Modern neutrino experiments require precise tuning of energy response parameters, a task complicated by the parameters’ nonlinear behavior and strong correlations. The authors present neural density estimators using normalizing flows and transformers integrating them with Bayesian nested sampling to achieve near-zero systematic biases and uncertainties limited only by statistics, offering a flexible framework for particle physics applications
- Arsenii Gavrikov
- Andrea Serafini
- Lucia Votano
