Calorimetry with deep learning: particle simulation and reconstruction for collider physics

Abstract

Using detailed simulations of calorimeter showers as training data, we investigate the use of deep learning algorithms for the simulation and reconstruction of single isolated particles produced in high-energy physics collisions. We train neural networks on single-particle shower data at the calorimeter-cell level, and show significant improvements for simulation and reconstruction when using these networks compared to methods which rely on currently-used state-of-the-art algorithms. We define two models: an end-to-end reconstruction network which performs simultaneous particle identification and energy regression of particles when given calorimeter shower data, and a generative network which can provide reasonable modeling of calorimeter showers for different particle types at specified angles and energies. We investigate the optimization of our models with hyperparameter scans. Furthermore, we demonstrate the applicability of the reconstruction model to shower inputs from other detector geometries, specifically ATLAS-like and CMS-like geometries. These networks can serve as fast and computationally light methods for particle shower simulation and reconstruction for current and future experiments at particle colliders.

GANplifying event samples

“…Fast precision simulation in particle physics starts with phase space integration [9,10], phase space sampling [11][12][13], and amplitude networks [14,15]. Especially interesting are NN-based event generation [16][17][18][19][20], event subtraction [21], detector simulations [22][23][24][25][26][27][28][29][30], or fast parton showers [31][32][33][34]. Deep generative models can also improve searches for physics beyond the Standard Model [35] or anomaly detection [36,37].…”

Towards a computer vision particle flow

“…The training and the evaluation of the NN models are performed on cells belonging to the topoclusters. Figure 4 shows a comparison between the predicted and truth cell energies for the Graph model for an inclusive energy range of [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] GeV. It illustrates the ability of the ML models to predict the cell fractions over a wide range of energy.…”

“…Image-based convolutional NN have been used to study calorimeter shower and also used for hadronic jet tagging, as can be found Refs. [11][12][13][14]. These approaches are based on uniform two or three-dimensional (2D or 3D) images.…”

Studying the Potential of Graphcore® IPUs for Applications in Particle Physics

“…Applications of GANs within particle physics are constantly appearing. GANs have been applied in both event generation [37][38][39][40][41][42][43] and detector modelling [44][45][46][47][48][49][50][51][52]. In this section the inference and training speeds of some of these particle physics based GANs are assessed on the IPU hardware and compared to results on the GPU and CPU described in Table 1.…”