Sofia Vallecorsa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Sofia Vallecorsa has a total of 1000 co-authors. Here are the top co-authors by number of shared publications:
- Michele Grossi
European Organization for Nuclear Research (CERN)
28 shared publications - Oriel Kiss
European Organization for Nuclear Research (CERN)
12 shared publications - Kerstin Borras
RWTH Aachen University, III. Physikalisches Institut A
8 shared publications - Antonio Mandarino
University of Gdańsk
6 shared publications - Dirk Krücker
48 Deutsches Elektronen-Synchrotron, Hamburg, Germany
6 shared publications - Eilas Fernandez Combarro
Department of Computer Science, University of Oviedo, Gijón, Spain
6 shared publications
Latest Publications
- Open Access
- 17 May 2026
- Published by Springer Nature in Communications Physics
- Vol. 9 (1)
Roughening and dynamics of an electric flux string in a (2+1)D lattice gauge theory
- 17 May 2026
- Published by Springer Nature in Communications Physics
- Vol. 9 (1)
Open AccessRoughening and dynamics of an electric flux string in a (2+1)D lattice gauge theory
- Open Access
- 1 April 2026
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 3206 (1)
Federated Learning Strategies of Generative Adversarial Networks for High Energy Physics Calorimeter Simulation
Calorimeter simulation using Monte-Carlo is a computationally demanding task both in terms of computing and storage and relies on the use of distributed computing clusters. Generative Adversarial Networks (GANs) provide a more efficient alternative by capturing the complex underlying data distribution and generating synthetic data rapidly. In most cases, existing prototypes simulate the traversal of individual particles through small detector volumes, and thus, the training datasets are of a manageable size and confined to a single cluster. The next natural step in this line of research is to train larger generative models that incorporate more complex physics processes and detectors (including various sub-detector combinations). This would increase the training dataset size, and most likely require merging independently produced samples. Given that Monte Carlo data is generated and distributed across multiple cluster nodes (such as those within the WLCG tiered infrastructure), the communication overhead involved in transferring data between servers for centralized training would inevitably rise. To address this issue, we propose federated learning as a decentralized training solution, wherein a group of collaborators trains a model by sharing training updates with an aggregator. Our approach combines federated learning with GAN-based calorimeter simulation by locally training individual GAN models on different portions of the Monte Carlo data and aggregating the resulting models using the FedAvg algorithm. Our experiments demonstrate the effectiveness of our approach for GAN-based calorimeter simulation, achieving comparable simulation accuracy. Our approach could contribute to developing more accurate and scalable simulation methods for particle physics experiments.
- 1 April 2026
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 3206 (1)
Open AccessFederated Learning Strategies of Generative Adversarial Networks for High Energy Physics Calorimeter Simulation
Calorimeter simulation using Monte-Carlo is a computationally demanding task both in terms of computing and storage and relies on the use of distributed computing clusters. Generative Adversarial Networks (GANs) provide a more efficient alternative by capturing the complex underlying data distribution and generating synthetic data rapidly. In most cases, existing prototypes simulate the traversal of individual particles through small detector volumes, and thus, the training datasets are of a manageable size and confined to a single cluster. The next natural step in this line of research is to train larger generative models that incorporate more complex physics processes and detectors (including various sub-detector combinations). This would increase the training dataset size, and most likely require merging independently produced samples. Given that Monte Carlo data is generated and distributed across multiple cluster nodes (such as those within the WLCG tiered infrastructure), the communication overhead involved in transferring data between servers for centralized training would inevitably rise. To address this issue, we propose federated learning as a decentralized training solution, wherein a group of collaborators trains a model by sharing training updates with an aggregator. Our approach combines federated learning with GAN-based calorimeter simulation by locally training individual GAN models on different portions of the Monte Carlo data and aggregating the resulting models using the FedAvg algorithm. Our experiments demonstrate the effectiveness of our approach for GAN-based calorimeter simulation, achieving comparable simulation accuracy. Our approach could contribute to developing more accurate and scalable simulation methods for particle physics experiments.
