As autonomous drone swarms are increasingly considered for military applications, ensuring that humans retain meaningful control over their behavior becomes both a legal and ethical imperative. The concept of Meaningful Human Control (MHC) has emerged as a central framework for addressing this challenge, yet it remains largely untested in concrete operational contexts, and the parallel field of human-swarm interaction has developed without systematic reference to MHC's requirements. This study bridges that gap. Using a simulation-based user study in a military surveillance scenario, five participants operated a semi-autonomous drone swarm through a Graphical User Interface. Their interactions were recorded and analyzed using the Critical Decision Method, and the resulting control failures were mapped to a consolidated MHC framework. The study makes two principal contributions. First, it identifies 32 specific control limitations across six themes and maps these to the technical and socio-technical layers of a consolidated MHC model, demonstrating concretely where the system falls short of MHC's conditions and where those conditions are themselves underspecified. Second, it introduces the distinction between Hard and Soft MHC. Hard MHC refers to binary requirements and functions that must exist for meaningful control to be possible at all. Soft MHC refers to the quality and operational usability of those functions, capturing the design-mediated failures that requirements alone cannot capture. Together, these contributions advance the operationalization of MHC.

Cable television networks distribute digital television by modulating each channel with quadrature amplitude modulation (QAM) and assigning it a fixed frequency slot in the cable's frequency spectrum. The digital upconverter is the subsystem responsible for interpolating the modulated baseband channels to the DAC's sampling rate and placing each channel at its assigned output frequency. In the existing implementation developed by WISI, channels are pre-grouped into fixed subbands prior to frequency shifting, which constrains operators to place each channel only within the frequency range of its assigned subband. The interpolation stage furthermore consumes a large share of the available DSP blocks on the target FPGA, making a per-channel replication of the current chain infeasible within the resource budget of the smaller FPGA platform planned for the next-generation product.

This thesis presents the design and evaluation of three digital upconverters that support arbitrary per-channel frequency placement while reducing FPGA resource utilization. All three architectures use a Farrow structure for arbitrary sample-rate conversion and pulse shaping. They differ in how interpolation to the final output sampling rate is performed: the first uses a cascaded integrator-comb interpolator, the second a shift-and-add FIR interpolator, and the third performs the entire conversion in the Farrow structure itself.

The Farrow coefficients were designed using a minimax framework. The resulting filter achieves a 78% reduction in DSP utilization compared to the Farrow filter in the existing implementation. All three architectures satisfy the spectral requirements across the full operating range of input symbol rates, and two of the three fit within the per-channel DSP budget on the target FPGA.

This paper explores how adolescent girls perceive skincare and skincare content on TikTok, and how such content influences their emotions and behavior. The study addresses the following research questions: 1.) How do adolescent girls perceive the growing prevalence of skincare practices among children? 2.) How do the girls believe that skincare practices on TikTok have affected their behavior and mindset? 3.) How do adolescent girls discuss skincare content shared on TikTok? Previous research highlights that particularly young adolescent girls are prone to peer pressure associated with social media and experience increased social comparisons because of the content they interact with. To investigate this issue in more depth, we conducted four group interviews in Sweden with a total of ten girls. Through thematic analysis we identified five key themes worthy of further exploration: Level of Interest and the Understanding & Use of Skincare Content, Suspicious Minds Persuasive Sources, Expectations and Unrealistic Beauty Standards, Lost Childhoods – Younger Children’s Consumption and Exposure to Skincare, and Adult Responsibility for Skincare Promotion and Distribution. Our findings suggest that the participants experience skincare content in a variety of ways. However, a common denominator was that the girls considered the content to negatively influence their self-esteem and judged it to have an active role in encouraging the desire to conform to homogenous and idealized beauty standards. The girls equally voiced that despite not actively or intentionally engaging with skincare content, they were nevertheless exposed to it and influenced by it, as well admitted to the trends portrayed on the platform having a subconscious effect on their emotions and habits. 

This thesis examines how the problem of female genital mutilation (FGM) is constructed within academic research on education and community- based FGM intervention strategies in Senegal, with a focus on how girls and girls’ agency are represented within these constructions. FGM is a harmful and oppressive practice affecting more than 230 million girls and women worldwide; infringing on their health, wellbeing and human rights. Researchers’ investigations into the issue demonstrate that there is currently a lack of adequate representations of women and girls’ agency in interventions, in addition to the prevalence of Eurocentric assumptions and perspectives surrounding FGM, which may create limitations towards eradicating the practice. Through a literature review, using Bacchi’s “What’s the Problem Represented to Be?” WPR method, this thesis examines the representation of girls’ agency within educational interventions and Eurocentric perspectives that underpin the prevalence of FGM. The first, second, and fourth questions of this approach are addressed, and the results are conclusive. The problem of FGM is found to be represented as a problem of tradition, collective social norms and structural inequalities such as generational power structures, with the underlying assumptions including socio-cultural norms and traditions, and what is not problematized is the lack of girls’ agency. From a WPR perspective, the silence surrounding girls’ agency is therefore shown to be an important, unproblematized factor within current research discourse on FGM interventions. 

Despite substantial economic growth in many western countries, income inequality has continued to rise, making the relationship between inequality and economic growth interesting to examine. This study investigates the impact of income inequality on economic growth through several literature-established transmission channels, each represented by several proxy variables. To address the research question, a two-step fixed effects approach is employed. In the first stage, the effect of income inequality on the transmission channels is examined, while the second stage analyzes the impact of these channels on economic growth. The analysis is based on an unbalanced panel covering 38 OECD countries over the period 1990-2023, thereby providing an updated contribution to the existing literature, which has a largely relied on older datasets. The empirical findings reveal that income inequality negatively affect economic growth through several proxies. More specifically, significant negative effects are found for fertility rates and the credit market, which is in line with the study's hypotheses. In addition, weaker and less robust evidence suggests that income inequality affects economic growth through patents and political unrest as well. 

This thesis evaluates whether Support Vector Regression (SVR) can improve financial curve construction by reducing functional instability without sacrificing market consistency. The study applies SVR as a post-processing smoother to benchmark-implied curves in both interest-rate and commodity markets. Three interest-rate curves, EUR-ESTR, USD-SOFR, and EURIBOR-6M-FWD, and one commodity forward curve, NYMEX-WTI, are examined. SVR models with polynomial, radial basis function, and sigmoid kernels are compared against traditional benchmark interpolation methods: Linear RT, Cubic Spline, and Hermite interpolation.

