Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Particle physics DL-simulation with control over generated data properties

Karol Rogoziński, Jan Dubiński, Przemysław Rokita, Kamil Deja

TL;DR

The paper tackles the high cost of traditional Monte Carlo simulations for the ALICE Zero Degree Calorimeter (ZDC) by adapting CorrVAE to enable controllable data generation. The authors introduce a three-latent-space architecture with a conditioning encoder to generate ZDC responses from particle properties, while maintaining independence between latent components. They demonstrate controllable generation by mapping latent dimensions to physical properties and show competitive fidelity using physics-aware metrics like Wasserstein distances across multiple channels. The approach offers a rapid, physically plausible alternative for ZDC simulations with explicit control over generated data properties, potentially accelerating analysis and hypothesis testing in high-energy physics.

Abstract

The research of innovative methods aimed at reducing costs and shortening the time needed for simulation, going beyond conventional approaches based on Monte Carlo methods, has been sparked by the development of collision simulations at the Large Hadron Collider at CERN. Deep learning generative methods including VAE, GANs and diffusion models have been used for this purpose. Although they are much faster and simpler than standard approaches, they do not always keep high fidelity of the simulated data. This work aims to mitigate this issue, by providing an alternative solution to currently employed algorithms by introducing the mechanism of control over the generated data properties. To achieve this, we extend the recently introduced CorrVAE, which enables user-defined parameter manipulation of the generated output. We adapt the model to the problem of particle physics simulation. The proposed solution achieved promising results, demonstrating control over the parameters of the generated output and constituting an alternative for simulating the ZDC calorimeter in the ALICE experiment at CERN.

Particle physics DL-simulation with control over generated data properties

TL;DR

The paper tackles the high cost of traditional Monte Carlo simulations for the ALICE Zero Degree Calorimeter (ZDC) by adapting CorrVAE to enable controllable data generation. The authors introduce a three-latent-space architecture with a conditioning encoder to generate ZDC responses from particle properties, while maintaining independence between latent components. They demonstrate controllable generation by mapping latent dimensions to physical properties and show competitive fidelity using physics-aware metrics like Wasserstein distances across multiple channels. The approach offers a rapid, physically plausible alternative for ZDC simulations with explicit control over generated data properties, potentially accelerating analysis and hypothesis testing in high-energy physics.

Abstract

The research of innovative methods aimed at reducing costs and shortening the time needed for simulation, going beyond conventional approaches based on Monte Carlo methods, has been sparked by the development of collision simulations at the Large Hadron Collider at CERN. Deep learning generative methods including VAE, GANs and diffusion models have been used for this purpose. Although they are much faster and simpler than standard approaches, they do not always keep high fidelity of the simulated data. This work aims to mitigate this issue, by providing an alternative solution to currently employed algorithms by introducing the mechanism of control over the generated data properties. To achieve this, we extend the recently introduced CorrVAE, which enables user-defined parameter manipulation of the generated output. We adapt the model to the problem of particle physics simulation. The proposed solution achieved promising results, demonstrating control over the parameters of the generated output and constituting an alternative for simulating the ZDC calorimeter in the ALICE experiment at CERN.
Paper Structure (11 sections, 3 figures, 1 table)

This paper contains 11 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. The simulations of the Zero Degree Calorimeter
  4. Controlling properties in generative modeling
  5. Problem Formulation
  6. Method
  7. Basic adaptation for ZDC simulation
  8. Model modifications
  9. Experiments
  10. Controlling the properties
  11. Conclusions

Figures (3)

  • Figure 1: Model structure presentation. Unlike the classic CorrVAE, the information is encoded in three latent spaces, where the additional encoder encodes particle properties.
  • Figure 2: Comparison of randomly selected simulations generated by different models.
  • Figure 3: Generated images of presented model by traversing two latent variables in $w$ for HEP dataset according to the mask between x, y position and eight-element $w$ vector. (a) Traversing on the $w_1$ that controls x position; (b) Traversing on the $w_2$ that controls y position; (c) Traversing on the $w_1$ and $w_2$ at the same time.