- Top Cited Papers
- 1 November 2025
- Published by IOP Publishing in Reports on Progress in Physics
- Vol. 88 (11) , 116201
CaloChallenge 2022: a community challenge for fast calorimeter simulation
We present the results of the ‘Fast Calorimeter Simulation Challenge 2022’—the CaloChallenge. We study state-of-the-art generative models on four calorimeter shower datasets of increasing dimensionality, ranging from a few hundred voxels to a few tens of thousand voxels. The 31 individual submissions span a wide range of current popular generative architectures, including variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows, diffusion models, and models based on conditional flow matching. We compare all submissions in terms of quality of generated calorimeter showers, as well as shower generation time and model size. To assess the quality we use a broad range of different metrics including differences in one-dimensional histograms of observables, KPD/FPD scores, AUCs of binary classifiers, and the log-posterior of a multiclass classifier. The results of the CaloChallenge provide the most complete and comprehensive survey of cutting-edge approaches to calorimeter fast simulation to date. In addition, our work provides a uniquely detailed perspective on the important problem of how to evaluate generative models. As such, the results presented here should be applicable for other domains that use generative AI and require fast and faithful generation of samples in a large phase space. Report Numbers : HEPHY-ML-24-05, FERMILAB-PUB-24-0728-CMS, TTK-24-43.
- 1 November 2025
- Published by IOP Publishing in Reports on Progress in Physics
- Vol. 88 (11) , 116201
Top Cited PapersCaloChallenge 2022: a community challenge for fast calorimeter simulation
We present the results of the ‘Fast Calorimeter Simulation Challenge 2022’—the CaloChallenge. We study state-of-the-art generative models on four calorimeter shower datasets of increasing dimensionality, ranging from a few hundred voxels to a few tens of thousand voxels. The 31 individual submissions span a wide range of current popular generative architectures, including variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows, diffusion models, and models based on conditional flow matching. We compare all submissions in terms of quality of generated calorimeter showers, as well as shower generation time and model size. To assess the quality we use a broad range of different metrics including differences in one-dimensional histograms of observables, KPD/FPD scores, AUCs of binary classifiers, and the log-posterior of a multiclass classifier. The results of the CaloChallenge provide the most complete and comprehensive survey of cutting-edge approaches to calorimeter fast simulation to date. In addition, our work provides a uniquely detailed perspective on the important problem of how to evaluate generative models. As such, the results presented here should be applicable for other domains that use generative AI and require fast and faithful generation of samples in a large phase space. Report Numbers : HEPHY-ML-24-05, FERMILAB-PUB-24-0728-CMS, TTK-24-43.
- Open Access
- 27 August 2025
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 6 (3)
Guided Graph Compression for Quantum Graph Neural Networks
Graph Neural Networks (GNNs) are effective for processing graph-structured data but face challenges with large graphs due to high memory requirements and inefficient sparse matrix operations on GPUs. Quantum Computing (QC) offers a promising avenue to address these issues and inspires new algorithmic approaches. In particular, Quantum Graph Neural Networks (QGNNs) have been explored in recent literature. However, current quantum hardware limits the dimension of the data that can be effectively encoded. Existing approaches either simplify datasets manually or use artificial graph datasets. This work introduces the Guided Graph Compression (GGC) framework, which uses a graph autoencoder to reduce both the number of nodes and the dimensionality of node features. The compression is guided to enhance the performance of a downstream classification task, which can be applied either with a quantum or a classical classifier. The framework is evaluated on the Jet Tagging task, a classification problem of fundamental importance in high energy physics that involves distinguishing particle jets initiated by quarks from those by gluons. We compare GGC to a model that uses the autoencoder as a preprocessing step and to a baseline classical GNN classifier. Our numerical results demonstrate that GGC outperforms both alternatives, while also facilitating the testing of novel QGNN ansatzes on realistic datasets.
- 27 August 2025
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 6 (3)
Open AccessGuided Graph Compression for Quantum Graph Neural Networks
Graph Neural Networks (GNNs) are effective for processing graph-structured data but face challenges with large graphs due to high memory requirements and inefficient sparse matrix operations on GPUs. Quantum Computing (QC) offers a promising avenue to address these issues and inspires new algorithmic approaches. In particular, Quantum Graph Neural Networks (QGNNs) have been explored in recent literature. However, current quantum hardware limits the dimension of the data that can be effectively encoded. Existing approaches either simplify datasets manually or use artificial graph datasets. This work introduces the Guided Graph Compression (GGC) framework, which uses a graph autoencoder to reduce both the number of nodes and the dimensionality of node features. The compression is guided to enhance the performance of a downstream classification task, which can be applied either with a quantum or a classical classifier. The framework is evaluated on the Jet Tagging task, a classification problem of fundamental importance in high energy physics that involves distinguishing particle jets initiated by quarks from those by gluons. We compare GGC to a model that uses the autoencoder as a preprocessing step and to a baseline classical GNN classifier. Our numerical results demonstrate that GGC outperforms both alternatives, while also facilitating the testing of novel QGNN ansatzes on realistic datasets.
