Optimal operation of cryogenic calorimeters through deep reinforcement learning

Abstract

Cryogenic phonon detectors with transition-edge sensors achieve the best sensitivity to light dark matter-nucleus scattering in current direct detection dark matter searches. In such devices, the temperature of the thermometer and the bias current in its readout circuit need careful optimization to achieve optimal detector performance. This task is not trivial and is typically done manually by an expert. In our work, we automated the procedure with reinforcement learning in two settings. First, we trained on a simulation of the response of three CRESST detectors used as a virtual reinforcement learning environment. Second, we trained live on the same detectors operated in the CRESST underground setup. In both cases, we were able to optimize a standard detector as fast and with comparable results as human experts. Our method enables the tuning of large-scale cryogenic detector setups with minimal manual interventions.

23 pages, 14 figures, 2 tables

References in corpus (13)

• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• Soft Actor-Critic Algorithms and Applications
• OpenAI Gym
• First results from the CRESST-III low-mass dark matter program
• Results on light dark matter particles with a low-threshold CRESST-II detector
• Optimizing Quantum Error Correction Codes with Reinforcement Learning
• A Microcalorimeter and Bolometer Model
• Results on sub-GeV Dark Matter from a 10 eV Threshold CRESST-III Silicon Detector
• Testing spin-dependent dark matter interactions with lithium aluminate targets in CRESST-III
• Latest observations on the low energy excess in CRESST-III
• First measurements of remoTES cryogenic calorimeters: easy-to-fabricate particle detectors for a wide choice of target materials
• Towards an automated data cleaning with deep learning in CRESST
• Generation of Noise Time Series with arbitrary Power Spectrum