In a joint fit with motion terms, the Sector 98 loss centroids of TOI-941.02 and b differ by 0.039 pixel

We tested the pixel source of the TOI-941.02 dip in TESS Sector 98. We fitted pixel flux with separate templates for the extra dip and the overlapping 8.51457-day signal b. The planned test compared the extra-dip loss image with the mean of three clean b images. In the eight-pixel SPOC aperture, this gives a difference of (+0.038, -0.128) pixel, or 0.134 pixel in distance. The clean b row centroids change by 0.270 pixel across the sector, alongside a 0.292 pixel change in recorded row motion. A residual bootstrap of the fixed mean reference does not include this change. We therefore added a local comparison of both signals from the same joint fit. With quadratic time and recorded motion terms, their centroid difference is (-0.015, -0.035) pixel, or 0.039 pixel in distance. Two-hour residual blocks give conditional standard errors of (0.053, 0.061) pixel. Thus this fit does not resolve an offset. Six local SPOC fits with motion terms give distances from 0.032 to 0.040 pixel. A fit with shapes measured from aperture flux gives 0.0275 pixel. Fifteen artificial overlap tests recover the imposed images; errors in their local centroid differences range from 0.005 to 0.084 pixel. The result adds a pixel source test for a known candidate and shows the effect of a moving reference. The fit does not confirm a planet, locate the earlier Sector 5 event, or exclude unresolved blends. The package preserves the planned and added tests, selected public pixels, hashes, code, figures, and all results. Repeated downloads give identical bytes, although the public FITS data-extension checksums fail.

TOI-941.02 · TOI-941 b · TIC 408137826 · TESS Sector 98 · difference images · pixel source test · pointing drift · overlapping transits · exoplanet candidate · reproducible astronomy