- Open Access
Roughening and dynamics of an electric flux string in a (2+1)D lattice gauge theory
We investigate the roughening transition in the pure $\mathbb{Z}_2$ lattice gauge theory in (2+1) dimensions. Using numerical simulations with matrix product states, we explore the static and dynamical properties of an electric flux string between two static charges as the coupling is varied and approaches the deconfinement phase transition from the confined phase. Within the roughening region, we obtain the universal Lüscher correction to the confining potential and observe the expected restoration of rotational symmetry. Our simulations of the out-of-equilibrium evolution of a string reveal that the growth of the entanglement entropy of the state and the string width exhibit qualitatively different behavior in the roughening region compared to the deeply confined one. In particular, we find that the rate of entropy growth is consistent with an effective description of the string excitations by a bosonic model in the roughening phase.Open AccessRoughening and dynamics of an electric flux string in a (2+1)D lattice gauge theory
We investigate the roughening transition in the pure $\mathbb{Z}_2$ lattice gauge theory in (2+1) dimensions. Using numerical simulations with matrix product states, we explore the static and dynamical properties of an electric flux string between two static charges as the coupling is varied and approaches the deconfinement phase transition from the confined phase. Within the roughening region, we obtain the universal Lüscher correction to the confining potential and observe the expected restoration of rotational symmetry. Our simulations of the out-of-equilibrium evolution of a string reveal that the growth of the entanglement entropy of the state and the string width exhibit qualitatively different behavior in the roughening region compared to the deeply confined one. In particular, we find that the rate of entropy growth is consistent with an effective description of the string excitations by a bosonic model in the roughening phase. - Open Access
- 28 May 2025
- Published by IOP Publishing in Physica Scripta
- Vol. 100 (6) , 065126
A study on quantum graph neural networks applied to molecular physics
This paper introduces a novel architecture for Quantum Graph Neural Networks, which is significantly different from previous approaches found in the literature. The proposed approach produces similar outcomes with respect to previous models but with fewer parameters, resulting in an extremely interpretable architecture rooted in the underlying physics of the problem. The architectural novelties arise from three pivotal aspects. Firstly, we employ an embedding updating method that is analogous to classical Graph Neural Networks, therefore bridging the classical-quantum gap. Secondly, each layer is devoted to capturing interactions of distinct orders, aligning with the physical properties of the system. Lastly, we harness SWAP gates to emulate the problem’s inherent symmetry, a novel strategy not found currently in the literature. The obtained results in the considered experiments are encouraging to lay the foundation for continued research in this field.
- 28 May 2025
- Published by IOP Publishing in Physica Scripta
- Vol. 100 (6) , 065126
Open AccessA study on quantum graph neural networks applied to molecular physics
This paper introduces a novel architecture for Quantum Graph Neural Networks, which is significantly different from previous approaches found in the literature. The proposed approach produces similar outcomes with respect to previous models but with fewer parameters, resulting in an extremely interpretable architecture rooted in the underlying physics of the problem. The architectural novelties arise from three pivotal aspects. Firstly, we employ an embedding updating method that is analogous to classical Graph Neural Networks, therefore bridging the classical-quantum gap. Secondly, each layer is devoted to capturing interactions of distinct orders, aligning with the physical properties of the system. Lastly, we harness SWAP gates to emulate the problem’s inherent symmetry, a novel strategy not found currently in the literature. The obtained results in the considered experiments are encouraging to lay the foundation for continued research in this field.
