PURE > Dspace - Faculdade de Ciências
Permanent URI for this collection
Browse
Recent Submissions
- Ocean-atmosphere interactions : the case of Marine Heatwaves in the North AtlanticPublication . Lopes,Beatriz Pereira; Department of Geographic Engineering, Geophysics and Energy; Gouveia,Célia Marina Pedroso; Oliveira,Ana Patrícia Pires MarquesExcessive greenhouse gas emissions are placing severe pressure on the oceans, which play a very important role in climate regulation. Consequently, the oceans are experiencing an exacerbated warming, leading to an increase in the occurrence of extreme events, such as Marine Heatwaves (MHWs). MHWs are characterized as prolonged periods of anomalously high sea surface temperatures (SSTs). Quantitatively, they are identified when SST anomalies exceed the 90th percentile of a reference climatology for at least five consecutive days. Understanding the occurrence of MHWs is challenging due to their rarity (less than 10 per cent of the surface temperature values, as the 90th percentile indicates) and the limited availability of consistent long-term observational data (last four decades). However, it is now known that they can be modulated by oceanic or atmospheric factors or a combination of both, affecting the ocean's ability to absorb incident solar radiation and dissipate it through currents and the mixing layer, leading to its warming. This study focused on large-scale atmospheric factors driving the MHW occurrence and characteristics in the North Atlantic (an area relatively understudied, especially when compared with the Mediterranean or with the Tasman Seas) from 1982 to 2022, with the objectives of (i) identifying spatial-temporal trends of MHWs, (ii) determining the atmospheric factors contributing to their occurrence, and (iii) exploring their relationship with prevalent climate variability modes. MHWs were identified using ESA Copernicus Climate Change Initiative Sea Surface Temperature data and Hobday et al. (2016)'s method. An event dataset generated using a new event detection algorithm developed by the team at +ATLANTIC CoLAB was used, with the events ranked by severity. Atmospheric data were obtained from ERA 5 reanalysis, including sea-level pressure, geopotential height, temperature, heat fluxes, solar radiation, and wind components. The study also analysed the North Atlantic Oscillation (NAO) index to examine its influence on MHWs. The results show positive trends in MHW frequency, duration, and intensity, especially since 1995, with regional variability. High-pressure systems with weak pressure gradients and associated reduced wind speeds, and increased solar radiation seem to be crucial for the formation of the analysed events. The annual NAO appears to modulate the spatial distribution of MHWs, with its positive phase favouring MHWs in mid-latitude regions, while the negative phase impacts subpolar and tropical regions.
- Study of central and exclusive production of charged Higgs boson pairs with forward proton tagging at the LHC, and projections for the HL-LHCPublication . Batista,Pedro Miguel Carvalho; Department of Physics; Pires,João Ramalho; Hollar,Jonathan JasonBeyond the Standard Model (BSM) scenarios suggest the existence of new particles or interactions to address some of the fundamental questions left unanswered by the Standard Model (SM). The majority of these scenarios suggest the existence of a charged Higgs boson, and its observation would directly imply the existence of BSM physics. With this in mind, this project focused on studying the central and exclusive production (CEP) of a single pair of charged Higgs bosons at the Large Hadron Collider (LHC) Run 3, including projections for the High-Luminosity LHC (HL-LHC) Phase-2, both at a centerof-mass energy of √ s = 13.6 T eV . Additionally, CEP processes are advantageous to study due to the lower number of competing background processes, but also because it’s possible to perform proton tagging with the Precision Proton Spectrometer (PPS) installed at the Compact Muon Solenoid (CMS). This project also takes advantage of the first CMS jet matching High Level Trigger (HLT) designed for studying processes with a four-jet system in the final states. This study used events simulated with the tools MadGraph5 aMC@NLO, Pythia8, and Delphes. This study started with the calculation of the cross sections for the signal and background events. The next part of the analysis focused on the detector configurations for the LHC Run 3 and the HL-LHC Phase-2, and several studies were performed on the impact of pileup interactions, on the acceptance of the PPS detector, on the efficiency of the HLT, and on the statistical sensitivity to exclude the signal process. A comparison between the LHC Run 3 and HL-LHC Phase-2 configurations was also made to highlight the key differences in event analysis. Lastly, remarks were made on possible HLT improvements for the HL-LHC Phase-2, as well as on future studies that could enhance event analysis.
