Neural network analysis of X-ray polarimeter data

TL;DR

This paper introduces deep neural network techniques to improve the analysis of X-ray polarimeter data, enhancing sensitivity and accuracy in measuring source polarization by classifying events and estimating parameters.

Contribution

The work presents novel neural network methods for event classification and parameter estimation in X-ray polarimetry, outperforming standard analysis techniques.

Findings

< 0.75x improvement in minimum detectable polarization

Neural networks effectively classify events and estimate polarization parameters

Enhanced sensitivity in X-ray polarimeter data analysis

Abstract

This chapter presents deep neural network based methods for enhancing the sensitivity of X-ray telescopic observations with imaging polarimeters. Deep neural networks can be used to determine photoelectron emission directions, photon absorptions points, and photon energies from 2D photoelectron track images, with estimates for both the statistical and model uncertainties. Deep neural network predictive uncertainties can be incorporated into a weighted maximum likelihood to estimate source polarization parameters. Events converting outside of the fiducial gas volume, whose tracks have little polarization sensitivity, complicate polarization estimation. Deep neural network based classifiers can be used to select against these events to improve energy resolution and polarization sensitivity. The performance of deep neural network methods is compared against standard data analysis methods, revealing a < 0.75x improvement in minimum detectable polarization for IXPE-specific simulations. Potential future developments and improvements to these methods are discussed.