Inverse Autoregressive Flows for Zero Degree Calorimeter fast simulation
Emilia Majerz, Witold Dzwinel, Jacek Kitowski
TL;DR
The paper tackles the high computational cost of ZDC simulations by introducing physics-informed surrogates based on a teacher–student Normalizing Flow framework that combines MAF teachers with IAF students. A physics-based channel loss, together with a diversity-based weighting, improves the mapping from input particles to detector responses, achieving better physical fidelity as reflected in $MAE_c$ and $MAE_{cw}$ while delivering substantial speedups. The authors demonstrate a 421× faster sampling compared with prior NF approaches and show that channel-based metrics better capture input–output dependencies than Wasserstein distance alone. This work provides a practical, high-fidelity surrogacy method for ZDC simulations that can accelerate physics analyses at ALICE and similar calorimeter systems.
Abstract
Physics-based machine learning blends traditional science with modern data-driven techniques. Rather than relying exclusively on empirical data or predefined equations, this methodology embeds domain knowledge directly into the learning process, resulting in models that are both more accurate and robust. We leverage this paradigm to accelerate simulations of the Zero Degree Calorimeter (ZDC) of the ALICE experiment at CERN. Our method introduces a novel loss function and an output variability-based scaling mechanism, which enhance the model's capability to accurately represent the spatial distribution and morphology of particle showers in detector outputs while mitigating the influence of rare artefacts on the training. Leveraging Normalizing Flows (NFs) in a teacher-student generative framework, we demonstrate that our approach not only outperforms classic data-driven model assimilation but also yields models that are 421 times faster than existing NF implementations in ZDC simulation literature.