TD-CARMA: Painless, Accurate, and Scalable Estimates of Gravitational Lens Time Delays with Flexible CARMA Processes
Fit CARMA processes to AGN light curves to estimate gravitational lens time delays. Uses MultiNest for Bayesian inference, to efficiently sample from multimodal posterior distribution of time delay parameter (and CARMA parameters).
- "main_gpcarma.py" : use to fit multiple TD-CARMA(p,q,m) models in parallel (but each individual TD-CARMA model on single core). Recommended for p <=3, q <=2, m <= 4.
- "new_td_gp.py": use to fit single TD-CARMA(p,q,m) model, using multiple cores. Recommended for p >= 4. [run using "mpiexec -n X python3 new_td_gp.py" where X is number of cores. Requires installation of MPI, see installation for pyMultiNest (https://johannesbuchner.github.io/PyMultiNest/)].
Link to the paper: https://iopscience.iop.org/article/10.3847/1538-4357/acbea1