Publikationer från Uppsala Universitet
Not a member yet
127603 research outputs found
Sort by
Structured models for scientific machine learning : From graphs to kernels
No full textThis thesis investigates the reciprocal relationship between science and machine learning, showing how embedding scientific principles within machine learning models enhances accuracy and interpretability in complex scientific domains. Through five contributions, this work addresses challenges spanning molecular modeling, fluid dynamics, and graph-based learning, illustrating how scientific insights can guide model development and improve performance across diverse applications. A central focus is developing models that directly incorporate physical symmetries and laws into their structure, creating novel, scientifically grounded approaches. For instance, the GeqShift model leverages E(3)-equivariant graph neural networks to predict NMR spectra with a significant accuracy boost by capturing three-dimensional molecular structures. Similarly, our SE(2)-equivariant graph neural network models rotational and translational symmetries to enhance data efficiency and performance in fluid dynamics simulations, demonstrating the strength of symmetry-aware models in complex physical domains. This thesis advances graph-based machine learning by developing physics-informed models. For example, we introduce a subsampling model that incorporates principles from the classic Ising model of magnetism, introducing a novel approach to graph subsampling that enhances tasks like graph explanation and mesh sparsification. Building on these graph-based techniques, we also present an approach to nonnegative matrix factorization (NMF), leveraging graph structures to accelerate low-rank factorization. In addition to symmetry-aware frameworks, we introduce the elliptical process, a flexible extension of the Gaussian process that adapts to non-Gaussian noise. This innovation allows the model to learn noise characteristics directly from data, producing robust predictions that address a broad spectrum of real-world challenges. These contributions underscore the dynamic exchange between scientific principles and machine learning, illustrating how physical knowledge enhances model performance and inspires new solutions. This thesis establishes a foundational framework for advancing scientific machine learning, paving the way for future breakthroughs in the field
Understanding Wintertime Pan-Atlantic Compound Weather Extremes : Statistical and Dynamical Insights
No full textThis thesis investigates the connections between extreme winter weather in North America and Europe through large-scale atmospheric patterns. The focus is on the co-occurrence of North American cold spells and European wet or windy extremes. The European extremes occur more frequently than normal during North American cold spells, yet the specific timing and location is dependent on the location of the cold spells. Three North American regions are identified where cold spells coincide with specific footprints of European extremes: 1. Eastern Canada: Cold spells predominantly precede wind extremes over the British Isles and Northern Europe. 2. Central Canada: Cold spells are predominantly preceded by wind and precipitation extremesin Iberia. 3. Eastern United States: Cold spells are both preceded and followed by wind and precipitationextremes in Iberia. These cold spells are associated with an intensified and more zonal North Atlantic jet stream, increasingthe frequency, and in some cases, intensity of extratropical cyclones affecting Europe. This in turn leads to an increased occurrence of wet and windy extremes. However, cold spells in a single North American region can arise due to different large-scale atmospheric circulations. Stratifying the cold spells by wave activity flux intensity indeed reveals differing atmospheric dynamics, referred to as "dynamical pathways". Further analysis using weather regimes and Fourier decomposition identifies five dynamical pathways, which produce with varying frequencies, cold spells across all three North American regions. Three pathways, associated with the North Atlantic Oscillation (NAO), result in specific footprints of wind extremes in western Europe. The remaining two pathways, associated with Scandinavian blocking, do not produce such extremes. The regional preferences of these pathways provide a mechanistic explanation for the observed spatial and temporal relationships between cold spells and European extremes. Climate models are found to effectively replicate these pathways, though they underestimate the frequency of NAO+ pathways. Under a high-emission future climate scenario, the frequency and dynamics of the pathways are not projected to change significantly. However, the model simulations point to both North American cold spells and western European wind extremes diminishing in severity
ESG:s inflytande vid företagsförvärv : En kvantitativ studie om hur ESG-poäng påverkar aktiemarknadens reaktion vid annonsering av företagsförvärv