- Top Cited PapersOpen Access
- 13 November 2024
- Published by Springer Nature in npj Quantum Information
- Vol. 10 (1) , 1-18
Trainability barriers and opportunities in quantum generative modeling
Quantum generative models provide inherently efficient sampling strategies and thus show promise for achieving an advantage using quantum hardware. In this work, we investigate the barriers to the trainability of quantum generative models posed by barren plateaus and exponential loss concentration. We explore the interplay between explicit and implicit models and losses, and show that using quantum generative models with explicit losses such as the KL divergence leads to a new flavor of barren plateaus. In contrast, the implicit Maximum Mean Discrepancy loss can be viewed as the expectation value of an observable that is either low-bodied and provably trainable, or global and untrainable depending on the choice of kernel. In parallel, we find that solely low-bodied implicit losses cannot in general distinguish high-order correlations in the target data, while some quantum loss estimation strategies can. We validate our findings by comparing different loss functions for modeling data from High-Energy-Physics.
- 13 November 2024
- Published by Springer Nature in npj Quantum Information
- Vol. 10 (1) , 1-18
Top Cited PapersOpen AccessTrainability barriers and opportunities in quantum generative modeling
Quantum generative models provide inherently efficient sampling strategies and thus show promise for achieving an advantage using quantum hardware. In this work, we investigate the barriers to the trainability of quantum generative models posed by barren plateaus and exponential loss concentration. We explore the interplay between explicit and implicit models and losses, and show that using quantum generative models with explicit losses such as the KL divergence leads to a new flavor of barren plateaus. In contrast, the implicit Maximum Mean Discrepancy loss can be viewed as the expectation value of an observable that is either low-bodied and provably trainable, or global and untrainable depending on the choice of kernel. In parallel, we find that solely low-bodied implicit losses cannot in general distinguish high-order correlations in the target data, while some quantum loss estimation strategies can. We validate our findings by comparing different loss functions for modeling data from High-Energy-Physics.
- Top Cited PapersOpen Access
- 14 October 2024
- Published by Springer Nature in Communications Physics
- Vol. 7 (1) , 1-11
Quantum anomaly detection in the latent space of proton collision events at the LHC
- 14 October 2024
- Published by Springer Nature in Communications Physics
- Vol. 7 (1) , 1-11
Top Cited PapersOpen AccessQuantum anomaly detection in the latent space of proton collision events at the LHC
- 11 October 2024
- Published by Springer Nature
- p. 35-50
Spooky Quantum Action: From Thought Experiments to Real World Quantum Technology Application
- 11 October 2024
- Published by Springer Nature
- p. 35-50
Spooky Quantum Action: From Thought Experiments to Real World Quantum Technology Application
- Open Access
- 13 August 2024
- Published by Springer Nature in The European Physical Journal Plus
- Vol. 139 (8) , 1-12
Assessment of few-hits machine learning classification algorithms for low-energy physics in liquid argon detectors
The physics potential of massive liquid argon TPCs in the low-energy regime is still to be fully reaped because few-hits events encode information that can hardly be exploited by conventional classification algorithms. Machine learning (ML) techniques give their best in these types of classification problems. In this paper, we evaluate their performance against conventional (deterministic) algorithms. We demonstrate that both Convolutional Neural Networks (CNN) and Transformer-Encoder methods outperform deterministic algorithms in one of the most challenging classification problems of low-energy physics (single- versus double-beta events). We discuss the advantages and pitfalls of Transformer-Encoder methods versus CNN and employ these methods to optimize the detector parameters, with an emphasis on the DUNE Phase II detectors (“Module of Opportunity”).
- 13 August 2024
- Published by Springer Nature in The European Physical Journal Plus
- Vol. 139 (8) , 1-12
Open AccessAssessment of few-hits machine learning classification algorithms for low-energy physics in liquid argon detectors
The physics potential of massive liquid argon TPCs in the low-energy regime is still to be fully reaped because few-hits events encode information that can hardly be exploited by conventional classification algorithms. Machine learning (ML) techniques give their best in these types of classification problems. In this paper, we evaluate their performance against conventional (deterministic) algorithms. We demonstrate that both Convolutional Neural Networks (CNN) and Transformer-Encoder methods outperform deterministic algorithms in one of the most challenging classification problems of low-energy physics (single- versus double-beta events). We discuss the advantages and pitfalls of Transformer-Encoder methods versus CNN and employ these methods to optimize the detector parameters, with an emphasis on the DUNE Phase II detectors (“Module of Opportunity”).