Jacobi Constant Time Series for One Million Cislunar Orbits Derived via Volunteer Computing

Dataset DescriptionThis dataset provides the Jacobi constant time series for one million unique cislunar orbits, derived from the public "One Million Open-source Cislunar Orbits" dataset generated by Lawrence Livermore National Laboratory (LLNL) (Yeager et al., 2025). The original orbits were numerically integrated for six years within the Earth-Moon circular restricted three-body problem (CR3BP) framework, using a mass parameter of μ = 0.01215. The trajectories are stored in the Earth-Moon rotating frame, with positions in meters and velocities in meters per second. To enable large-scale stability analysis, we recomputed the Jacobi constant for every time step of each orbit—over 16 billion individual values in total—by applying the CR3BP energy integral. The computation was distributed across thousands of volunteer devices worldwide using the BOINC (Berkeley Open Infrastructure for Network Computing) platform. The application code used for this processing is archived separately (Gao, 2026).The data are organized into 1,000 ZIP archives, each containing 20 JSON files. Every JSON file corresponds to a batch of 50 consecutive orbits. File naming follows the pattern orb_id_XXXXX_to_XXXXX_output.json (or .zip for the archives), where XXXXX is the starting orbit ID without leading zeros.Each JSON file includes the following fields:- orbit_results: a dictionary keyed by orbit ID, each entry containing the jacobi_constants list (float64) for that orbit, along with optional classification metadata (e.g., major_class) based on the classical energy thresholds of the CR3BP.- global_jacobi_statistics: summary statistics (min, max, mean, std, range) for the entire batch.- execution_summary: processing time and status information.The total uncompressed dataset size is approximately 300 GB, comprising 16,057,065,559 data points.A supplementary archive charts.zip is provided, containing:- histogram_all_jacobi.png – a histogram of all Jacobi constants.- region_iv_(orbit_754482)_jacobi.png – the Jacobi constant time series for the single orbit classified as Region IV (orbit ID 754482).- orbit_XXXX_jacobi.png – individual Jacobi constant versus point index plots for each of the 173 orbits whose maximum Jacobi constant exceeds 100. The filename includes the orbit ID (e.g., orbit_1791_jacobi.png).Temporal and spatial contextEach orbit spans six simulated years, but the original time stamps are not included; the Jacobi constant values are given in the same temporal order as the original position/velocity time series. The underlying trajectories reside in the Earth-Moon rotating frame, with the characteristic length of 3.843856e8 m and characteristic velocity of 1.022e3 m/s used for non-dimensionalization where applicable. All Jacobi constants reported here are dimensionless.Data quality and completenessAll 1,000,000 orbits were successfully processed; there are no missing records. The calculations were performed in double precision (float64). The numerical integration error inherent in the original LLNL dataset may cause slight variations in the Jacobi constant along an orbit; however, we performed a redundant conservation check (with a tolerance of 1e-12) on a random sample of points and confirmed that the computed values are consistent with the underlying trajectories. The maximum Jacobi constant encountered exceeds 6e5, corresponding to high-energy (those which have Jacobi Constants of more than 100.0) orbits; these values have been verified independently.File format and accessibilityThe data are stored in standard JSON format, which can be read by any programming language (Python, R, MATLAB, etc.) and by many text editors. No specialized software is required. Each JSON file is self-contained and can be processed independently. For users who prefer to work with the compressed archives, standard tools such as WinRAR, 7-Zip, or the Python zipfile module can extract the files.Scientific value and reuse potentialThis dataset enables large-scale statistical analysis of cislunar orbital stability, including the identification of rare orbit families (e.g., the sole Region IV orbit) and the characterization of energy distributions across one million trajectories. It serves as a benchmark for machine learning models aimed at predicting orbital behavior and for validating simplified dynamical models. The data are fully open and reusable under the Creative Commons Attribution 4.0 International (CC BY 4.0) license, with proper attribution to both the original LLNL dataset and this derived work.