FlameNEST: Explicit Profile Likelihoods with the Noble Element Simulation Technique

TL;DR

FlameNEST introduces an efficient, data-driven likelihood evaluation framework for noble element detectors, enabling statistical analysis without extensive Monte Carlo simulations, thus advancing dark matter research.

Contribution

It provides an explicit likelihood evaluation method using analytic probability elements convolved in TensorFlow, reducing reliance on costly simulations and facilitating collaborative analyses.

Findings

Enables event-by-event likelihood evaluation

Reduces computational costs compared to Monte Carlo methods

Supports collaborative data analysis in dark matter experiments

Abstract

We present FlameNEST, a framework providing explicit likelihood evaluations in noble element particle detectors using data-driven models from the Noble Element Simulation Technique. FlameNEST provides a way to perform statistical analyses on real data with no dependence on large, computationally expensive Monte Carlo simulations by evaluating the likelihood on an event-by-event basis using analytic probability elements convolved together in a single TensorFlow multiplication. Furthermore, this robust framework creates opportunities for simple inter-collaborative analyses which will be fundamental for the future of experimental dark matter physics.