- Extracting n-ary Relations from Biomedical Literature using Deep-Learning TechniquesPublication . Fernandes,João Lucas Matias; Department of Chemistry and Biochemistry; Couto,Francisco José MoreiraThe rapid growth of biomedical literature makes it challenging for researchers to stay up-to-date. Text mining has become essential for efficiently extracting knowledge from unstructured texts. Abstracts offer a focused alternative to full-text articles, but extracting meaningful insights remains difficult. Key tasks such as Named Entity Recognition (NER) and Named Entity Linking (NEL) face issues like ambiguous terminology, entity variability, and incomplete knowledge bases, especially when handling novel or NIL (not-in-lexicon) entities. Relation Extraction (RE) systems also face challenges, including limited scope, lack of interpretability, and a focus on binary relations that do not fully capture complex biomedical interactions. This thesis introduces a small gold-standard dataset created by expanding 31 abstracts from the 600-document BioRED corpus. The dataset adds CellTypeOrAnatomicalConcept and NIL entities, serving as a resource to test and improve the Biomedical Entity Annotator (BENT) tool for NER and NEL. It also enables the extension of relation extraction from binary to n-ary relations, starting with ternary relations. Compared to BioRED, NER performance was generally lower across most entity types, while NEL showed particularly low scores for GeneOrGeneProduct, CellTypeOrAnatomicalConcept, and NIL entities, reflecting the challenges of novel entity annotation. For n-ary relation extraction, the K-RET system, built on BERT-based models, was employed with SciBERT and BioMedBERT. In the binary setting, the system achieved an F1-score of 0.775 compared to BioRED’s 0.7562. Ternary relations were evaluated against BioRex, a state-of-the-art study, yielding F1- scores of approximately 0.65. Despite being lower than BioRex, the results provide a promising baseline for n-ary relation extraction across a broader set of entity types.
- Quantum Approaches to Portfolio OptimizationPublication . Matias,Rodolfo Crispim Inácio; Department of Physics; Souto,André Nuno Carvalho; Evans,Guiomar Gaspar de AndradeThe relentless quest for optimized financial portfolios has led to the exploration of advanced computational techniques capable of navigating financial markets’ complex and dynamic landscape. This dissertation presents a solution for portfolio optimization by harnessing the synergistic potential of quantum computing and machine learning. We propose a hybrid framework that integrates quantum algorithms’ optimization capabilities with machine learning’s predictive accuracy to address the multifaceted challenges of portfolio management. The study begins with a comprehensive review of traditional and contemporary portfolio optimization methods and an in-depth analysis of quantum computing principles relevant to optimization problems. Next, we examine machine learning models commonly used for forecasting financial time series, which serve as critical inputs for the quantum optimization process. The proposed solution is evaluated through simulations and real-world financial data to demonstrate its efficacy in achieving optimized asset allocations with enhanced risk-adjusted returns. This research contributes to the theoretical advancement of financial optimization techniques and provides practical insights for investors and portfolio managers seeking to leverage emerging technologies for strategic decision-making.
- Identifying Interpretable Imaging Biomarkers for Survival prediction and NAT response assessment in PDACPublication . Araújo,Marta Vieira; Department of Physics; Matela,Nuno Miguel de Pinto Lobo e; Marquering,Henk A.Neoadjuvant therapy (NAT) has significantly improved outcomes in PDAC, yet patient responses vary considerably, highlighting the need for reliable predictive biomarkers. We developed an interpretable framework using 3D convolutional autoencoders (CAEs) to learn pre-NAT latent features for survival modeling and difference-map features (pre-post) for response modeling in locally advanced pancreatic cancer (LAPC), with clustering and explainability to relate imaging phenotypes to clinical/radiomics data. CTs were preprocessed, a CAE was pre-trained on the PANORAMA dataset (n=676) and finetuned on the LAPC cohort (LAPC, n=127). Survival pipeline: three CAE configurations were trained from pre-NAT scans: (1) scans only; (2) scans+clinical; (3) scans+clinical+Pre-NAT radiomics (PCA). Response pipeline: three CAE configurations from pre- and post-NAT scans were trained: (1) scans only; (2) scans+clinical; (3) scans+clinical+delta radiomics (PCA). Performance was estimated with stratified 5-fold CV. We clustered embeddings from the best CAE models using K-means, tested associations with clinical variables, PCA radiomics, and CA 19-9, and explained cluster drivers with UMAP plus a random-forest surrogate and SHAP. In both pipelines, adding clinical+radiomics during training/fine-tuning improved the latent space geometry (higher Silhouette score, lower Davies-Bouldin). Post-training, K-means on embeddings showed: (i) response pipeline – moderate separation but no association with CA 19-9 change or survival; clusters and SHAP aligned with delta radiomics PCs, not routine clinical variables; (ii) survival pipeline – coherent pre-NAT latent geometry with clusters that were non-prognostic; clusters and SHAP aligned with pre-NAT radiomics PCs, again weakly with clinical variables. The proposed CAE framework yields interpretable, spatially localized representations that capture treatment effect and partly overlap with handcrafted radiomics; however, unsupervised clusters from these embeddings did not translate into prognostic groups in this cohort. Next step: outcomeguided multimodal learning (shared latent space for survival and response), CA 19-9 trajectory modeling, and aligned pre/post CTs with tumor-focused analysis to convert these interpretable signals into actionable biomarkers.
