Study of the tracking performance of a liquid Argon detector based on a novel optical imaging concept

The Deep Underground Neutrino Experiment (DUNE) is a long-baseline accelerator experiment designed to make a significant contribution to the study of neutrino oscillations with unprecedented sensitivity. The main goal of DUNE is the determination of the neutrino mass ordering and the leptonic CP violation phase, key parameters of the three-neutrino flavor mixing that have yet to be determined. An important component of the DUNE Near Detector complex is the System for on-Axis Neutrino Detection (SAND) apparatus, which will include GRAIN (GRanular Argon for Interactions of Neutrinos), a novel liquid Argon detector aimed at imaging neutrino interactions using only scintillation light. For this purpose, an innovative optical readout system based on Coded Aperture Masks is investigated. This dissertation aims to demonstrate the feasibility of reconstructing particle tracks and the topology of CCQE (Charged Current Quasi Elastic) neutrino events in GRAIN with such a technique. To this end, the development and implementation of a reconstruction algorithm based on Maximum Likelihood Expectation Maximization was carried out to directly obtain a three-dimensional distribution proportional to the energy deposited by charged particles crossing the LAr volume. This study includes the evaluation of the design of several camera configurations and the simulation of a multi-camera optical system in GRAIN.

Unsupervised reinforcement learning via state entropy maximization

Reinforcement Learning (RL) provides a powerful framework to address sequential decision-making problems in which the transition dynamics is unknown or too complex to be represented. The RL approach is based on speculating what is the best decision to make given sample estimates obtained from previous interactions, a recipe that led to several breakthroughs in various domains, ranging from game playing to robotics. Despite their success, current RL methods hardly generalize from one task to another, and achieving the kind of generalization obtained through unsupervised pre-training in non-sequential problems seems unthinkable. Unsupervised RL has recently emerged as a way to improve generalization of RL methods. Just as its non-sequential counterpart, the unsupervised RL framework comprises two phases: An unsupervised pre-training phase, in which the agent interacts with the environment without external feedback, and a supervised fine-tuning phase, in which the agent aims to efficiently solve a task in the same environment by exploiting the knowledge acquired during pre-training. In this thesis, we study unsupervised RL via state entropy maximization, in which the agent makes use of the unsupervised interactions to pre-train a policy that maximizes the entropy of its induced state distribution. First, we provide a theoretical characterization of the learning problem by considering a convex RL formulation that subsumes state entropy maximization. Our analysis shows that maximizing the state entropy in finite trials is inherently harder than RL. Then, we study the state entropy maximization problem from an optimization perspective. Especially, we show that the primal formulation of the corresponding optimization problem can be (approximately) addressed through tractable linear programs. Finally, we provide the first practical methodologies for state entropy maximization in complex domains, both when the pre-training takes place in a single environment as well as multiple environments.