Every star that hosts a planet performs an intricate dance, wobbling ever so slightly as gravitational forces tug it back and forth. These minute radial velocity variations—often just meters per second—are among our most powerful tools for discovering worlds beyond our solar system. Yet extracting planetary signals from noisy stellar observations requires sophisticated statistical analysis that can distinguish genuine orbital signatures from instrument artifacts and stellar activity.
KIMA (Kinematic Inference for Massive Astrophysical data) tackles this challenge head-on with a robust C++ framework built around Bayesian inference methods. The toolkit employs advanced Markov Chain Monte Carlo algorithms to model radial velocity time series, automatically determining the optimal number of planetary companions while accounting for observational uncertainties and correlated noise. Its hierarchical modeling approach can simultaneously fit multiple datasets from different instruments, extract orbital parameters with rigorous uncertainty quantification, and detect planets that would be lost in traditional periodogram analysis.
Developed for professional astronomers and researchers, KIMA has become an essential tool in exoplanet surveys worldwide, contributing to discoveries that expand our census of nearby worlds. Its modular architecture allows researchers to customize models for specific observational scenarios, from hot Jupiters with obvious signals to Earth-mass planets requiring years of precise measurements to confirm.
⭐ Stars: 11
💻 Language: C++
🔗 Repository: kima-org/kima