Paper detail

$\it COD:$ An Algorithm for Shape Reconstruction of Transiting Celestial Bodies through Topological Optimization

We introduce a novel algorithm, $\textit{COD}$ -- Compact Opacity Distribution, for shape reconstruction of a celestial body that has been observed to occult a star, using the photometric time-series observations of the occultation. $\textit{COD}$ finds a solution to the light-curve inversion problem for an optically thick occulter having an approximately convex shape, together with an estimate of its size, impact parameter and velocity, relative to the occulted star. The algorithm is based on an optimization scheme that uses topological constraints and an objective function for the geometry of the occulter. The constraints of the problem follow linear relations, which enable the use of linear programming optimization as the mathematical framework. Multiple tests of the algorithm were performed, all of which resulted in high correlations between the simulated and obtained shapes of the occulting objects, with errors within $5\%$ in their projected velocities and horizontal sizes, and within $0.1$ in their impact parameters. These tests include a video of a solar eclipse by Phobos, as seen by NASA's Curiosity rover, which was collapsed into its corresponding light curve and reconstructed afterwards. We applied $\textit{COD}$ to the mysterious case of VVV-WIT-08 -- a single deep occultation ($\sim 96 \%$) of a giant star lasting for over 200 days. The analysis, which did not assume any specific shape of the occulter, suggested an object with a projected opacity distribution resembling an ellipse with an eccentricity of $\sim 0.5$, tilted at $\sim 30$ degrees relative to the direction of motion, with a semi-minor axis similar to the stellar radius.