SciPost Submission Page
Interdisciplinary Digital Twin Engine InterTwin for calorimeter simulation
by Corentin Allaire, Vera Maiboroda, David Rousseau
Submission summary
| Authors (as registered SciPost users): | Vera Maiboroda · David Rousseau |
| Submission information | |
|---|---|
| Preprint Link: | https://arxiv.org/abs/2509.26527v1 (pdf) |
| Date submitted: | Oct. 1, 2025, 1 p.m. |
| Submitted by: | Vera Maiboroda |
| Submitted to: | SciPost Physics Proceedings |
| Proceedings issue: | The 2nd European AI for Fundamental Physics Conference (EuCAIFCon2025) |
| Ontological classification | |
|---|---|
| Academic field: | Physics |
| Specialties: |
|
| Approach: | Experimental |
The author(s) disclose that the following generative AI tools have been used in the preparation of this submission:
ChatGPT for grammar checks
Abstract
Calorimeter shower simulations are computationally expensive, and generative models offer an efficient alternative. However, achieving a balance between accuracy and speed remains a challenge, with distribution tail modeling being a key limitation. Invertible generative network CaloINN provides a trade-off between simulation quality and efficiency. The ongoing study targets introducing a set of post-processing modifications of analysis-level observables aimed at improving the accuracy of distribution tails. As part of interTwin project initiative developing an open-source Digital Twin Engine, we implemented the CaloINN within the interTwin AI framework.
Current status:
Reports on this Submission
Report
Requested changes
Fig. 1: vertical axis labels
Section 3:
Its implementation within itwinai benefits from the framework’s experiment tracking, configuration management, and workflow orchestration, thereby ensuring reproducibility and facilitating large-scale experiments.
-> Can this be made more concrete?
Section 4:
Perhaps refrain from statements like "of the most promising generative model, CaloINN." without a detailed evaluation why this should be the *most* promising model.
"trained on datasets with artificially enhanced tails" -> provide the equation/mechanism used to create this
"These ratios are subsequently applied during inference to reweight and correct the generated distributions," -> provide equations and references (at least the description provided here does not sound like FastPerfekt, but rather DTCR)
Recommendation
Ask for minor revision