Using machine learning to disentangle LHC signatures of Dark Matter candidates

Charanjit Kaur Khosa,Michael Soughton,Veronica Sanz

doi:10.21468/scipostphys.10.6.151

Charanjit Kaur Khosa, Michael Soughton + Show 1 more

Open Access

https://doi.org/10.21468/scipostphys.10.6.151

Copy DOI

Abstract

We study the prospects of characterising Dark Matter at colliders using Machine Learning (ML) techniques. We focus on the monojet and missing transverse energy (MET) channel and propose a set of benchmark models for the study: a typical WIMP Dark Matter candidate in the form of a SUSY neutralino, a pseudo-Goldstone impostor in the shape of an Axion-Like Particle, and a light Dark Matter impostor whose interactions are mediated by a heavy particle. All these benchmarks are tensioned against each other, and against the main SM background (Z+jets). Our analysis uses both the leading-order kinematic features as well as the information of an additional hard jet. We explore different representations of the data, from a simple event data sample with values of kinematic variables fed into a Logistic Regression algorithm or a Fully Connected Neural Network, to a transformation of the data into images related to probability distributions, fed to Deep and Convolutional Neural Networks. We also study the robustness of our method against including detector effects, dropping kinematic variables, or changing the number of events per image. In the case of signals with more combinatorial possibilities (events with more than one hard jet), the most crucial data features are selected by performing a Principal Component Analysis. We compare the performance of all these methods, and find that using the 2D images of the combined information of multiple events significantly improves the discrimination performance.

Highlights

Principal Component Analysis (PCA) correlations pTj2 η j1 η j2 ∆φ j1 j2 MET ∆φMj2ETEFT PCA correlations η j1-0.00 -0.00 0.00 0.71 -0.03 -0.71 0.00 -0.00 η j2SUSY1, Mχ10 = 100 GeVackground pTj1 (GeV) PCA correlationsSUSY2, Mχ10 = 200 GeV PCA correlationsWe show here results for signal to background classification
This difficult task calls for the use of Machine Learning (ML), which is emerging as a powerful tool for New Physics searches at Large Hadron Collider (LHC)
Because images are constructed from highly processed information (PDFs are already an abstract concept) we find no sizeable improvement respect to the Deep Neural Networks (DNN) results

Summary

Introduction

After the Higgs boson discovery [1,2] at the Large Hadron Collider (LHC), some of the discovery potential focus has shifted towards Dark Matter (DM) searches. The challenge of disentangling different DM candidates would likely benefit from the implementation of new and more sophisticated techniques, beyond the conventional search strategies This difficult task calls for the use of Machine Learning (ML), which is emerging as a powerful tool for New Physics searches at LHC (see e.g. the recent review by Radovic et al [8] and references therein). We will use as benchmarks of comparison three types of models: a heavy WIMP dark matter from SUSY [23], Axion-Like Particles (ALPs) [24, 25] and a simplified DM model with a spin mediator These will provide enough variety of characteristics to analyse differences and degeneracies among models.

Description of Benchmark Models

Kinematic Distributions

DM Characterization using Machine Learning

Conclusions and Outlook

Findings

Background