The empirical evaluation is based on two criteria: out-of-sample repricing accuracy, measured by mean squared error, and functional stability, measured by the total variance of daily curve changes. The results show that SVR does not provide a superior overall trade-off between market consistency and functional stability. The RBF kernel generally achieves the lowest repricing errors among the SVR models, but often at the cost of increased curve instability. Polynomial and sigmoid kernels can reduce total variance in several cases, but typically produce materially larger pricing errors. Overall, the benchmark interpolation methods remain more effective for the curves studied. The findings suggest that machine-learning methods for financial curve construction may need to incorporate financial pricing structure more directly in order to add practical value in a banking context.

As digital infrastructure grows in scale and complexity, reliable server monitoring becomes increasingly critical. This is particularly important within healthcare IT systems, where a disruption can prevent clinical staff from being able to do their job. Traditional methods for anomaly detection include statistical techniques or a manually defined threshold, where both struggle to capture complex temporal patterns. They also require separate manual configurations for each monitored metric, something that is not scalable in large and diverse server infrastructures.

This thesis investigates whether a single generalized unsupervised univariate machine learning model can detect anomalies across diverse server metrics in Sectra’s monitoring system without server-specific customization. To achieve this, a transformer autoencoder was developed and trained on data from 63 different hospitals, totalling more than 210 million data points. To find the best performing model, six configurations with varying hyperparameters were trained and evaluated. The best configuration was compared against the pretrained foundation model MOMENT in both zero-shot and fine-tuned settings, a static threshold classifier, and Claude Sonnet 4.6 as a large language model (LLM) baseline. In addition to the labeled evaluation, the best transformer autoencoder configuration and MOMENT variant were also tested in two practical scenarios: ranking metrics during known ongoing system problem periods and detecting behavioral changes following a system update.

The best performing transformer autoencoder configuration achieved a weighted VUS-PR of 0.625 and a weighted PA-F1 of 0.912. With that, it outperformed both MOMENT and the LLM, while also being approximately 19 times faster than MOMENT at inference. In the applied scenarios, both models surfaced known problematic metrics, consistently ranking them in top positions, and identified post-update behavioral changes that had gone unnoticed through manual monitoring. The univariate approach proved well suited for Sectra’s diverse infrastructure, as it allowed a single model to generalize across different hospitals without requiring server specific customization.

Future work includes expanding the labeled evaluation dataset, training a model on more data, further investigation of threshold strategies, and implementing relevance-based weighting of metrics to better prioritize alerts for support personnel.

Heating and cooling of buildings account for a significant share of overall energy use and therefore play a central role in the transition towards a more sustainable and resource-efficient society. At the same time, urban areas are continuing to expand and develop, increasing the demand for sustainable energy system solutions. District heating is highlighted as an important part of the EU’s strategy for reducing carbon emissions and improving energy efficiency. Despite the significant role of district heating in the energy transition, knowledge regarding the climate impact of the construction phase of district heating systems is limited. In particular, comparisons between different system configurations and associated equipment such as district heating substations are scarce. Therefore, this study aims to increase the understanding of the climate impact associated with the construction phase of district heating systems at neighborhood scale in Sweden through life cycle assessment (LCA). The study focuses on fourth-generation district heating (4GDH) as well as hybrid solutions combining centralized and decentralized system principles from 4GDH and fifth-generation district heating and cooling (5GDHC).

The study consists of three LCAs conducted in accordance with the SS-EN 15978:2026 standard using the software SimaPro. The LCAs identify the climate impact associated with the construction phase of district heating distribution pipes, district heating substations and cooling systems at neighborhood scale. A study object considered representative of neighborhood-level district heating systems in a Swedish context was used to derive dimensions and design requirements corresponding to Swedish average conditions. The LCA of district heating distribution pipes quantifies the climate impact of distribution pipes designed for two alternative system configurations, a 4GDH system and a hybrid solution. The LCAs of the remaining subsystems consider only the 4GDH system. Furthermore, the LCAs investigate how the climate impact of the construction phase varies under different assumptions regarding the use of recycled materials in manufacturing, the reuse of excavation materials, and changes in the technical service life of system components.

The results of the study show that the combined climate impact from the construction phase of the assessed subsystems in the study object is 1 366 tCO2-eq. The district heating pipes account for the largest share of the climate impact from the construction phase, followed by the cooling system and the district heating substations. The results further show that the processes related to the raw material extraction and manufacturing account for the largest share of the climate impact during the construction phase of district heating systems at neighborhood scale.

The quantification of the climate impact from the district heating substations is considered one of the key contributions of the study, given the identified knowledge gap. Climate impacts from district heating substations with different power levels were calculated and used to develop linear relationships that enable estimation of climate impacts for power levels beyond those included in the study. In addition, the climate impacts of district heating distribution pipes with different dimensions were assessed for the two system configurations. The hybrid distribution pipes had lower climate impacts than the pipes used in 4GDH systems, unless a high proportion of recycled steel was used in the production of the 4GDH pipes. The results therefore identify the use of plastic service pipes as a potential technical measure for reducing climate impact during the construction phase. Furthermore, increasing the share of recycled material content, reusing excavated materials during pipe installation, and extending the service life of system components were identified a important strategies for reducing climate impact.

Supporting students who experience difficulties with writing constitutes an important part of the special education teacher’s role. Writing is a fundamental and frequent activity in school, which makes knowledge of how to effectively support and motivate students with writing difficulties highly valuable. This study aims to contribute knowledge about motivation and writing from the perspective of teachers. It is based on seven semi-structured interviews, focusing on what teachers describe as significant for motivating students with writing difficulties, as well as how teachers design writing tasks to support such motivation. The study adopts a thematic analysis approach.

The findings indicate that teachers identify task content as motivating, particularly when writing assignments allow students to be creative. Furthermore, the results show that adaptations can promote students’ motivation to write by reducing barriers for those who experience difficulties. A third theme concerns the sense of competence, referring to how teachers describe students’ self-confidence as crucial for their motivation. The final theme is social interaction, where teachers highlight that students are motivated by opportunities to demonstrate their abilities. Additionally, the study shows how students with writing difficulties are influenced by how they are perceived in the classroom during writing activities.

The study highlights four key points. First, students with writing difficulties are motivated when their interests are acknowledged in writing instruction, as this enhances enjoyment in writing. Second, by removing obstacles to writing, teachers can foster motivation through a sense of participation. Third, enabling students to participate and experience competence contributes to increased motivation. Fourth, students with writing difficulties are motivated by the opportunity to present their work in ways that do not involve significant social risks.

