Detection is truncation: studying source populations with truncated marginal neural ratio estimation

Noemi Anau Montel and Christoph Weniger

Submitted on 8 November 2022


Statistical inference of population parameters of astrophysical sources is challenging. It requires accounting for selection effects, which stem from the artificial separation between bright detected and dim undetected sources that is introduced by the analysis pipeline itself. We show that these effects can be modeled self-consistently in the context of sequential simulation-based inference. Our approach couples source detection and catalog-based inference in a principled framework that derives from the truncated marginal neural ratio estimation (TMNRE) algorithm. It relies on the realization that detection can be interpreted as prior truncation. We outline the algorithm, and show first promising results.


Comment: Accepted for the NeurIPS 2022 workshop Machine Learning and the Physical Sciences; 8 pages, 3 figures

Subjects: Astrophysics - Instrumentation and Methods for Astrophysics; Astrophysics - Cosmology and Nongalactic Astrophysics; Astrophysics - High Energy Astrophysical Phenomena


Illustration of our inference framework (see~\cref{sec:method} for details). \textit{Left panel}: A source detection network $r_1$ and the corresponding sensitivity network $r_2$ are trained based on the full simulation model. \textit{Right panel}: Bright sources are constrained in the truncated simulation model, while sub-threshold sources vary freely. Two inference networks are trained to capture information from sub-threshold sources ($r_3$) and detected sources ($r_4$).