- A Dynamical System Approach to Uni and Multi-Variate Weather Analysis over Portuguese TerritoryPublication . Belime,Timothée Franck; Department of Physics; Nunes,Ana Maria Ribeiro Ferreira; Câmara,Carlos do Carmo de Portugal e Castro daThe main objective of this work is to investigate and reproduce some of the existing research in the analysis of weather dynamics with an approach based on dynamical system. Most studies of climate extremes rely on statistical approaches or machine learning, but chaotic dynamics can provide a deeper understanding of atmospheric behavior. The method can easily be extended to multivariate systems. The general observation that the atmosphere displays chaotic dynamics leads us to numerical analysis of some chaotic attractors. The Lorenz attractor is used extensively in this study. We use these systems to see how much information can be obtained from the time series of a few metrics derived from a discrete trajectory in phase space. This trajectory must be long enough to represent the system globally, meaning that it must be longer than the characteristic time of the system. After probing with numerically integrated systems, we transpose the acquired knowledge to practical applications of real data over Portuguese territory, from 1979 to 2020. We search for the physical meaning behind each metric and investigate if they can provide a better understanding of weather extremes and compound events. Practical applications are explored by crossing the data with registered wildfire activity in the same period of time. We mainly focus on temperature and wind speed and assessment is made of their role in major wildfires. According to their dynamical signatures, warm-windy compound extremes fall into two different classes, one with a more predictable character, and the other less predictable.
- Decomposition of inflationary systems into scale-free networksPublication . Ricardo José Velho,Ramos; Department of Physics; Pires Da Cruz,João; Cristovão,DiasDue to their overly complex nature, real-world networks cannot be understood through typical mathematic tools, such as diferential equations, for that reason we use networks and analyse them through algorithms to understand them. This work focuses on decomposing scale-free networks into Barabási-Albert networks to better understand them. We are able to make an analysis of these networks through their decomposition and to identify redundant data to reduce size and complexity. We use tools such as autoencoders to create the algorithms and to go from integer-dimensional spaces to networks and vice versa. An integer dimension space, such as a Euclidian space is a projection of a network, for the case of BA (Barabási-Albert) networks, in the limit they tend to a one-dimension space, or a curved line. We can make use of this notion to try and recreate the original integer dimension space from the created BA networks. We can tested these concepts using a network whose projection is a Euclidian space and a market network, whose projection is a fractal space, which are created with the help of autoencoders. Autoencoders are also used to try and understand their differences and to validate the results. An accuracy analysis will be made in order to identify redundant data in the networks and possibly simplify them. This work can help to simplify and better comprehend real-world networks, such as the stock market and social networks.
