Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 29: Deep Learning II

T 29.2: Vortrag

Dienstag, 26. März 2019, 16:15–16:30, H06

Antiproton to proton ratio determination using deep neural networks with AMS-02 — •Sichen Li — I. Physikalisches Institut B, RWTH Aachen, Germany

AMS-02 is a high precision detector for charged cosmic rays installed on the International Space Station. A discrepancy occurs between the current observation for antiproton to proton ratio and the prediction from collisions of ordinary cosmic rays, which could be explained by dark matter annihilation or other astrophysical phenomena. To test models precisely, an antiproton to proton ratio in a larger energy range is required.

Charge confusion occurs when a proton is mis-reconstructed as an antiproton. The reasons for this are interactions with AMS materials and detector resolution. To extend the energy range of the measurement, a rejection power against charge-confused protons in excess of 1 in 1 million is needed, due to the tiny fraction of antiprotons in cosmic rays.

We build a deep neural network to improve the separation for charge confusion from interactions. With this approach, we have a good potential to extend energy range for antiproton to proton ratio.