The epochalypse catalog: simulated Gaia DR4-like epoch astrometry for substellar-companion orbit-modeling benchmarks

Simulated Gaia DR4-like epoch astrometry for testing orbit-modeling pipelines on a common set of injected systems with known parameters. Onto each of the 5,040,328 main-sequence Gaia stars within 500 pc in the G23H catalog we inject a zero-, one-, and two-companion system, for 15,120,984 systems in total. Companion masses and semimajor axes are drawn log-uniformly over 1 M_Mars to 80 M_Jup and 0.03 to 100 au, with no occurrence model and no detectability cut, so the companion-free population serves as a matched null. The one- and two-companion populations each carry a high-SNR sub-catalog (every companion with SNR_tot >= 5: 439,503 and 26,379 systems). Every system has been characterized with the kepmodel astrometric periodogram on a 15,777-period grid; the thresholds are calibrated on the companion-free population to a 1% false-positive rate. This record accompanies Yahalomi et al. (2026), "Preparing for the Gaia DR4 Epochalypse: Simulated Epoch Astrometry for Substellar Companion Orbit-Modeling Benchmarks". The generator, the periodogram characterization, the paper-figure notebooks and an orbit-fitting tutorial (NumPyro + jaxoplanet) are at https://github.com/dyahalomi/epochalypse. Files The record is flat; download_catalog.sh (included, also at catalog/download_catalog.sh in the repository) fetches it into the data/ and periodograms/ layout the notebooks and the generator expect, and unpacks the epoch shards on request (--shards 1_companion, --shards all). MD5SUMS.txt lists an MD5 for every file; README.md describes the layout. Stellar sample and injected truth (data/): - stars.csv: the 5,040,328-star parent sample: G23H astrometry, Gaia DR3 photometry and astrometric metadata, Hipparcos-Gaia proper motions, the per-star noise-model terms, and the interpolated main-sequence mass and radius- stars_index.parquet: gaia_source_id to row number in stars.csv- scanlaw_index.parquet: gaia_source_id to the offset and length of the star's block of predicted DR4 transits in the generator's scanning-law table (the per-epoch times, scan angles and parallax factors are in the epoch shards)- injected_solutions_{0,1,2}_companion.parquet: the injected truth for every system: seeds, astrometric parameters, n_transits_dr4, and per companion sma, ecc, mass, inc, Omega, omega, M_anom, period, alpha_mas, snr_single, snr_eff, snr_total (plus coplanarity and period-ratio flags for two companions)- injected_solutions_{1,2}_companion_high_snr.parquet: the high-SNR selections of the above- simulated_astrometry_{0,1,2}_companion.tar: uncompressed tar of 640 parquet shards per population: 320 epochs shards (one row per along-scan epoch: obs_time_tcb, scan_pos_angle, parallax_factor_al, centroid_pos_al, centroid_pos_error_al, field_of_view) and the 320 matching truths shards; unpacks to data/simulated_astrometry/<population>/ Periodogram characterization (periodograms/): - characterization_{0,1,2}_companion.parquet and the two _high_snr tables: the kepmodel search result per system: five-parameter chi2, best period, class (undetected / unimodal / multimodal / broad), peak powers, delta-BIC, acceleration test, calibrated detected and period_reliable flags, FAP- period_grid.parquet: the 15,777 trial periods- calibration.json: the orbit and acceleration thresholds, calibrated on the companion-free population to a 1% false-positive rate and applied to all three- manifest.json: the run's configuration and software version Other: - README.md: the file layout, how to unpack and read the shards, citation- MD5SUMS.txt: an MD5 for every file in the record- download_catalog.sh: fetches the record into the data/ and periodograms/ layout, with --shards to unpack the epoch shards