The purpose of the study is to examine special education teachers’ perceptions of the phenomenon of inclusive pedagogical learning environment based on a phenomenographic qualitative approach. The study, carried out in a Swedish school context, also highlights which special didactic teaching strategies are considered important in an inclusive pedagogical learning environment. The data collection method used is semi-structured interviews with six special education teachers in four municipal upper secondary schools in the same municipality and in one national school agency. The theoretical framework is the concept of inclusion and the three special educational perspectives – the categorical, the critical and the dilemma perspectives. The analysis has resulted in four descriptive categories which present different ways of perceiving the phenomenon. The results show that an inclusive pedagogical learning environment is characterized by accessible teaching, consideration of students' perceptions of the teaching, students´ sense of belonging and cooperation between teachers and special education teachers. These categories can be considered when planning and implementing teaching. Successful special didactic teaching strategies are teachers´ ability to plan their teaching in ways which allow all students to learn through differentiated and explicit teaching. Teachers' relational competence and consideration of students' perceptions of the teaching environment are crucial for students' sense of belonging. Collaboration between special education teachers and teachers is an important prerequisite for the organization of inclusive learning environments. This result can serve as guidance in discussions of pedagogical learning environments in upper secondary schools. 

Keywords Inclusive education, pedagogical learning environment, special didactics, teaching strategies, special educator 

The High Level Architecture (HLA) standard for distributed simulation lacks a native Python API, despite the language's widespread adoption in scientific and engineering domains. While developing such an interface provides immediate engineering value, the underlying academic challenge lies in evaluating how Foreign Function Interfaces (FFIs) can be used for computation offloading. Despite the widespread use of FFIs to bridge performance gaps, there is still a lack of research into how different architectural granularities impact offloading performance.

This thesis explores FFI offloading strategies using the development of a Pythonic HLA API as the experimental domain. Three strategies were evaluated: No Offloading, Coarse-Grained Offloading, and Fine-Grained Offloading. The study was conducted in three distinct phases. First, an evaluation outside the HLA domain revealed that different FFI libraries exhibit distinct performance preferences for varying granularities. Second, an evaluation within the HLA domain identified the superiority regions for each offloading strategy based on the structural complexity of the data in the context of encoding HLA data types. The findings show that HLA encoding is a computationally inexpensive operation where the overhead costs of FFI offloading outweigh the performance gains of a faster guest language for most data structures. Finally, a prototype Python API was evaluated against the traditional HLA Java API in terms of performance and programming complexity. Results indicate that while performance is generally lower than its Java counterpart, the gap in update frequency narrows significantly as network load increases. Furthermore, quantitative metrics show a definitive decrease in programming complexity, increasing the usability and lowering the barrier to entry for distributed simulation development.

The think-aloud protocol is a well-established method in usability testing in which participants verbalize their thoughts, reasoning, and decisions while interacting with a system. Although the method provides rich insights into users’ cognitive processes and encountered usability issues, it traditionally relies on a human facilitator to guide sessions, encourage verbalization, and manage task progression.

This thesis presents the design, implementation, and evaluation of an AI-based evaluator intended to support think-aloud usability testing. A real-time pipeline was developed to capture microphone input, transcribe spoken utterances using the Whisper framework, and facilitate structured think-aloud sessions through standardized prompts delivered via Kyutai text-to-speech synthesis. The system was designed to partially automate session facilitation while preserving human responsibility for qualitative analysis.

The proposed framework was evaluated in a user study involving eight participants performing predefined tasks in Microsoft PowerPoint using the think-aloud protocol. All sessions were completed successfully without direct human moderation, demonstrating the technical feasibility of AI-based facilitation. Post-session questionnaires indicated consistently positive participant perceptions regarding clarity, responsiveness, and behavioral consistency.

The findings suggest that AI-supported facilitation can maintain procedural integrity and structured data collection in think-aloud usability testing. While the system does not perform automated usability analysis, it establishes a baseline framework for AI-assisted evaluation and provides a foundation for future research exploring more advanced or agentic extensions of usability testing workflows.

This thesis evaluates an existing Generalised LL (GLL) parser for the Storm programming language platform and compares it to the current Generalised LR (GLR) parser. The focus is on parsing performance, error recovery, and syntax highlighting when syntax errors are present.

A simple character-based recovery mechanism is implemented for the GLL parser. The evaluation uses generated Storm inputs, both with and without introduced syntax errors. Valid inputs are measured using strict parsing, while erroneous inputs are measured using approximate parsing. Syntax highlighting is evaluated by first checking which correct inputs produce syntax colour, and then comparing how much highlighting remains after errors are introduced.

The results show that the GLL parser performs especially well on valid expression-chain inputs, where it is up to 106 times faster than the GLR parser in the evaluated tests. For more deterministic inputs, the difference is smaller. With syntax errors, the results are more mixed: GLL remains faster for the simple expression grammar, but can be slower on some long erroneous inputs in the more realistic Storm language.

For inputs that produce syntax-highlighted output in the baseline test, the highlighting results are mixed. GLL preserves useful highlighting in several error cases, while GLR preserves more of the expected colours in some missing-closing-brace cases. Overall, the results indicate that the GLL parser is a promising alternative to the current GLR parser in Storm, but that the recovery strategy needs further improvement before full editor integration.

In this master thesis, the role of smartphone apps in Linköping, and people’s ability to move around there and about, by the help of the mobile apps, will be investigated in detail. The apps aim at helping people be more mobile while all transport alternatives are aimed at being accessible and distributed fairly. The transport options at focus are car, bus and train.    

The theoretical framework of the thesis is mobility justice and innovation resistance theory. Both help in explaining the background with mobility and how it connects to different aspects of urban planning and human behaviour. Also, explanations as to why people have the experiences and opinions that they do regarding innovation and digitalisation.  

This while the method used is qualitative semi-structured interviews with elderly people above the age of 60, and respondents who are within the numerous traffic organizations are the ones being interviewed. The respondents who are not elderly are from Linköping municipality, Dukaten parkering, Östgötatrafiken and Mälardalstrafik.  

Also, in the thesis it is analysed what the consequences with the apps are for the society of Linköping. Both positive and negative aspects are listed, and each section of the thesis take us closer to the thesis aim and answering all the research questions.  

One of the last sections of the thesis describes future outlooks in detail. This is meant to work as inspiration for upcoming research in this field and making sure that the field of urban planning continues to be relevant for the foreseeable future.   

