Dipòsit Digital de la Universitat de Barcelona

Dades

Docència

Institucional

Mediateca

Programari

Recerca

Tesis Doctorals

Treballs de l'alumnat

El Dipòsit Digital de la Universitat de Barcelona és el repositori institucional que conté en format digital els materials derivats de l'activitat docent, investigadora i institucional de la comunitat universitària.

Enviaments recents

Epidemic spreading in hyperbolic geometry
(2026-06) Rojo Mérida, Adrià; Boguñá, Marián
Epidemic spreading on complex networks is strongly influenced by the underlying network structure. In this work we simulate stochastic SIR epidemics on networks generated with the S¹H² model and compare the results with non-geometric models, including Erdős–Rényi, Barabási-Albert and Configuration model. Using distance-dependent infection rates from the H² embedding and native S¹H² coordinates, we analyse the epidemic dynamics through the variability of the recovered nodes and the spatio-temporal distribution of infection events. We find that hyperbolic networks exhibit lower epidemic thresholds than non-geometric networks, consistent with a hierarchical radial placement of hubs in the centre of the hyperbolic plane enhancing the epidemic transmissibility. The average hyperbolic distance reached by the infection follows a universal rescaled behaviour largely independent of the network topology. Furthermore, only the hyperbolic model displays an approximately coherent wavefront-like propagation through latent space, whereas non-geometric networks exhibit almost instantaneous spreading with weak distance dependence. These results show that hidden hyperbolic geometry qualitatively amplifies epidemic propagation and shapes the dynamical organisation of the spreading process.
A machine learning approach for CP asymmetry detection
(2026-06) Rodríguez Reus, Àlex; Marin Benito, Carla; Souza de Almeida, Felipe Luan
A novel model-independent approach to search for CP violation based on machine learning is presented, together with two standard approaches in the literature: the energy test and the Miranda method. The latter and the novel technique are tested using simulated D± → K∓K±π± samples of 200 pseudo-experiments generated to analyze the phase space of these decays. Results from the machine learning approach are compared against those of the Miranda method, which employs a binning scheme adapted to this specific dataset. While this binning approach is shown to be a superior detection technique for the considered dataset, potential enhancements to the machine learning strategy are suggested to achieve higher statistical significance than the Miranda method
Quantifying Non-Newtonian Reasoning in Mechanics
(2026-06) Rodríguez Calderón, Nuria; Vilà i Arbonès, Anna Maria
This thesis studies non-Newtonian reasoning in mechanics through Force Concept Inventory responses. It combines a large-scale dataset with a reduced ten-item survey administered to Physics and Engineering students in Barcelona. Answers are classified as Newtonian, Aristotelian, or residual, and are analyzed with response frequencies, Bao–Redish model analysis, and transition matrices. Overall, Aristotelian reasoning persists beyond instruction and academic level, and student responses are better described as mixed states than pure models.
Testing the performance of image deconvolution for crowded field photometry of Cepheid stars
(2026-06) Rocamora i Martorell, Ferran; Courbin, Frédéric
The systematic errors in the distance-ladder measurement of the Hubble constant (H0) have been under close scrutiny in recent years. In this work, we test a method not previously applied in this context, image deconvolution with STARRED, to examine the effects of crowding on photometry and, consequently, on the distance-ladder measurement of H0. Using simulations of crowded fields from HST images of NGC 4258, we assessed the possible bias over a range of signalto-noise ratios. The point spread function used in the simulations was extracted from another HST field, NGC 136. Ten thousand injections were performed for each crowding regime with the recovered amplitudes showing small relative errors compared with the injected amplitudes, even in low signalto-noise regimes. Our results suggest that this method is effectively unbiased within tested regimes, which could be a substantial improvement in constraining possible Cepheid photometry systematic errors.
Memristive devices for improving AI energy efficiency
(2026-06) Rivero Artalejo, Elena; Cirera Hernández, Albert; Aran Vilà, Pere
Resistive switching memories (ReRAM) are emerging devices that can exhibit both volatile and non-volatile behaviour, making them promising candidates for next-generation memory and neuromorphic computing applications. In this work, inkjet printed Ag/h-BN/Au memristive devices based on an hexagonal boron nitride dielectric, which is a two-dimensional material, are experimentally investigated. Electrical characterization is performed through current–voltage measurements, where reproducible switching between high- and low-resistance states is observed, associated with the formation and rupture of conductive filaments driven by ionic migration. Both volatile and non-volatile regimes are obtained depending on the applied electrical conditions. The role of current compliance in filament stabilization during the switching process is considered. In addition, the devices are analysed in terms of cycle-to-cycle and device-to-device variability to evaluate their reproducibility and reliability. The results highlight the stochastic nature