Extended Data Fig. 6: Detailed fits and residuals of eclipses and overall HiPERCAM light curve.
From: A dense 0.1-solar-mass star in a 51-minute-orbital-period eclipsing binary

a, Our best-fit model of the primary eclipse, with the model shown as a solid black line, the binned data as red points, and the model without an eclipse as the dashed black line. In this figure, a linear polynomial component (of the functional form y = a × t, where t is the time from mid-eclipse (we applied separate corrections for the primary and secondary), has been subtracted out of both the model and the data for better visualization. We constructed the model by simultaneously fitting data around the primarily eclipse as well as data from the secondary eclipse (panel c). b, A best-fit of the eclipse-derived model constructed from the eclipse data shown in panels a and c to the full HiPERCAM light curve. While the model roughly reproduces the correct amplitude of ellipsoidal variations, it does not to fully capture the structure seen in the full light curve. c, Our best-fit model to the secondary eclipse, with a linear correction subtracted out of both the data and models. d, The residuals of the best fit of the eclipse-derived model to the full dataset shown in panel b. As is readily apparent, the strongest residuals occur out-of-eclipse, and are likely to arise because of a combination of effects from the accretion disk, and an O’Connell effect associated with the donor.