No full textThrough an event study, this research essay examines the stock market's reaction to mergers and acquisitions, focusing on ESG (Environmental, Social and Governance) as a potential value-creating factor. The study analyzes 113 transactions conducted by Swedish publicly traded companies between 2015 and 2023. Despite the intent to create value, many mergers result in failures that instead destroy value. Using regression analysis and an event study method, the hypothesis that higher ESG scores lead to positive abnormal returns for the acquiring company is tested. In recent years, interest in sustainability has grown rapidly, with an increasing focus on integrating ESG considerations into corporate operations and financial reporting. The study finds no statistically significant abnormal returns on the Swedish stock market that can be linked to ESG as a value-creating variable, despite previous research supporting the long-term value-enhancing potential of ESG.Genom en eventstudie undersöks aktiemarknadens reaktion vid företagsförvärv med fokus på om ESG (Environmental, Social and Governance) kan skapa värde. Studien analyserar 113 transaktioner utförda av svenska noterade bolag mellan 2015 och 2023. Trots syftet att skapa värde resulterar många företagsförvärv i misslyckanden som istället leder till värdeförstörande som förluster, ineffektivitet och minskat aktieägarförtroende. Genom att använda regressionsanalys och en eventstudie testas hypotesen om att högre ESG-poäng leder till positiv avvikelseavkastning för det förvärvande företaget. Samtidigt har intresset för hållbarhet vuxit snabbt under senare år, med ett allt större fokus på att integrera ESG i företagens verksamheter och finansiella rapporter. I studien observerades ingen signifikant avvikelseavkastning hos den svenska aktiemarknaden som kan kopplas till ESG som värdeskapande variabel, trots stöd i tidigare forskning för ESG:s långsiktiga värdeskapande.
Microsimulation models on child and adolescent health : a scoping review protocol
No full textIntroduction Microsimulation models are computer-based models, which can be employed to simulate the behaviour of microagents, such as children and adolescents, to understand the potential behavioural and economic effects of health interventions or policies. As a result, these models can be useful tools to help guide decision-making. A comprehensive review of the literature on child and adolescent microsimulation models has yet to be undertaken. Moreover, an evaluation of the quality of the existing models can be useful to understand their strengths and limitations and thus inform the development of future models. The aim of this scoping review will be to retrieve, synthesise and critically appraise the literature on existing microsimulation models focused on child and adolescent health.Methods and analysis We will conduct a scoping review using established methods. We will search PubMed (until 23 September 2024), Embase (until 18 September 2024), CINAHL (until 9 September 2024), PsyclNFO (11 September 2024), EconLit (until 9 September 2024) and Scopus (until 10 September 2024), with an update closer to the time of manuscript submission. We will also undertake snowballing, Google searches and searches on specific journal (eg, International Journal of Microsimulation) and websites (eg, https://www.microsimulation.ac.uk/) to complement database searches. We will extract relevant data on all studies retrieved and use the Quality Assessment Reporting for Microsimulation Models checklist to assess the reporting quality of each model. We will use a narrative synthesis with summary tables to describe our findings. Findings will be synthesised by type of health condition, if/where possible.Ethics and dissemination Given that primary data will not be collected in this study, research ethics approval is not required. We will present our findings at relevant conferences and publish our results in an appropriate peer-reviewed academic journal. In addition, we will use this information to guide the development of a microsimulation model on child and adolescent health for use in the Swedish context.Registration details https://osf.io/a8txn
Inulin as a Biopolymer; Chemical Structure, Anticancer Effects, Nutraceutical Potential and Industrial Applications : A Comprehensive Review
No full textInulin is a versatile biopolymer that is non-digestible in the upper alimentary tract and acts as a bifidogenic prebiotic which selectively promotes gut health and modulates gut-organ axes through short-chain fatty acids and possibly yet-to-be-known interactions. Inulin usage as a fiber ingredient in food has been approved by the FDA since June 2018 and it is predicted that the universal inulin market demand will skyrocket in the near future because of its novel applications in health and diseases. This comprehensive review outlines the known applications of inulin in various disciplines ranging from medicine to industry, covering its benefits in gut health and diseases, metabolism, drug delivery, therapeutic pharmacology, nutrition, and the prebiotics industry. Furthermore, this review acknowledges the attention of researchers to knowledge gaps regarding the usages of inulin as a key modulator in the gut-organ axes
Charging strategies and battery ageing for electric vehicles: A review
