Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CaloQVAE : Simulating high-energy particle-calorimeter interactions using hybrid quantum-classical generative models

Sehmimul Hoque, Hao Jia, Abhishek Abhishek, Mojde Fadaie, J. Quetzalcoatl Toledo-Marín, Tiago Vale, Roger G. Melko, Maximilian Swiatlowski, Wojciech T. Fedorko

TL;DR

The paper addresses the computational bottleneck of simulating high-energy particle showers in calorimeters for HL-LHC analyses. It introduces CaloQVAE, a hybrid quantum-classical generative model that uses a D-Wave quantum annealer to sample the RBM-based latent space, replacing expensive block Gibbs sampling and enabling faster generation of calorimeter showers while maintaining fidelity to GEANT4-based data. The approach builds on CaloDVAE by incorporating a Chimera RBM mapped to an Ising Hamiltonian for quantum sampling, aided by an iterative $eta_{ ext{eff}}^{*}$ temperature calibration to align quantum and classical distributions. Results show that CaloQVAE reproduces key shower-shape distributions and energy conditioning with fidelity close to GEANT4, and that the quantum sampler offers a potential speedup due to the ~20 μs anneal time, suggesting practical viability for real-time or large-scale simulations in collider experiments.

Abstract

The Large Hadron Collider's high luminosity era presents major computational challenges in the analysis of collision events. Large amounts of Monte Carlo (MC) simulation will be required to constrain the statistical uncertainties of the simulated datasets below these of the experimental data. Modelling of high-energy particles propagating through the calorimeter section of the detector is the most computationally intensive MC simulation task. We introduce a technique combining recent advancements in generative models and quantum annealing for fast and efficient simulation of high-energy particle-calorimeter interactions.

CaloQVAE : Simulating high-energy particle-calorimeter interactions using hybrid quantum-classical generative models

TL;DR

The paper addresses the computational bottleneck of simulating high-energy particle showers in calorimeters for HL-LHC analyses. It introduces CaloQVAE, a hybrid quantum-classical generative model that uses a D-Wave quantum annealer to sample the RBM-based latent space, replacing expensive block Gibbs sampling and enabling faster generation of calorimeter showers while maintaining fidelity to GEANT4-based data. The approach builds on CaloDVAE by incorporating a Chimera RBM mapped to an Ising Hamiltonian for quantum sampling, aided by an iterative temperature calibration to align quantum and classical distributions. Results show that CaloQVAE reproduces key shower-shape distributions and energy conditioning with fidelity close to GEANT4, and that the quantum sampler offers a potential speedup due to the ~20 μs anneal time, suggesting practical viability for real-time or large-scale simulations in collider experiments.

Abstract

The Large Hadron Collider's high luminosity era presents major computational challenges in the analysis of collision events. Large amounts of Monte Carlo (MC) simulation will be required to constrain the statistical uncertainties of the simulated datasets below these of the experimental data. Modelling of high-energy particles propagating through the calorimeter section of the detector is the most computationally intensive MC simulation task. We introduce a technique combining recent advancements in generative models and quantum annealing for fast and efficient simulation of high-energy particle-calorimeter interactions.
Paper Structure (5 sections, 1 equation, 3 figures, 1 table)

This paper contains 5 sections, 1 equation, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Model Setup
  3. CaloQVAE Training and Inference
  4. Performance Evaluation
  5. Conclusion

Figures (3)

  • Figure 1: Histogram showing the probability of obtaining sample $s$ with Ising Energy computed by $E(s) = s^{T}Js+h^{T}s$ in dimensionless units for a trained CaloQVAE model. The energy distribution of QPU samples (solid line) are very close to the classical samples (shaded) after accounting for the $\beta_{eff}^{*}$ factor.
  • Figure 2: Comparison of selected shower shape variables for three clusters ($\pi^+$, $\gamma$, and $e^+$). The rows of sub-panels show the distributions of three key variables: (1) Total Energy: The total energy observed across all layers. (2) Energy Fraction: The fraction of energy deposited in the layer 1 (middle layer) compared to the total energy in all layers. (3) Sparsity: The ratio of the number of hits in the layer 1 to the total number of hits across all layers. Each sub-panel illustrates the close alignment of results from the GEANT4, DVAE (classical), and QVAE (quantum) models.
  • Figure 3: Comparison of energy spectra between CaloDVAE (classical) and CaloQVAE (quantum) models against GEANT4 test data for incident particle energies ranging from 45 to 60 GeV. The data encompasses both pion (top panel) and positron (bottom panel) clusters.