Stiftelser skiljer sig från andra associationsformer eftersom de saknar ägare eller medlemmar som kan kontrollera att förvaltningen sker i enlighet med ändamålsbestämmelsen i stiftelseförordnandet. Det medför en risk att förvaltningen av stiftelsens tillgångar används i strid med ändamålsbestämmelsen.

För att säkerställa att förvaltningen överensstämmer med ändamålsbestämmelsen finns tre kontrollmekanismer: räkenskap och revision, tillsyn samt skadestånd. Räkenskaper och revision har en central och viktig funktion för att uppdaga avvikelser från ändamålet. Denna kontrollmekanism utgör också ett underlag för de två andra kontrollmekanismerna. Tillsynen grundar sina ingripanden på informationen som uppdagas i räkenskaperna och revisionen. Tillsynen har en ingripande och reaktiv funktion. Skadeståndet är både reaktivt och preventivt, den reaktiva funktionen bygger på avvikelser som uppdagas genom räkenskaperna och revisionen. Skadeståndet skapar ett incitament för stiftelsens företrädare, eftersom de ska följa stiftelseförordnandet och andra bestämmelser, vilket är den preventiva funktionen.

I uppsatsen konstateras att kontrollmekanismerna var för sig inte är tillräckliga, men i samspel med varandra säkerställer de att ändamålsbestämmelsen i stiftelseförordnandet uppfylls. Kontrollmekanismerna fungerar i samspel eftersom de kompletterar varandra, genom att fylla ut deras brister sinsemellan, vilket resulterar i ett enhetligt system. Därtill konstateras att stiftelserna innehar en stark självständighet, men att denna begränsas av statligt ingripande i viss mån. Det konstateras också att stiftarens vilja kan påverka de tre kontrollmekanismerna, men att stiftaren inte kan stadga föreskrifter som strider mot lagstiftarens uppsatta minimikrav. Avslutningsvis konstateras att räkenskaper och revision har en central funktion eftersom tillsynens ingripande och skadeståndets reaktiva utformning är beroende av informationen som fordras från räkenskaper och revision.

Denna uppsats utgör en litteraturstudie med syfte att analysera begreppet science capital (naturvetenskapligt kapital) och dess relevans för gymnasieelevers relation till naturvetenskap. Tre frågeställningar undersöks: begreppets teoretiska grund och centrala dimensioner, hur science capital relaterar till science identity, samt sambanden med elevers engagemang, aspiration och studieval.

Analysen bygger på 21 vetenskapliga källor och tillämpar reflexiv tematisk analys enligt Braun och Clarke (2022). Science capital har sin teoretiska förankring i Pierre Bourdieus kapitalteori och introducerades formellt av Archer med flera (2015) i samband med det longitudinella ASPIRES-projektet. Begreppet operationaliseras i tre dimensioner, vetenskapsrelaterat kulturellt kapital, vetenskapsrelaterade beteenden och praktiker, samt vetenskapsrelaterat socialt kapital, vilka inte enbart är mätverktyg utan formar, kännetecknar och operationaliserar begreppet på ontologisk, epistemologisk respektive metodologisk nivå. Analysens begrepp och relationer sammanfattas i en analysmodell där Bourdieus kapitalteori utgör det övergripande ramverket och elevers engagemang, aspirationer och studieval utgör de analytiska utfallen.           

Resultaten visar att science capital är ojämnt fördelat längs strukturella linjer av klass, kön och etnicitet, och att dessa mönster förstärks över tid under skolgången. Relationen mellan science capital och science identity är nära men inte deterministisk. Identitetsutvecklingen är beroende av socialt erkännande och tillhörighet samtidigt som akademisk prestation inte är tillräcklig för långsiktigt engagemang. Tillsammans erbjuder begreppen en mer nyanserad förståelse av elevers naturvetenskapliga val än traditionella intresse- och attitydbaserade modeller, men begreppets förklaringskraft varierar med nationell kontext. Analysen pekar på ett tydligt behov av ämnesspecifik, kontextberoende och longitudinell forskning, särskilt i en svensk utbildningskontext.

The increasing use of satellite constellations creates a need for dispenser systems capable of carrying and deploying multiple spacecraft efficiently. Conventional cylindrical dispensers may require additional interface structures when spacecraft mounting points are located far from the cylindrical surface. This work investigates whether a polygonal dispenser structure can provide a structurally feasible alternative by placing the load-carrying structure closer to the spacecraft interfaces.

The study has two main objectives: to identify the spacecraft sizes and configurations for which a polygonal dispenser is most suitable as an alternative to a cylindrical dispenser, and to determine how such a structure must be adapted relative to a cylindrical reference design in order to satisfy the structural requirements defined by the launch vehicle provider.

The study is divided into three separate parts: an initial study, a parametric study, and a concept study. The initial study examines the structural behaviour of cylindrical and polygonal thin-walled structures without attached spacecraft, focusing on stiffness, eigenfrequencies, and buckling. The parametric study evaluates the influence of geometric parameters, such as the number of sides, radius, corner radius, and launch vehicle adapter length, for different spacecraft configurations. Finally, a selected polygonal configuration is further investigated and assessed against structural requirements for stiffness, static strength, and buckling.

The results show that polygonal cross-sections can provide advantages in global bending stiffness, particularly for slender configurations with fewer sides. However, their performance is limited by the flexibility of the flat faces, especially for less slender configurations. Increasing the corner radius or introducing face curvature reduces the effective flat-face span, improves local stiffness, and reduces stress concentrations.

A four-sided dispenser with one tier of medium-sized spacecraft was selected for the concept study. Under the same modelling assumptions as the cylindrical reference, the final polygonal concept satisfied the considered stiffness, strength, and buckling requirements after introducing face curvature, local laminate reinforcement, and additional stiffening at the upper interface. Although the concept increased the mass relative to the cylindrical reference structure alone, it may reduce the need for additional interface components. The results indicate that polygonal dispensers could be advantageous for selected configurations, particularly when reduced system complexity is prioritised.

