A new approach to handling factorial moment correlations through principal component analysis

TL;DR

This paper introduces a novel method using Principal Component Analysis to improve intermittency analysis of factorial moments in high-energy collision data, effectively handling scale correlations without data subdivision.

Contribution

It presents a new PCA-based approach to address correlations in factorial moments analysis, enhancing the detection of power-law scaling in high-energy physics data.

Findings

PCA effectively manages scale correlations in factorial moments analysis.

The method improves the reliability of intermittency analysis in critical point searches.

It eliminates the need for data subdivision in scale correlation handling.

Abstract

Intermittency analysis of factorial moments is a promising method used for the detection of power-law scaling in high-energy collision data. In particular, it has been employed in the search of fluctuations characteristic of the critical point (CP) of strongly interacting matter. However, intermittency analysis has been hindered by the fact that factorial moments measurements corresponding to different scales are correlated, since the same data are conventionally used to calculate them. This invalidates many assumptions involved in fitting data sets and determining the best fit values of power-law exponents. We present a novel approach to intermittency analysis, employing the well-established statistical and data science tool of Principal Component Analysis (PCA). This technique allows for the proper handling of correlations between scales without the need for subdividing the data sets available.