Generative Unfolding with Distribution Mapping
This provides an incremental improvement for high-energy physics researchers needing precise cross-section measurements from complex data.
The paper tackles the problem of ensuring generative models learn correct conditional probabilities when unfolding data from simulations, extending Schrödinger Bridges and Direct Diffusion techniques to achieve accuracy comparable to state-of-the-art conditional generative unfolding methods. Results show this approach works on a standard single jet substructure benchmark and a new 22-dimensional Z + 2-jets dataset.
Machine learning enables unbinned, highly-differential cross section measurements. A recent idea uses generative models to morph a starting simulation into the unfolded data. We show how to extend two morphing techniques, Schrödinger Bridges and Direct Diffusion, in order to ensure that the models learn the correct conditional probabilities. This brings distribution mapping to a similar level of accuracy as the state-of-the-art conditional generative unfolding methods. Numerical results are presented with a standard benchmark dataset of single jet substructure as well as for a new dataset describing a 22-dimensional phase space of Z + 2-jets.