a constant flux offset

We tested public TESS photometry for rotation warnings in a fixed list of 57 unassigned HARPS radial-velocity (RV) signals on 52 hosts. An exact join with the default TESS All-Sky Rotation Survey (TARS) catalog gives two adopted periods. Each uses one sector; neither matches an RV period at tolerances from 1% to 10%. No candidate passes our rule for a clean catalog period from more than one sector. We then searched all 70 returned TARS light curves for the hosts of 20 RV signals with periods at or below 40 days. The files cover 19 signals on 17 hosts. We used dominant Lomb-Scargle peaks from 0.1 to 20 days, with a floating mean and separately after linear trend removal. For the 38.420921-day GJ 9103 A RV signal, three sectors appear to match twice the photometric period within 5% under both choices. All three peaks reach the 20-day upper limit. Raising the limit to 30 days moves the constant-offset peaks to 20.58, 27.72, and 27.57 days. All three 5% matches then fail; no sector matches under both processing choices. One of the three remains a match at a 10% tolerance. A direct sine-and-cosine least-squares calculation checks the numerical result. Across all 20 short-period signals, none has two non-edge sector matches under both processing choices. This audit shows how a shared search limit can produce repeated apparent period matches in these products. It does not determine a rotation period, identify the RV cause, or confirm or reject a planet. We preserve the source tables, exact archive query, all selected light curves, hashes, code, figures, results, and reproduction instructions.

TESS · HARPS · TARS · radial velocity · stellar rotation · period search · search boundary · GJ 9103 A · public data · reproducible research