No full textIntroducing electric vehicles in society requires access to charging infrastructure and a robust electric grid. This development concerns strategic planning of policymakers. However, much uncertainty still exists about how new charging strategies affect the vehicle batteries, due to the fast development. This review article provides an overview of recent literature on how electric vehicle batteries are aged during different charging strategies, such as conductive charging, inductive charging, and battery swapping. This study finds that some charging conditions, such as fast charging at low temperatures, degrade batteries faster. Battery ageing is a non-linear process and depends on, for example, temperature, charging current, and state-of-charge. The high charging rates strongly influence battery degradation. It is concluded that there is a trade-off between faster charging and a longer battery lifetime. There is an interest in vehicle-to-grid to add revenues and grid flexibility. However, the related battery degradation needs to be further investigated. A key challenge in the decision-making process is to plan for charging infrastructure suitable for electric vehicle owners while contributing to the long life of electric vehicle batteries and ensuring socially acceptable and economically viable solutions. More interdisciplinary research in this field is recommended to support the clean energy transition.Laddningsstrategins inverkan på elbilens batterisystem (LIEB)
Long-term energy scenario ranking with MCDA analysis : The case of Reunion Island
No full textIsland territories rely heavily on fossil fuel resources, and transitioning to other energy sources is essential for their progress. To accomplish this, it is imperative to utilize local renewable energy sources. In this study, an analysis of energy planning for Reunion Island has been conducted where Multi-Criteria Decision Analysis (MCDA) methodologies have been used to evaluate the sustainability of energy scenarios for 2050. This evaluation is carried out considering criteria encompassing technical, economic, environmental, and social aspects of the scenarios. Further, perspectives of local actors were considered in the evaluation of these criteria. The results indicate that the greenhouse gas emissions and job creation criteria are considered to be of utmost significance, whilst technical criteria were regarded as the least significant by these local stakeholders. PROMETHEE II and TOPSIS MCDA analysis reveal that the Combined scenario, a scenario where all local energy resources are exploited to their maximum, is the preferred electricity generation scenario. The findings are used to formulate policy recommendations for island planners, helping them to adjust the island's current energy strategies. This study also serves as a valuable resource for other non-interconnected islands undergoing an energy transition
Depression in adolescence and the understanding of health : A phenomenographic study
No full textHealth is multifaceted, with divergent interpretations in diverse cultures and groups of individuals. The ways individuals understand health might aid in developing future interventions. There is scant knowledge on how adolescents with depression conceptualise health. A descriptive qualitative design with a phenomenographic approach was used to describe the different ways adolescents with depression conceptualise health. Interviews were performed with adolescents 13-17 years old (n = 33) who participated in a randomised controlled trial evaluating the effects of aerobic group exercise versus leisure group activities for adolescents with depression. The results were interpreted into four metaphors to embody the understanding of health as described by adolescents with depression: 1) establishing sound routines by managing everyday life, 2) connecting to others by having access to social resources, 3) managing depression symptoms by having control over the mental illness, and 4) attaining inner drive by experiencing joy in everyday life. The various conceptions of health among adolescents with depression provide valuable insights for enhancing evidence-based treatments with person-centred care. Key aspects include establishing routines, fostering connections, finding strategies for symptom control, and incorporating joy through exercise. Given that adolescents highlighted these aspects as essential to health, future research could explore individualised health promotion, particularly focusing on routine-building, social connections, or finding an inner drive as an add-on to evidence-based treatments for adolescent depression
Improving the thermoelectric performance of scandium nitride thin films by implanting helium ions
No full textIon implantation is a widely used technique to introduce defects in low-dimensional materials and tune their properties. Here, we investigate the thermoelectric properties of scandium nitride thin films implanted with helium ions, revealing a positive impact of defect engineering on thermoelectric performance. Transport properties modeling and electron microscopy provide insights on the defect distribution in the films. The electrical resistivity and Seebeck coefficient increase significantly in absolute values after implantation and partially recover upon annealing as some of the implantation-induced defects heal. The thermal conductivity decreases by 46 % post- implantation due to the formation of extended defects and nanocavities. Consequently, the thermoelectric figure of merit zT doubles for the sample annealed at 673 K. These findings highlight the potential of controlled ion implantation to enhance thermoelectric properties in thin films, paving the way for further optimization through defect engineering