Swedish healthcare is based on the principle of equitable care, meaning that healthcare should be provided according to need and on equal terms for the entire population. Despite this, previous research has demonstrated significant variations in access to care, both between and within regions. Specialized palliative care is an essential and increasingly important part of healthcare, aimed at meeting complex needs and improving quality of life for patients at the end of life. This study therefore examines variations in access to and resource utilization within specialized palliative care between the central and eastern parts of Region Östergötland during the period 2021–2025. The study applies a quantitative, descriptive, and comparative approach based on secondary data from Statistics Sweden, Region Östergötland, and the Swedish Register of Palliative Care. The analysis includes demographic structure, mortality, specialized palliative care bed capacity, care utilization, healthcare costs, and access to specialized palliative care at the end of life. The results reveal clear and systematic differences between the two areas. The eastern part of the region has higher mortality and greater potential care needs, while simultaneously having lower access to specialized palliative care. The eastern area has lower specialized palliative care bed capacity, lower care utilization, and fewer patients receiving care. Cost patterns also indicate organizational differences between the two areas. The study demonstrates that these variations are related to healthcare organization and resource allocation. The findings indicate that the principle of equitable care is not realized within the region and highlight how decentralized governance may contribute to structural inequalities within the same healthcare system.

Sociala medier har förändrat marknadsföringens förutsättningar och bidragit till att influencer marketing blivit en central marknadsföringsstrategi. Genom influencers integreras kommersiella budskap i personligt innehåll, vilket kan göra det svårt för konsumenter, särskilt barn, att skilja reklam från annat innehåll. Eftersom barn utgör en särskilt sårbar konsumentgrupp aktualiseras frågor om hur väl de skyddas mot otillbörlig marknadsföring i digitala miljöer.

Syftet med uppsatsen är att undersöka det marknadsföringsrättsliga skyddet för barn vid influencer marketing enligt svensk rätt och relevanta EU-rättsliga bestämmelser. Särskilt fokus riktas mot influencers ansvar vid marknadsföring riktad till barn samt vilka förbättringsmöjligheter som kan identifieras inom svensk rätt.

Uppsatsen bygger på en rättsdogmatisk metod, kompletterad med en EU-rättslig metod och en komparativ metod. Analysen grundas på lagstiftning, förarbeten, rättspraxis, doktrin samt vägledningar från relevanta myndigheter och organisationer.

Resultatet visar att det finns ett rättsligt skydd för barn genom marknadsföringslagen och EU-rättsliga regler om otillbörliga affärsmetoder. Samtidigt kvarstår flera utmaningar. Influencer marketing kännetecknas av att kommersiella budskap ofta integreras i influencers personliga innehåll, vilket kan försvåra reklamidentifiering och öka risken för dold marknadsföring. Studien visar även att influencers har en central roll i marknadsföringen och att de kan bära ett självständigt ansvar för marknadsföring som strider mot gällande rätt.

Uppsatsen drar slutsatsen att den nuvarande regleringen inte fullt ut är anpassad till de utmaningar som digital marknadsföring medför. Ett starkare skydd för barn kan uppnås genom tydligare reglering, effektivare sanktioner, ökad tillsyn samt mer omfattande vägledning för influencers och företag som verkar på sociala medier.

Denna uppsats behandlar diskrimineringslagen (2008:567) (DL) med fokus på en situation

där en arbetsledare, som likställs med arbetsgivaren, oavsiktligt diskriminerar en äldre

underordnad arbetstagare. Diskriminering innebär att en person eller grupp misgynnas genom

att antingen behandlas sämre än någon annan i en jämförbar situation eller att misgynnas

genom ett kriterium som framstår som neutralt men som kan komma att missgynna personer

eller grupper. Uppsatsen behandlar även huruvida avsikt spelar någon roll vid bedömningen

om diskrimineringsfall förekommer, från arbetsgivarens sida, genom att göra en analogisk

tolkning av skadeståndslagen (1972:207) (SkL) och hur culpa påverkar potentiella

skadeståndsfall.

Vidare kommer uppsatsen behandla lag (1982:80) om anställningsskydd (LAS) och på djupet

skapa en förståelse för lagens innebörd och hur lagen syftar till att skydda arbetstagaren.

Lagen redogör för hur en arbetstagare kan bli skild från sin anställning och med vilka grunder

det är tillåtet. Det ena sättet är på arbetstagarens egna initiativ, genom att denne själv säger

upp sig. Det andra är från arbetsgivarens sida, genom antingen en uppsägning eller ett

avskedande.

Syftet med den här uppsatsen är att ge läsaren en mer tydlig bild gällande diskriminering och

uppsägning. Framförallt ska läsaren i slutet av analysen få en förståelse för hur

diskrimineringsfall kan kategoriseras. Detta är av yttersta vikt i vår uppsats, eftersom

skillnaden mellan direkt och indirekt diskriminering kan vara svår att förstå för personer med

begränsad kunskap om DL. Läsaren kommer även kunna ta del av huruvida oavsiktlig

diskriminering utgör sakligt skäl för uppsägning från en arbetsgivares sida.

This exploratory case study examines whether Swedish coastal municipalities visualization methods for flood management promote adaptive spatial planning. The study provides an insight into how digital twins can enhance the efficiency of flood analyses in line with national flood management strategies. The study uses a qualitative approach where the adaptive planning theory and diffusions of innovation theory work as a theoretical framework. The empirical data consists of four policy documents and five interviews for Helsingborg, Ängelholm, Gothenburg and Lindholmen Science Park/ Visual Arena.  The policy documents give an insight into how the methods relate to flood strategies and act as a complement to the interviews.

The research shows that Swedish coastal municipalities mostly use static visualization tools that usually not prepare the community for different scenarios. The digital twin has the potential to enable adaptive urban planning, but their use is currently limited. The municipalities struggle with challenges such as resource and competence shortage, data difficulties, and unclear guidelines from the government.

This master’s thesis investigates computational methods for vibroacoustic analysis of underwater structures, with the purpose of supporting the development of quieter underwater vehicles. The study addresses the connection between structural vibrations and acoustic radiation in water by evaluating how numerical simulations can be correlated with controlled physical experiments.

A simplified box geometry was designed and dimensioned to generate measurable vibration and acoustic responses within the constraints of the available test environment. Physical impact tests were conducted in both dry and wet conditions using accelerometers, a microphone, and a hydrophone to capture structural response and acoustic pressure. The experimental results were then compared with LS-DYNA simulations using three acoustic solvers: BEM, Rayleigh and Kirchhoff.

The results show that the transition from dry to wet conditions significantly affects the dynamic response of the structure, mainly through added mass and increased damping. In the dry condition, both BEM and Rayleigh showed good agreement with the experimental acoustic response, but Rayleigh at a lower computational cost. BEM was identified as the most reliable method for estimating acoustic pressure, where as the Rayleigh method proved to be an efficient alternative for simplified geometries. Kirchhoff was not reliable in the present implementation due to numerical instability.

