Search papers, labs, and topics across Lattice.
This paper introduces a generative surrogate model using Normalizing Flows (NFs) to efficiently simulate the responses of the ALICE Zero Degree Calorimeter (ZDC) at the LHC, addressing the computational challenges posed by traditional Monte Carlo methods. By employing transfer learning and fine-tuning specialized models for various particle types, the authors enhance the simulation accuracy with new metrics that account for conditional dependencies in the data. The resulting ensemble of models achieves a Wasserstein distance of $1.61 \pm 0.02$, significantly surpassing existing baselines and providing a robust framework for LHC detector simulations.
Fine-tuned Normalizing Flows can dramatically reduce simulation costs while improving accuracy in high-energy physics experiments.
Simulating the ALICE Zero Degree Calorimeter (ZDC) neutron detector responses at the LHC is computationally expensive, requiring complex Monte Carlo chains. We develop a generative surrogate, focusing on Normalizing Flows (NFs). Through transfer learning, we pre-train on the full imbalanced dataset and fine-tune specialized models for different particle types ($\gamma$, $n$, $\Lambda$, $K_S^0$, $\Sigma^+$) using two gradual-unfreezing schemes. As standard ZDC metrics like Wasserstein distance overlook conditional structure, we introduce refined metrics: conditional weighted MAE, dispersion ratio, and Jaccard co-activation error, that better capture physics-relevant input-output dependencies and response variability. Our ensemble of fine-tuned models achieves a Wasserstein distance of $1.61 \pm 0.02$, outperforming baselines across all metrics. This work provides a generalizable NF-based framework for LHC detector simulation, combining NFs, conditional fine-tuning, and physics-motivated evaluation.
