Algorithms for Identification of Nearly-Coincident Events in Calorimetric Sensors

Area of Science:

• Experimental physics
• Particle physics
• Signal processing

Background:

• High arrival rates in experiments necessitate reliable identification of nearly-coincident events.
• Unidentified pulse pile-ups are a primary error source in calorimetric neutrino mass measurements (e.g., HOLMES).
• Existing Wiener filtering methods struggle with pulse-shape variations and deconvolution instability.

Purpose of the Study:

• To develop an advanced processing method for accurately detecting nearly-coincident events.
• To overcome limitations of traditional pile-up identification techniques.
• To improve the precision of neutrino mass measurements.

Main Methods:

• Exploited singular value decomposition (SVD) for signal processing.
• Developed a method to separate single-pulse from piled-up records in training data.
• Constructed a model of single-pulse records accounting for amplitude, arrival time, and baseline variations.

Main Results:

• Successfully separated single-pulse and piled-up event records.
• Created a robust model for single-pulse shapes under varying conditions.
• Demonstrated the capability to detect nearly-coincident events with high accuracy.

Conclusions:

• The SVD-based processing method significantly enhances pile-up identification.
• This advancement can relax detector and readout system performance requirements.
• Enables larger sensor arrays and tighter constraints on neutrino mass measurements.