41199 research outputs found
Sort by
Lower mass bounds on FIMP Dark Matter
Get PDFOver the last decades, strong evidence for dark matter (DM) has been accumulated. However, its nature is still unknown in fundamental physics. A cold relic produced via thermal freeze-out has been the leading paradigm for a long period, both for its simplicity and versatility to include candidates offered by extensions of the Standard Model of particle physics. Despite a large experimental
effort to detect these candidates, no conclusive evidence has been found yet. The outcome of these negative searches has been motivating the community to explore
alternative paradigms to the freeze-out. One of these is the freeze-in scenario, first introduced by Hall et al. (2009), which shares the advantages of the freeze-out paradigm but in which DM is produced out of thermal equilibrium. In this work, after a review of the evidence of DM and the Boltzmann equation formalism, we present the main features of the freeze-in paradigm proving its efficiency in reproducing the relic density. Despite the usually smaller couplings between DM and visible matter in the freeze-in scenario, frozen-in DM candidates, called Feebly Interacting Particles (FIMPs), can have interesting observational features through which freeze-in models can be constrained. Among them, stringent bounds can come from structure formation data. In particular, too light FIMP DM is in tension with small-scales structures such as the Lyman- forest. We develop a model independent procedure to constrain the parameter space of a FIMP model and to extract the value of the minimum DM mass allowed, which is of crucial importance in model building. The methodology is based on the comparison of the linear matter power spectrum, computed with the CLASS code, from the non-thermal FIMP phase-space distribution, with the limit power spectrum obtained from a Warm Dark Matter (WDM) model, found by M. Viel et al. (2017).
To test our procedure and compare with the literature, we consider three simple scalar FIMP DM toy models involving renormalizable interactions with hypothetical scalars belonging to the thermal bath. These are benchmarks models in which DM production is dominated by decays and scatterings and can be used to draw general conclusions on FIMP DM produced via these mechanisms.
The developed procedure can also be applied to concrete freeze-in models and frameworks in which the DM production occurs in modified cosmologies and can be generalized to include also additional external bounds on the linear matter power spectrum
Leptonic contributions to muon-electron scattering at NNLO
Get PDFConfirmation of the long-standing muon g-2 discrepancy requires both experimental and theoretical progress. On the theory side, the hadronic corrections are under close scrutiny, as they induce the leading uncertainty of the Standard Model prediction. In this context, the MUonE experiment has been proposed at CERN to provide a new direct determination of the leading hadronic contribution to the muon g-2, measuring the differential cross section of muon-electron scattering. As the shape of this differential cross section must be measured with a systematic uncertainty of 10 ppm or better, an analogous precision is required in its theoretical prediction. In this thesis we calculated the leptonic QED corrections to muon-electron scattering at next-to-next-to-leading order (NNLO). They arise from two-loop diagrams with leptonic vacuum polarization insertions in the photon propagator. Non-factorizable two-loop diagrams were
computed using the dispersive approach. These results will play a crucial role in the analysis of MUonE's data
Deep Reinforcement Learning methods for StarCraft II Learning Environment
Get PDFReinforcement Learning (RL) is a Machine Learning framework in which an agent learns to solve a task by trial-and-error interaction with the surrounding environment.
The recent adoption of artificial neural networks in this field pushed forward the boundaries of the tasks that Reinforcement Learning algorithms are able to solve, but also introduced great challenges in terms of algorithmic stability and sample efficiency.
Game environments are often used as proxies for real environments to test new algorithms, since they provide tasks that are typically challenging for humans and let the RL agents make experience much faster and at a cheaper price than if they were to make it in the real world.
In this thesis state-of-the-art Deep Reinforcement Learning methods are presented and applied to solve four mini-games of the StarCraft II learning environment.
StarCraft II is a real-time strategy game with large action and state space, which requires learning complex long-term strategies in order to be solved; StarCraft mini-games are auxiliary tasks of increasing difficulty that test an agent's ability to learn different dynamics of the game.
A first algorithm, the Advantage Actor-Critic (A2C), is studied in depth in the CartPole environment and in a simple setting of the StarCraft environment, then is trained on four out of seven StarCraft mini-games.
A second algorithm, the Importance Weighted Actor-Learner Architecture (IMPALA), is introduced and trained on the same mini-games, resulting approximately 16 times faster than the A2C and achieving far better scores on the two hardest mini-games, lower score on one and equal score on the easiest one.
Both agents were trained for 5 runs for each mini-game, making use of 20 CPU cores and a GPU and up to 72 hours of computation time.
The best scores from the 5 runs of IMPALA are compared with the results obtained by the DeepMind team author of the paper StarCraft II: A New Challenge for Reinforcement Learning, which report the best scores out of 100 runs that used approximately two orders of magnitude more of training steps than our runs.
Our IMPALA agent surpasses the performance of the DeepMind agent in two out of the four mini-games considered and obtains slightly lower scores on the other two
Convolutional Neural Network data analysis development for the Large Sized Telescope of CTA and broadband study of the blazar 1ES 1959+650
Get PDFThis thesis representes the summary of the activities I performed in the field of very-high-energy (VHE) gamma-ray astronomy, and it is articulated in two distinct parts.
VHE gamma-ray astronomy is the science studying the photons emitted at TeV energies in cataclysmic events of the Universe. When these highly-energetic gamma-rays interact with the high atmosphere of the Earth, they produce cascades of particles that emit flashes of Cherenkov light. Imaging Atmospheric Cherenkov Telescopes (IACTs) detect these flashes and convert them into shower images than can be analyzed to extract the properties of the primary gamma ray. Dominating background for IACTs is constituted by images produced by cosmic hadrons, with typical noise-to-signal ratios of several orders of magnitude.
The standard machine learning technique adopted to separate gamma-rays from hadrons is based on a set of parameters extracted from the images. On the other hand, state-of-the-art Deep Learning techniques such as Convolutional Neural Networks could enhance the analysis, since they are able to autonomously extract features from raw images, exploiting the pixel-wise information irreversibly washed out during the parametrization process.
In the first part of this work, I present the development of a novel approach to the analysis of the images produced by a new-generation IACT, the Large Sized Telescope (LST) of CTA. I use Convolutional Neural Networks to separate gamma rays from the dominating background of cosmic hadrons and to reconstruct their properties, showing that this technology performs remarkably better than the standard analysis technique.
In the second part, I present the study of the emission of the blazar 1ES 1959+650, that is an ive galaxy emitting two extremely energetic jets of plasma in the outer space. First, I describe the analysis of the VHE gamma-ray emission observed during 2017 by the MAGIC IACTs, then I perform a multiwavelength characterization of the activity exhibited by the source between 2016 and 2020, modeling its broadband emission. The results of this investigation give insights on the mechanisms at work in the jets, preluding to a wider study that will probe the physics of this kind of sources to a deeper degree
