arXiv:2410.22870 · npj Quantum Information

Conditioned quantum-assisted deep generative surrogate for particle-calorimeter interactions

J. Quetzalcóatl Toledo-Marín, Sebastian Gonzalez, Hao Jia, Ian Lu, Deniz Sogutlu, Abhishek Singh, C. Gay, Eric Paquet, Roger G. Melko, Geoffrey Fox, M. Swiatlowski, Wojciech T. Fedorko

Published Jul 7, 2025• 3 citations

arXiv HTML PDF

Abstract

Particle collisions at accelerators like the Large Hadron Collider (LHC), recorded by experiments such as ATLAS and CMS, enable precise standard model measurements and searches for new phenomena. Simulating these collisions significantly influences experiment design and analysis but incurs immense computational costs, projected at millions of CPU-years annually during the high luminosity LHC (HL-LHC) phase. Currently, simulating a single event with Geant4 consumes around 1000 CPU seconds, with calorimeter simulations especially demanding. To address this, we propose a conditioned quantum-assisted generative model, integrating a conditioned variational autoencoder (VAE) and a conditioned restricted Boltzmann machine (RBM). Our RBM architecture is tailored for D-Wave's Pegasus-structured advantage quantum annealer for sampling, leveraging the flux bias for conditioning. This approach combines classical RBMs as universal approximators for discrete distributions with quantum annealing's speed and scalability. We also introduce an adaptive method for efficiently estimating effective inverse temperature, and validate our framework on Dataset 2 of CaloChallenge.

DOI: 10.1038/s41534-025-01040-x

Related papers

Zephyr quantum-assisted hierarchical Calo4pQVAE for particle-calorimeter interactions

Lu, Ian, Hao Wei Jia, Gonzalez, Sebastian +10 more

With the approach of the High Luminosity Large Hadron Collider (HL-LHC) era set to begin particle collisions by the end of this decade, it is evident that the computational demands of traditional collision simulation methods are becoming increasingly unsustainable. Existing approaches, which rely heavily on first-principles Monte Carlo simulations for modeling event showers in calorimeters, are projected to require millions of CPU-years annually -- far exceeding current computational capacities. This bottleneck presents an exciting opportunity for advancements in computational physics by integrating deep generative models with quantum simulations. We propose a quantum-assisted hierarchical deep generative surrogate founded on a variational autoencoder (VAE) in combination with an energy conditioned restricted Boltzmann machine (RBM) embedded in the model's latent space as a prior. By mapping the topology of D-Wave's Zephyr quantum annealer (QA) into the nodes and couplings of a 4-partite RBM, we leverage quantum simulation to accelerate our shower generation times significantly. To evaluate our framework, we use Dataset 2 of the CaloChallenge 2022. Through the integration of classical computation and quantum simulation, this hybrid framework paves way for utilizing large-scale quantum simulations as priors in deep generative models.

arXiv:2412.04677Dec 6, 20240 citations

CaloQVAE for electromagnetic shower synthetic data generation in the ATLAS calorimeter

Wojciech T. Fedorko

As CERN approaches the High-Luminosity LHC (HL-LHC) run, the computational cost of first-principles detector simulation has become a critical bottleneck, with projections indicating millions of CPU-years required to generate the necessary Monte Carlo datasets — a substantial share of which is consumed by calorimeter shower simulation. Building on our earlier proof-of-concept demonstrating Quantum-assisted Generative AI for synthetic calorimetry data generation, we present what is, to our knowledge, the first full-scale demonstration of this approach applied to ATLAS detector simulated data. Our model couples a Restricted Boltzmann Machine (RBM), mapped onto the logical qubit topology of D-Wave's quantum annealing processor, with a deep generative architecture to produce realistically scaled calorimeter showers. We extend our earlier framework with several techniques critical to scaling the approach to production-grade ATLAS geometry and granularity: advanced prior conditioning strategies, Spin Reversal Transforms to mitigate intrinsic qubit biases, orbit averaging to reduce sampling noise, and QPU-aware weight normalization to stabilize the sampling process. We show how these advances yield quantum processor generated calorimeter datasets that maintain high fidelity in physics features relative to first-principles GEANT4 simulation while promising substantial improvements in generation throughput over both traditional simulation and purely classical deep learning generative models. We discuss the implications of these results for scaling quantum-assisted generative modeling to full ATLAS fast-simulation pipelines ahead of the HL-LHC era.

