Project: CM 2019-T1/TIC-13194 TALENTO M1 Comunidad de Madrid
C. Rappold (IP)
PhD student: Samuel Escrig Lopez
DeepHyp - DEEP LEARNING TECHNIQUES FOR THE STUDY OF EXOTIC HYPERNUCLE
Within the WASA@FRS collaboration, the upcoming S447 experiment of hypernuclear spectroscopy is scheduled for March 2022 at the GSI-FAIR facility, Germany. In 2019, the central part of the WASA detector was transported to GSI-FAIR. In the S447 experiment, the WASA central detector system will be combined with the fragment separator (FRS) of GSI. Fragments emitted forward from the hypernuclear decay will be measured by the second-half of the FRS (S2-S4) with an excellent momentum resolution. In the meantime, light particles from the decay of the hypernuclei will be measured by the WASA detectors. The S447 experiment was approved a second time during the General Program Advisory Committee of GSI-FAIR during the meeting of Sept. 2020. The DeepHyp project focus first on the preparation of the S447 experiment, and then on the data analysis of the obtained experimental data. The development of new analysis techniques based on deep learning techniques aims to improve the hypernuclear spectroscopy. The main objective is to tackle the current hypertriton lifetime puzzle and the confirmation of the nnΛ observation. The preparation of the experiment has been a main part for the DeepHyp project plan of 2020. The finalization of the Geant4 simulations of the realistic experimental apparatus was carried out. More realistic efficiencies of the particle tracking and event reconstruction were estimated. The ion-optics of S2-S4 was optimized for a high fragment acceptance with a expected resolving power of 5200. New deep learning algorithms are in development for the track finding procedure using graph neural networks. We have started the R&D of a new silicon micro vertex detector for precise estimation of the interaction vertex of the nuclear collisions.
Project: AEI PID2020-118009GA-I00 - C. Rappold (IP)
HYP@FRS = MACHINE LEARNING METHODS APPLIED TO THE SPECTROSCOPY OF HYPERNUCLEI: THE PUZZLE OF THE HYPERTRITON LIFETIME AND OF THE NNLAMBDA EXISTENCE STUDIED AT WASA@FRS
The goal of Hyp@FRS is to study precisely light hypernuclei with ion-induced reactions by employing the fragment separator of GSI and FAIR facilities in Darmstadt, Germany, as a high-resolution forward spectrometer. The objectives of Hyp@FRS is the improvement of spectroscopy through machine learning. The proposed activity revolves around the WASA@FRS project that takes place at the GSI-FAIR facility. With the upcoming experiment, the proposed project aims to validate or discard the discoveries that we published from our first experimental campaign HypHI – Phase 0. The development of innovative methods in machine learning on the hypernuclear spectroscopy experiments will be carried out. Convolutional neural networks and long short-term memory architectures will be investigated first for the non-linear modeling of ion optics in the magnetic spectrometer, and secondly in the tracking of measured particles in the experimental apparatus. The WASA@FRS experiment is scheduled for March 2022. The data analysis will use the machine learning methods explained above. Additionally, the R&D for the electronic readout of the silicon micro vertex detectors and its construction is carried out by this project. The final goal of the project is to tackle the current hypertriton lifetime puzzle and the confirmation of the observation of the nnΛ bound state (a state of two neutrons and a Λ hyperon)..