For the wet condition, the correlation between simulations and experiments was more challenging. Reflections in the test tank, fluid damping, and limitations in the explicit solid mechanics solver treatment of fluid interaction contributed to discrepancies between numerical and experimental results. The study concludes that accurate vibroacoustic prediction requires careful modelling of both the structural and fluid domains. Further work is needed to improve fluid interaction coupling in explicits imulations and to validate the workflow on more complex underwater structures

This study examines how different groups of residents in Linköping perceive urban densification, both in the city as a whole and in their own local neighborhood, and which consequences are connected to their attitudes. The study is based on a survey with 258 respondents in the urban area of Linköping, combined with an interview with planners from Tyréns. The results show that attitudes toward densification vary clearly depending on whether it takes place in the city as a whole or in the respondent’s own neighborhood. Age and type of housing are the sociodemographic factors most strongly related to attitudes. The consequences seen as most important are a more crowded living environment, loss of neighborhood character, and reduced access to daylight. The study uses communicative planning as its theoretical framework and analyzes the results through the concepts of antigrowth, general acceptance, OIMBY and NIMBY. The largest attitude groups are antigrowth, followed by general acceptance, while NIMBY attitudes are much less common than expected and OIMBY is the smallest group. The study contributes to a greater understanding of how people view and experience densification and contrasts these views with how densification is presented in earlier research and by public authorities. 

The electronics industry is among the fastest growing contributors to global waste, and it is increasingly recognised as a key enabler for a circular economy to counteract the effects of resource depletion, e-waste and raw material dependency. There is an increasing policy ambition and academic debate on the circular economy in electronics but less is known about how Swedish electronics companies put these ambitions into practice. This thesis aims at filling this gap by examining the challenges and opportunities for implementing circular economy strategies in the Swedish electronics industry with an emphasis on reuse, repair, remanufacturing, and recycling.

It is a qualitative study following a total of 16 semi-structured interviews with actors along the Swedish electronics value chain, ranging from OEMs to EMS, recyclers, a science park and a municipal environmental authority. The sample is weighed towards industrial electronics, where the products are primarily business-to-business, long-lived, and subject to high reliability and trade-compliance requirements are high. Thematic analysis was used to analyse the interviews, following the approach of Braun and Clarke (2006) and drawing on the Gioia methodology, supported by NVivo software. The empirical analysis was framed by a conceptual lens developed through a narrative literature review.

The results reveal that the electronics industry in Sweden remains structurally linear, in that recycling acts as the default end-of-life route, while other value retention solutions, such as reuse and remanufacturing, are found in isolated examples. Barriers exist at several levels and are interconnected with regard to organisational, design and business model issues such as the lack of take-back systems, dependence on suppliers, design choices that limit recyclability, and uncertainties in the regulatory framework. Enabling factors include shared sustainability goals, regulatory drivers, second-life value recovery, and emerging traceability tools such as the Digital Product Passport. The study also identifies less-discussed dynamics like the dependency of EMS providers on OEMs for design, the tension between trade compliance and circular design and the conservative attitude of B2B customers in safety-critical sectors. The thesis adds empirical knowledge on implementation of the circular economy in industrial electronics and provides concrete leverage points for the ReLoop project and the industrial electronics industry in general.

This study examines how total, core, and energy inflation affect income distribution in advanced OECD economies. Using panel data from 29 OECD member states over the period 1995 to 2022, fixed effects models are estimated with the top 10 percent and bottom 20 percent pre-tax income shares as dependent variables.

Results show that core inflation significantly reduces the bottom 20 percent income share, consistent with the wage lag mechanism whereby nominal wages fail to keep pace with rising prices. For the top 10 percent, inflation loses its statistical significance once asset prices are controlled for, suggesting that it is rising financial and housing values rather than consumer price inflation that drives the income share of the highest earners. Energy inflation shows no significant distributional effect for either group in the fully specified model. The findings indicate that the distributional consequences of inflation differ substantially by inflation type and income group, with low income earners bearing a disproportionate share of the burden from domestic price pressures. 

Denna rapport beskriver ett kandidatarbete som genomfördes i samband med utbildningarna civilingenjör i datateknik och civilingenjör i mjukvaruteknik vid Linköpings Universitet. Kandidatarbetet bestod av ett mjukvaruutvecklingsprojekt med uppdraget att vidareutveckla den digitala vågformsvisaren Surfer för att expandera användarnas möjligheter att skapa annoteringar i systemet och att samarbeta med andra på samma vågformsfil. Rapporten beskriver den tekniska implementeringen av denna vidareutveckling och förklarar de tekniska val som gjorts i projektet. Vidare reflekterar rapporten kring den arbetsprocess som arbetsgruppen tillämpat. Syftet med rapporten är att dokumentera hur arbetsgruppen har arbetat med projektet, hur projektet kan skapa värde för kunden, vilka erfarenheter som kan dokumenteras från projektet och som kan vara intressanta för framtida projekt, vilket stöd som kan fås genom att skapa och följa upp en systemanatomi, och hur ett delat systemtillstånd kan implementeras utan att behöva skicka all data varje gång det uppdateras. Arbetsgruppen använde sig av det agila ramverket Scrum och utvärderade processen med regelbundna mätningar. Projektet skapade värde för kunden genom att implementera funktionalitet som användarna har efterfrågat och genom användartester skapat ett system som enligt SUS-ramverket anses vara ett användarvänligt system. Flera erfarenheter dokumenterades, bland annat att arbetsprocesser som Scrum-ramverket kan anpassas efter gruppens behov för att bäst passa det gruppen behöver. Systemanatomin visade sig vara en värdefull tillgång för projektgruppen som bidrog till skapandet av milstolpar och delmål, samt som stöd under gruppens sprintmöten. Genom att implementera förändringsbaserad synkronisering där ändringarna sker genom meddelanden kan systemet skicka endast den data som förändrats mellan klienter och server och därmed kringgå problemet att skicka om hela systemtillståndet vid varje ändring av systemet. 

Software testing has always been an important part of software development. As long as software testing has existed, there have also been attempts to automate it. Nowadays, with the rise of Large Language Models (LLMs) in our daily lives, many attempts have been made to use them for test generation, especially for unit tests.

Despite the growing demand, there is still no benchmark specialized for industrial use that evaluates unit test generation in Java. This work aims to address this gap by creating an automated benchmark for unit test generation, which uses dynamic prompting techniques and five metrics to evaluate the performance of each LLM. It covers three difficulty levels (Simple, Complex, and Industry), which makes it applicable in an industry setting.