- Automated reporting framework for task-specific biomechanical analysis and performance assessment during forward braking and backward acceleration in elite athletesPublication . Oliveira,António Manuel Gaspar; Department of Physics; Matela,Nuno Miguel de Pinto Lobo e; Veloso,António PrietoAnterior Cruciate Ligament (ACL) injuries are among the most common and debilitating sports injuries, posing potential career-threatening consequences and carrying a high risk of re-injury. Movements involving sudden deceleration are particularly relevant in the context of ACL injuries, as deceleration actions are common high-impact tasks where non-contact ACL injuries typically occur. Accurate evaluation of ACL injuries is essential to determine when an athlete is ready to return to sport. Forward braking and backward acceleration is a functional test that mimics the abrupt deceleration common in various sports. This task replicates the real-life physical demands of abrupt decelerations and can simulate the stress on the knee encountered during such activities. Manual analysis of movement data can be time-consuming and prone to human error. An automated report framework offers several advantages in the context of motion capture and subsequent analysis, improving efficiency and consistency. Therefore, the primary objective of this work was to develop an automated framework specific to the biomechanical analysis of forward braking and backward acceleration in elite athletes. Using Qualisys Track Manager (QTM) and its Project Automation Framework (PAF) tools, two automation packages were created: one for marker-based and another for markerless motion capture data. These frameworks streamline the entire workflow, from motion capture to the generation of a customized biomechanical report. The frameworks automatically handle tasks such as model building, event detection, data processing, and report generation. The resulting report includes kinematic, kinetic, and electromyography (EMG) data specific to this task, and was designed in consultation with a professional who regularly uses this functional test. This automated approach eliminates the need for manual data processing, reducing errors and improving workflow efficiency. The final framework offers a practical tool for assessing this task in elite athletes, with applications in performance and injury evaluations.
- Combinatorial and algebraic properties of Lucas-AnaloguesPublication . Hipólito,João Tomás Coelho Lhansol Urbano; Department of Mathematics; Torres,Maria Manuel CorreiaThis thesis investigates a generalization of Lucas polynomials through combinatorial and algebraic approaches, including a multivariable extension. Each chapter addresses a unique perspective of these polynomials, building on foundational recurrence relations and exploring their applications in combinatorics and algebra. Chapter 2 develops a combinatorial interpretation of Lucas polynomials, focusing on their representation through tiling problems and other discrete structures. This perspective highlights the connections between recurrence relations and enumeration techniques. Chapter 3 exposes an algebraic framework, examining the properties of Lucas polynomials through transformations and identities. This chapter establishes links with broader algebraic objects and explores how recurrence relations encode these structures. Chapter 4 extends the scope to multivariable Lucas polynomials, presenting a rigorous generalization that integrates combinatorial and algebraic principles. This chapter lays the groundwork for future exploration of multivariable polynomial systems and their applications in higher-dimensional combinatorics. Chapter 5 introduces a recent approach to generating rational functions associated with Lucas polynomials. By transforming previously known recurrence relations, we derive explicit formulas and construct generating functions, offering new insights into these polynomials. This thesis provides a cohesive framework for understanding and extending Lucas polynomials, contributing with ideas for new methodologies and interpretations with relevance to both discrete mathematics and algebraic combinatorics.
- Importância da fauna de profundidade na dieta de aves marinhas pelágicasPublication . Ribeiro,Jéssica Andreia Pacheco; Departamento de Biologia Animal; Dias,Maria Ana de Figueiredo Peixe; Silva,Mónica Sérvulo Correia Carneiro daThe deep-sea fauna, composed mainly of fish, cephalopods and crustaceans, inhabits the mesopelagic (200 m and 1000 m depth) and bathypelagic (1000 m and 4000 m depth) layers of the ocean. These layers are vast and difficult to access and are therefore the least studied in the ocean. Although the deep-sea species are found in deep waters during the day, likely to avoid predation, some migrate at night towards the surface to feed, performing diel vertical migration (DVM), becoming more accessible to surface predators, such as pelagic seabirds. This study aims to characterize, for the first time, the importance of deep-sea fauna in the diet of bird species belonging to the families Oceanitidae, Hydrobatidae and Procellariidae, which include all smaller Procellariiformes. In order to characterize this importance, a meta-analysis was carried out based on an extensive literature review of scientific articles, reports and books with diet data. I identified and analysed 177 studies on the diet of 126 target species published between 1969 and August 2024. The results revealed a lack of knowledge about the diet of these birds, not only in terms of the target species, but also in terms of the level of detail of the studies published so far. This study showed that deep-sea species, mainly fish, are key components of the Pacific and Antarctic trophic food webs, but also in the Atlantic, being an important fraction of the diet of many pelagic seabirds that occur in these oceans. Deep-sea fauna is particularly important component of the diet of seabirds that are threatened with extinction, which highlights the importance of this fauna and the need to conserve this resource that is extremely understudied but is already being targeted as commercially relevant.