Search for tau neutrinos using multi-PMT sensors

Details

Supervisors

de Wasseige, Gwenhaël

Faculty

Faculté des sciences

Degree label

Master [120] en sciences physiques, à finalité approfondie

Abstract

enThis Master's thesis investigates whether data obtained by multi-PMT DOMs in KM3NeT can be used to distinguish tau neutrinos from the other two neutrino flavours. Monte Carlo simulations were employed to obtain events, which were then made to resemble the standard KM3NeT output. With this dataset, different methods were employed in order to ascertain whether tau-non-tau separation was feasible or not. Data visualisation tools such as t-SNE, and UMAP were used to visualise the five dimensional parameter space. Following this, various methods such as parameter cuts, manual cuts, machine learning algorithms, and multiple-threshold data, were utilised to attempt to distinguish tau events from non-tau events. Parameter cuts obtained the best overall results, closely followed by the machine learning algorithm. As such, the results indicate that the current task is indeed feasible, with a likely optimal strategy involving a specially crafted algorithm in combination with parameter cuts.