Viterbi decoding of CRES signals in Project 8

Abstract

Cyclotron radiation emission spectroscopy (CRES) is a modern approach for determining charged particle energies via high-precision frequency measurements of the emitted cyclotron radiation. For CRES experiments with gas within the fiducial volume, signal and noise dynamics can be modelled by a hidden Markov model. We introduce a novel application of the Viterbi algorithm in order to derive informational limits on the optimal detection of cyclotron radiation signals in this class of gas-filled CRES experiments, thereby providing concrete limits from which future reconstruction algorithms, as well as detector designs, can be constrained. The validity of the resultant decision rules is confirmed using both Monte Carlo and Project 8 data.

Deep learning based event reconstruction for cyclotron radiation emission spectroscopy

“…Taking the ratio of the effective trapping volumes between minimum pitch angles 89.33 • and 89 • , we expect only about 44% of all simulated events to be visible (detectable) after trapping conditions are satisfied. Furthermore, following theoretical modeling of CRES signals with the Viterbi algorithm [41], the minimal reconstructable track length given Phase II conditions is approximately 3 pixels, diminishing the percentage of visible electrons by 59%. Finally, making a cut at the minimal Phase II reconstructable SNR of 4 as a realistic approximation, we keep only 76% of this fraction of events.…”