SciPost Submission Page
Generative Unfolding with Distribution Mapping
by Anja Butter, Sascha Diefenbacher, Nathan Huetsch, Vinicius Mikuni, Benjamin Nachman, Sofia Palacios Schweitzer, Tilman Plehn
Submission summary
| Authors (as registered SciPost users): | Nathan Huetsch · Sofia Palacios Schweitzer · Tilman Plehn |
| Submission information | |
|---|---|
| Preprint Link: | https://arxiv.org/abs/2411.02495v1 (pdf) |
| Date submitted: | 2025-01-07 10:33 |
| Submitted by: | Huetsch, Nathan |
| Submitted to: | SciPost Physics |
| Ontological classification | |
|---|---|
| Academic field: | Physics |
| Specialties: |
|
| Approaches: | Experimental, Computational, Phenomenological |
Abstract
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, Schr\"odinger 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.
Author indications on fulfilling journal expectations
- Provide a novel and synergetic link between different research areas.
- Open a new pathway in an existing or a new research direction, with clear potential for multi-pronged follow-up work
- Detail a groundbreaking theoretical/experimental/computational discovery
- Present a breakthrough on a previously-identified and long-standing research stumbling block
Current status:
Reports on this Submission
Strengths
- It is a very clear presentation of a rather involved subject, with important benefits for the practitioners.
- Section 2 has a valuable discussion of general aspects of distribution mapping with stochastic differential equations.
- The proposed methodologies are studied thoroughly, not just by means of one example.
- Taken together, the toy example and the real-world application make a very compelling case with low-threshold entry points for the reader.
- The quality of the draft is excellent. Fonts in the body and the figures were made consistent.
- The code is public.
Weaknesses
Very small editorial remarks:
p2
2nd paragraph: "These results are promising, but due to the ill-posed nature of the problem, it is essential to have alternative methods."
-> It is not clear what is ill-posed here. Please clarify.
last sentence: " ... as well as any unfolding method." -> Please clarify which class this applies to. Clearly, you don't have traditionally binned unfolding in mind here.
p4 There are so few results on unbinned unfolding that it is a stretch to say that Omnifold is a "standard". Please rephrase.
Fig. 8 has too small labels.
Report
This paper draft presents techniques to improve unfolding techniques that traditionally suffer from improperly learning the conditional probabilities in the training data. It fully meets the quality standards of the journal; in fact, I had difficulties in finding any relevant weaknesses.
Requested changes
Please see above.
Recommendation
Publish (surpasses expectations and criteria for this Journal; among top 10%)