Theo Heimel

Theo Heimel

Postdoctoral researcher in particle physics and machine learning

About me

I am a postdoctoral researcher in particle physics and machine learning at the Centre for Cosmology, Particle Physics and Phenomenology in Louvain-la-Neuve. The main focus of my research is accelerating event generation for the High-Luminosity LHC and beyond using generative networks. In particular, the MadNIS project aims to make sampling hard-scattering events in MadGraph faster using neural importance sampling. Furthermore, I am interested in ML-based methods to extract more information from LHC data. This includes unfolding with generative networks, simulation-based inference, and combining machine-learned detector effects with theory knowledge in the matrix element method.

Career

Postdoc, 2024 –

Centre for Cosmology, Particle Physics and Phenomenology
UCLouvain

PhD in Physics, 2021 – 2024

Advisor: Tilman Plehn
Heidelberg University

MSc in Physics, 2018 – 2021

Advisor: Tilman Plehn
Heidelberg University

BSc in Physics, 2015 – 2018

Advisor: Tilman Plehn
Heidelberg University

Publications

Precision calibration of calorimeter signals in the ATLAS experiment using an uncertainty-aware neural network

G. Aad et al.
arXiv:2412.04370

Profile Likelihoods on ML-Steroids

T. Heimel, T. Plehn, N. Schmal
arXiv:2411.00942

Differentiable MadNIS-Lite

T. Heimel, O. Mattelaer, T. Plehn, R. Winterhalder
arXiv:2408.01486, SciPost Phys. 18, 017 (2025)

The landscape of unfolding with machine learning

N. Huetsch, J. Villadamigo, A. Shmakov, S. Diefenbacher, V. Mikuni, T. Heimel et al.
arXiv:2404.18807, SciPost Phys. 18, 070 (2025)

The MadNIS reloaded

T. Heimel, N. Huetsch, F. Maltoni, O. Mattelaer, T. Plehn, R. Winterhalder
arXiv:2311.01548, SciPost Phys. 17, 023 (2024)

Precision-machine learning for the matrix element method

T. Heimel, N. Huetsch, R. Winterhalder, T. Plehn, A. Butter
arXiv:2310.07752, SciPost Phys. 17, 129 (2024)

Returning CP-observables to the frames they belong

J. Ackerschott, R. Barman, D. Gonçalves, T. Heimel, T. Plehn
arXiv:2308.00027, SciPost Phys. 17, 001 (2024)

How to understand limitations of generative networks

R. Das, L. Favaro, T. Heimel, C. Krause, T. Plehn, D. Shih
arXiv:2305.16774, SciPost Phys. 16, 031 (2024)

MadNIS - Neural multi-channel importance sampling

T. Heimel, R. Winterhalder, A. Butter, J. Isaacson, C. Krause, F. Maltoni et al.
arXiv:2212.06172, SciPost Phys. 15, 141 (2023)

Modern Machine Learning for LHC Physicists

T. Plehn, A. Butter, B. Dillon, T. Heimel, C. Krause, R. Winterhalder
arXiv:2211.01421

Two invertible networks for the matrix element method

A. Butter, T. Heimel, T. Martini, S. Peitzsch, T. Plehn
arXiv:2210.00019, SciPost Phys. 15, 094 (2023)

Machine learning and LHC event generation

A. Butter et al.
arXiv:2203.07460, SciPost Phys. 14, 079 (2023)

Generative networks for precision enthusiasts

A. Butter, T. Heimel, S. Hummerich, T. Krebs, T. Plehn, A. Rousselot et al.
arXiv:2110.13632, SciPost Phys. 14, 078 (2023)

Measuring QCD Splittings with Invertible Networks

S. Bieringer, A. Butter, T. Heimel, S. Höche, U. Köthe, T. Plehn et al.
arXiv:2012.09873, SciPost Phys. 10, 126 (2021)

QCD or What?

T. Heimel, G. Kasieczka, T. Plehn, J. Thompson
arXiv:1808.08979, SciPost Phys. 6, 030 (2019)

Contact

theo.heimel@uclouvain.be

UCLouvain, CP3
Chemin du Cyclotron, 2 bte L07.01.05
1348 Louvain-la-Neuve
Belgium