This benchmark was tested on four LLMs (DeepSeek-R1, Mistral-Nemo, Llama3, and Qwen2.5). The results show the effectiveness of dynamic prompting compared to static prompt templates. We also observe that, while models show promising results in the Simple and Complex categories, the results in the Industry category still have significant room for improvement.

Furthermore, the results demonstrate that the size of a model is not always correlated to its performance in this task, which highlights the importance of benchmarking.

Den pågående gröna omställningen kräver en omfattande elektrifiering av samhället, vil-ket ökar belastningen på dagens elnät då de ursprungligen inte är byggda för att stödja denutökade elkonsumtionen från exempelvis elbilar. För att möjliggöra utvärdering av smartladdning krävs tillförlitliga simuleringsverktyg. Inom ramen för detta kandidatprojekt harsimuleringsverktyget PowerFlow vidareutvecklats för att möta detta behov.Vidareutvecklingen har genomförts med en iterativ arbetsprocess baserad på Scrum,där kontinuerlig erfarenhetsinsamling och anonyma enkäter använts för att utvärdera ochoptimera gruppens arbetssätt. En systemanatomi upprättades också tidigt som ett verktygför att hantera systemets arkitektur och samordna parallella utvecklingsflöden.Verktygets tidigare prestandabegränsningar vid optimering har lösts genom att intro-ducera analytiska gradienter via algoritmisk differentiering. I praktiken innebar detta en29 gånger snabbare optimering. En ytterligare funktionalitet var att kunna jämföra simu-lerade värden med de dokumenterade värdena för att identifiera och uppdatera felaktigaparametrar. Två metoder utvecklades för detta ändamål, en baserad på majoritetsröstningoch en baserad på linjär regression. Vid utvärdering valdes regressionsmetoden eftersomden visade sig vara mest tillförlitlig. Vid testning av algoritmen visade sig skillnaden mel-lan de beräknade kabelparametrarna och de faktiska värdena vara mindre än 1%. Utöverdetta har PowerFlow byggts ut för direkt avläsning av data från CSV-filer för att omgåendekunna ta fram bland annat spänning.Projektets samlade erfarenheter visar att arbete mot en kund kräver tidig kravdefinie-ring och att parallell kunskapsinhämtning effektiviseras av interna workshops. Dessutombelyser resultaten vikten av konsekvent koddokumentation vid vidareutveckling av en be-fintlig kodbas, såsom PowerFlow.

Nocturnal short-wave infrared (SWIR) sensing is relevant for passive low-light observation and imaging applications, but the available SWIR illumination under moonless nighttime conditions can vary due to atmospheric effects and natural night-sky emissions. This variability makes short-term forecasting of nocturnal SWIR irradiance a challenging problem.

To investigate whether this irradiance can be forecasted using data-driven approaches, this thesis evaluates machine learning models for one-hour-ahead prediction of moonless nocturnal SWIR irradiance using measured SWIR irradiance together with selected meteorological variables. Ridge Regression, XGBoost, Random Forest, and LSTM models are evaluated and compared against a persistence baseline.

The results show that forecasting nocturnal SWIR irradiance was difficult with the available dataset and covariates. While the LSTM model achieved the best overall performance, the improvements over the persistence baseline were limited, and additional meteorological and temporal covariates did not consistently improve prediction accuracy. Overall, the study highlights practical limitations in forecasting moonless nocturnal SWIR irradiance using the available measurements and covariates.

Detta dokument är en rapport som handlar om projektet ”TAK Simulator” som utfördes som en del av kursen TDDD96 vid Linköpings universitet. Projektet genomfördes på uppdrag av en kund. Projektets mål var att skapa ett program som kunde emulera flera TAK-klienter samt skapa och exekvera scenarier med dessa emulerade klienter. De emulerade klienterna skulle efterlikna de centrala funktionerna hos riktiga klienter. Rapporten syftar till att dokumentera projektet. Programmet implementerades i Python, och arbetet organiserades enligt den agila arbetsmetoden Scrum. Resultatet var ett program som uppfyllde kundens krav. Programmet kan skapa flera emulerade klienter som kan kommunicera enligt TAKs kommunikationsprotokoll. Programmet kan även skapa scenarier utifrån filer och därefter exekvera dessa. Lösningen skapade värde för kunden genom programmets funktionalitet, kodkvalitet och möjligheter till vidareutveckling.

This Bachelor’s report aims to describe the work process and the results of the Bachelor’s project that has been produced by the group PUM1 within the framework of the course TDDD96 Bachelor’s Project in Software Development given at Linköping University. The report presents four central questions: how the working method contributed to customer value, what experiences could be documented, what support is provided by creating and following up a system anatomy and how the user-friendliness of the user portal differs from the customer’s previous system. The group’s agile working method could be used to create customer value and a number of process- and technology-related experiences could be documented. The system anatomy provided support in the form of a mutual understanding of the user portal, but less support during its implementation. Finally, the result was that the user portal was more user-friendly than the customer’s previous system.

This thesis explores what previous research says about the use of horror fiction in different educational contexts to find out how it can function in the English upper-secondary classroom in Swedish schools. It focuses on three areas: teaching methods, ethical concerns, and how horror affects students’ literary learning. The study finds that horror fiction can support transformative learning, increase motivation, and help students develop empathy and genre awareness. However, most of the available research is theoretical, and there are not any studies that look specifically at the Swedish school context. Even so, the findings suggest that horror fiction can be a valuable and effective resource in English teaching when used thoughtfully. The recommendation for teachers is to use horror fiction with caution and that more research is needed before its use can be fully justified in classroom contexts

Without access to real-time profitability data, traffic planners are forced to make decisions based solely on experience, which risks leading to poor planning and lost revenue. This report describes the development of Project Visir, a web-based decision support system that enables traffic planners to assess the profitability of planned transports in real time. The system retrieves planning data via an external API and calculates estimated revenue using a five-step hierarchical model, where historical data is matched with successively lower precision until sufficient data is found. The work was carried out in close collaboration with the customer, with Kanban-based task management and continuous testing. The delivered system allows traffic planners to filter truck configurations by areas and date, view estimated profitability directly in the interface and test alternative loading combinations. The project demonstrates that clear structure and close early collaboration with the customer are essential for a well-grounded result and that the hierarchical model adds value beyond the estimate itself by indicating how reliable the figure is, which is particularly valuable for planners with limited experience.

