Gast ::
| Hauptseite > Normsätze > Projekte > Datensatz #885779 |
ZT-I-PF-5-3
Deep Generative models for fast and precise physics Simulation (DeGeSim)
| Coordinator | Krücker, Dirk |
| Grant period | 2020-2023 |
| Funding body | Helmholtz Gemeinschaft Deutscher Forschungszentren |
| HGF | |
| Identifier | G:(DE-HGF)2020_ZT-I-PF-5-3 |
⇧ Impuls- und Vernetzungsfonds ⇧
Note: This project explores the capability of Deep Generative Models, like Generative Adversarial Networks (GANs) or Variational Autoencoders (VAEs), trained on large physics-based data sets to create reliable models for the fast production of precise simulations. We study two types of urgently needed simulations in elementary particle physics in the context of the future High Luminosity Phase of the Large Hadron Collider. The challenging aspect of our project is the modeling of hadronic processes in two different settings: once in hadronic showers in calorimeters and secondly in general soft-hadronic processes in proton-proton collisions. Due to the complexity of the interactions both cases are computationally demanding and are, even under the present conditions, not modeled satisfactorily using traditional simulation approaches. The challenge becomes even more demanding under the intense future beam conditions with high event pile-up and with novel detector devices with magnitudes more signal channels than used in present devices. Deep Generative Models are generally assumed to be able to master these challenges for fast, but precise simulations. We are part of the Helmholtz-TRIUMF Cooperation and a joined project with the Jülich Supercomputing Centre (JSC) at Forschungszentrum Jülich GmbH which operates one of the most powerful HPC infrastructures in Europe, enabling scientists and engineers to solve highly complex and socially relevant problems by simulations, and which is the host of Helmholtz AI Local dedicated to basic and applied Deep Learning research.All known publications ...
Download: BibTeX | EndNote XML, Text | RIS |
Journal Article
Adversarial domain adaptation to reduce sample bias of a high energy physics event classifier
Machine learning: science and technology 3(1), 015014 (2022) [10.1088/2632-2153/ac3dde]
Files Fulltext by OpenAccess repository
BibTeX |
EndNote:
XML,
Text |
RIS
Poster (After Call)
DeGeSim and the High Granularity Calorimeter for the CMS Experiment at the Large Hadron Collider
Helmholtz AI virtual conference 2021, onlineonline, Germany, 14 Apr 2021 - 15 Apr 2021
Files Fulltext Fulltext by OpenAccess repository
BibTeX |
EndNote:
XML,
Text |
RIS
All known publications ...
Download: BibTeX | EndNote XML, Text | RIS |