Sep 25, 20260 citations

Quantum-Assisted Generative AI for Simulation of the Calorimeter Response

Wojciech T. Fedorko, J. Quetzalcóatl Toledo-Marín, Geoffrey Fox +7 more

As CERN approaches the launch of the High Luminosity Large Hadron Collider (HL-LHC) by the decade’s end, the computational demands of traditional simulations have become untenably high. Projections show millions of CPU-years required to create simulated datasets - with a substantial fraction of CPU time devoted to calorimetric simulations. This presents unique opportunities for breakthroughs in computational physics. We show how Quantumassisted Generative AI can be used for the purpose of creating synthetic, realistically scaled calorimetry dataset. The model is constructed by combining D-Wave’s Quantum Annealer processor with a Deep Learning architecture, increasing the timing performance with respect to first principles simulations and Deep Learning models alone, while maintaining current state-of-the-art data quality

EPJ Web of ConferencesJan 1, 20250 citations

Quantum-Assisted Generative AI for Simulation of the Calorimeter Response

Wojciech T. Fedorko, J. Quetzalcóatl Toledo-Marín, Geoffrey Fox +7 more

Oct 7, 20250 citations

A Quantum Generative Adversarial Network model for the Simulation of LHC Calorimeter Data

PHILIP CORSINI

Quantum Computing has emerged as a viable alternative to classical computing for a number of different tasks, with the idea of a potential speedup in mind. One of the most promising fields of research in quantum computing is Quantum Machine Learning. An interesting area of application for Quantum Machine Learning can be found in High Energy Physics: one key component of creating experiments in a particle accelerator like the Large Hadron Collider is the simulation of those experiments beforehand, in order to gain a better understanding of the results actually measured. Those types of simulations, typically based on Monte-Carlo methods, require significant computing resources and, as we approach CERN's High Luminosity Phase, there is a strong need for more efficient and cost controlled simulations. In this thesis a model based on Quantum Generative Adversarial Networks (QGANs) is implemented, with the intention of exploring potential quantum computing advantages within a hybrid machine learning model that is tasked with the generation of calorimeter samples. The model is initially trained on an ideal simulator to ensure that the results are comparable to those of a similar classical model. Finally, the model is modified in order to effectively run on a real quantum device and the training is executed in its entirety on a 54-qubit quantum processor.

Apr 15, 20260 citations

Multi-Mode Quantum Annealing for Generative Representation Learning with Boltzmann Priors

Gilhan Kim, Daniel K. Park

Energy-based models provide a natural bridge between statistical physics and machine learning by representing data through structured energy landscapes. Boltzmann machines are a particularly compelling class of such models for capturing complex interactions among latent variables, but their use in modern generative learning has been limited by the classical intractability of sampling from general (non-restricted) Boltzmann distributions. Here we develop a quantum-annealing-based framework that enables variational autoencoders with general Boltzmann priors. The framework employs three complementary annealing modes tailored to different stages of learning and deployment: diabatic quantum annealing provides unbiased Boltzmann samples for efficient training, slower annealing concentrates samples near low-energy configurations of the learned prior for unconditional generation, and conditional annealing with external fields steers the learned energy landscape toward attribute-specific regions for conditional generation and semantic editing. Using up to 2000 qubits on a D-Wave Advantage2 processor, we demonstrate stable training and high-quality generation on MNIST, Fashion-MNIST, and CelebA, achieving faster convergence and lower reconstruction loss than a Gaussian-prior VAE with the same encoder-decoder architecture. Beyond generation, the learned energy function provides out-of-distribution detection signals that add discriminative power beyond reconstruction loss. We demonstrate that these scores separate in-distribution samples from held-out digit classes in one-class MNIST experiments and improve the detection of market regime shifts in financial data. These results establish quantum annealing as a practical and controllable physical mechanism for energy-based representation learning and generative modeling beyond the reach of tractable classical approaches.

quant-pharXiv:2604.00919Apr 2, 20260 citations