Controlled orbital modification of planetary-scale bodies within chaotic multi-body environments demands simulation engines capable of managing non-conservative perturbations with sub-meter numerical stability. This work—the fourth installment of the Stellar Death Clock framework—presents a rigorous data science pipeline designed to automate the calibration of the three-dimensional non-gravitational force tensor (Marsden et al., 1973) without relying on traditional manual parameter tuning. We address the limitations of standard N-body algorithms (such as REBOUND utilizing its adaptive 15th-order Gauss-Radau core IAS15) when tracking Small Solar System Bodies (SSSBs) subjected to severe outgassing or radiation pressure. To mitigate catastrophic prediction drift, we formalize a temporal cross-validation algorithm that systematically maps the bias-variance tradeoff across variable observational windows. Using telemetry of the near-Earth asteroid (99942) Apophis compiled up to June 1st, 2026, the framework isolates an optimal 15-year historical data window. This calibration window minimizes short-term observational noise overfitting while avoiding long-term physical data-drift, yielding a high-fidelity prediction of the critical flyby on April 13th, 2029. Furthermore, we stress-test the computational limits of this architecture under extreme retrograde hyperbolic boundary conditions using real-world telemetry from the interstellar object 3I/ATLAS. By coupling the optimized tensor with a precise barycentric-to-heliocentric state vector transformation, the pipeline achieves an absolute sub-meter convergence floor (0.000049 km) against NASA/JPL Horizons ephemerides over an annual tracking cycle. These results validate a platform-independent, verified digital twin architecture capable of guiding long-term, high-frequency multi-body orbital control tasks with absolute dynamical sensitivity.
Moisés Frutos Plaza (Sun,) studied this question.