Transferable Fast Calorimeter Shower Generation via Multi-Geometry Pre-training
Abstract
Detailed Geant4 simulation of calorimeter showers dominates the computing budget of high-energy physics experiments. Deep generative surrogates reduce this cost, but they have remained tied to the detector they were trained on, so each new geometry needs a large in-domain dataset. We study whether a single point cloud shower generator can be pre-trained on multiple detectors and transferred to unseen calorimeters. The pre-training geometries come from synthetic geometric variation rather than real-detector data. We introduce SimpleBox, a family of 104 box calorimeters spanning the plane of sampling fraction and longitudinal segmentation, and benchmark it against pre-training on realistic detectors. On a calorimeter unseen in pre-training, with 103 target showers for fine-tuning, the two priors reduce the aggregated sliced Wasserstein distance to Geant4 by factors of 5.2 (synthetic) and 8.0 (realistic) relative to training from scratch. At larger target sizes the synthetic prior performs better than the realistic one. Geometric diversity alone is therefore a practical way to pre-train a transferable shower generator.
Versions
Related
- Cross-geometry transfer and model collapse in point cloud calorimeter shower generation
- Calibration of electromagnetic shower features in the CMS calorimeter with machine-learning techniques
- Generalized Forcing Method: Generation of Diverse Data for Training Linear Transport PDE Closure Models
- Temporal Geometry of Deep Networks: Hyperbolic Representations of Training Dynamics for Intrinsic Explainability
- AnchorPose for Geometry-Aware MOF Assembly through Meso-Grained Pose Generation
BibTeX
@misc{buss2026transferable,
title = {{Transferable Fast Calorimeter Shower Generation via Multi-Geometry Pre-training}},
author = {Thorsten Buss and Henry Day-Hall and Frank Gaede and Gregor Kasieczka and Katja Krüger and Peter McKeown and Lorenzo Valente},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2608.18233},
eprint = {2608.18233},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.18233},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Transferable Fast Calorimeter Shower Generation via Multi-Geometry Pre-training
AU - Thorsten Buss
AU - Henry Day-Hall
AU - Frank Gaede
AU - Gregor Kasieczka
AU - Katja Krüger
AU - Peter McKeown
AU - Lorenzo Valente
PY - 2026
DA - 2026/10/29
PB - arXiv
DO - 10.48550/arxiv.2608.18233
UR - https://arxiv.org/abs/2608.18233
AB - Detailed Geant4 simulation of calorimeter showers dominates the computing budget of high-energy physics experiments. Deep generative surrogates reduce this cost, but they have remained tied to the detector they were trained on, so each new geometry needs a large in-domain dataset. We study whether a single point cloud shower generator can be pre-trained on multiple detectors and transferred to unseen calorimeters. The pre-training geometries come from synthetic geometric variation rather than real-detector data. We introduce SimpleBox, a family of 10^4 box calorimeters spanning the plane of sampling fraction and longitudinal segmentation, and benchmark it against pre-training on realistic detectors. On a calorimeter unseen in pre-training, with 10^3 target showers for fine-tuning, the two priors reduce the aggregated sliced Wasserstein distance to Geant4 by factors of 5.2 (synthetic) and 8.0 (realistic) relative to training from scratch. At larger target sizes the synthetic prior performs better than the realistic one. Geometric diversity alone is therefore a practical way to pre-train a transferable shower generator.
ER -