This report describes the design and development of a web-based e-commerce platform for the company Garn Extrem in Linköping. The purpose of the project is to create a functional Minimum Viable Product (MVP) that enables the digital sale of the company's products, while also providing practical experience from a large-scale software development project. The system was developed as a full-stack application based on a client-server architecture, in which the frontend and backend communicate through a REST API. The platform provides core features such as product browsing, filtering, searching, a shopping cart, and payment integrations, as well as a modern administrative interface for managing products and orders. Particular emphasis was placed on security, performance, and usability. The project was carried out using an iterative development methodology, with continuous data collection through retrospectives, version control, and testing. The results show that the system's core functionality fulfills the defined requirements for an MVP, with strong performance and a stable technical foundation. Furthermore, the report analyzes the project's development process and technical decisions. Key lessons learned include the importance of early risk management, clearly defined API contracts, and continuous code reviews. The report therefore contributes both a practical solution for the client and general insights that are relevant to future software development projects.

This thesis examines whether systematic risk, measured as CAPM beta, varies between bull and bear market regimes on the Swedish stock market. Using monthly data for 142 stocks listed on Nasdaq Stockholm between February 1998 and December 2025, grouped into eight aggregated ICB sectors, a dual-beta-CAPM model is estimated in which beta is allowed to differ across regimes. Market regimes are identified primarily through a two-state Markov model and complemented by a 12-month moving average and a consumer confidence index in order to assess the sensitivity of the results to the choice of regime definition. The main inference is based on a sector-level panel Wald test with stock fixed effects. In the main sample, the hypothesis of symmetric beta is rejected under all three regime definitions. However, the robustness checks show that the strength of the results varies. The Markov result remains after including delisted firms, the CCI result weakens but remains significant, whereas the moving-average result is not robust in the extended sample. The sectoral patterns are heterogeneous. The technology sector displays the clearest procyclical pattern under the Markov definition, while several other sector estimates are sensitive to sample composition and regime definition. Overall, the findings support the presence of regime-dependent beta in the studied stock population, but also show that conclusions about beta asymmetry are method- and sample-sensitive. The results suggest that an unconditional CAPM specification with a constant beta does not fully capture systematic risk on the Swedish stock market. 

Over the past decade, digital commerce has grown substantially, contributing to the emergence of new payment solutions such as Buy Now, Pay Later (BNPL). Sweden has become one of the most prominent markets for these services, while their impact on consumers' financial behavior has attracted increasing attention. In particular, concerns have been raised regarding whether deferred payment solutions may encourage impulsive consumption and whether certain socioeconomic groups are more vulnerable than others.   The purpose of this study is to examine the relationship between the use of BNPL services and consumer behavior, as well as whether this relationship varies across individuals with different income levels. The study is based on a quantitative survey comprising 182 respondents, of whom 142 individuals with experience using BNPL were included in the regression analysis. The analysis was conducted using an ordinary least squares (OLS) regression model, supported by behavioral economic theories such as present bias, mental accounting, and bounded rationality.   The results show a statistically significant positive relationship between higher BNPL usage and a higher consumption index, indicating more impulsive consumption behavior. However, no statistically significant evidence was found to suggest that this relationship varies across income groups. Age was the only control variable found to have a significant relationship with consumer behavior, with older age being associated with less impulsive consumption.   This study contributes to the understanding of how modern payment solutions are associated with consumer behavior and suggests that behavioral mechanisms may be more influential than income levels in this context. However, the findings should be interpreted with caution, as the study is based on cross-sectional data and a convenience sample, which limits the ability to draw causal conclusions and generalize the results.    

Keywords: Behavioral Economics, Buy Now, Pay Later (BNPL), Consumer Behavior, Impulsivity, Income groups, Mental Accounting, Present bias, Self-Control 

This thesis examines how J.R.R. Tolkien’s On Fairy-stories functions as a foundational poetics of Lord of the Rings trilogy and The Silmarillion and shapes the narrative’s mythopoetic structure, symbolism and moral vision. Drawing on Jacques Derrida’s notion of bricolage within deconstruction theory, the study argues that Tolkien’s theories of sub-creation, Secondary Belief and eucatastrophe are not merely theoretical ideas but active narrative principles realised within the text itself. Through close textual analysis of key elements such as Gandalf, Lothlórien and the One Ring, the thesis demonstrates how Tolkien transforms inherited Norse, Celtic and classical mythological traditions into a coherent Secondary World with its own ethical and symbolic framework. In doing so, Tolkien embodies the role of bricoleur, working on pre-existing mythic linguistic and cultural materials and recombining them into a new literary structure that simultaneously preserves traces of its origins while generating new meanings. Rather than reproducing mythic materials directly, Tolkien reinterprets and reshapes it to explore the themes of power, loss, hope, morality and transcendence. The study concludes that The Lord of the Rings trilogy represents the practical realisation of Tolkien’s fairy-story theory and that fantasy in Tolkien’s conception, functions as a serious literary mode capable of engaging with fundamental human concerns.

In this paper, we derive the distribution of a two-dimensional (complex) random walk in which the angle of each step is restricted to a subset of the circle. This setting appears in various domains, such as in over-the-air computation in signal processing. In particular, we derive the exact joint and marginal distributions for two steps, numerical solutions for a general number of steps, and approximations for a large number of steps. Furthermore, we provide an exact characterization of the support for an arbitrary number of steps. The results in this work provide a reference for future work involving such problems.

This paper investigates how a transformation of datacan be structured and applied to an IFC-file. The datashould be transformed into a format that integrates wellwith a relational database. The data should also onlycontain information relevant for quantity take-off as itwas meant for integration into the calculation phase ofa project in C7 Projects. This was achieved by parsingan IFC-file using xBim and aggregating the lines basedon a combination of object name, floor and building. To-tal amounts and quantity units were calculated for eachrelevant IFC-object and included in the aggregate. Thetransformed data was then compared to a second ag-gregate done using basic text search to confirm correctquantities. Due to lack of access to professionally donequantity take-offs and an IFC-file belonging to the sameproject this comparison could not be made to verify datacorrectness or fitness of use. Irregardless of lack of accessto previously mentioned information the prototype ma-nages to flatten the structure of the hierarchical file intoa format compatible with relational databases withoutlosing the spatial structure.

Function calling enables large language models to interact with external toolsby emitting structured calls that an application can execute. Asthe number of available tools grows, reliably selecting the correct tool andarguments becomes harder. A controlled benchmark is presented that varies thesize of the available tool set while holding task structure constant, enablingdirect comparison of function-calling performance at scale. A set ofproprietary and open-source models is evaluated on tasks that require a singletool call, and scored on tool-selection accuracy, call format validity,argument correctness, and strict success rate. The benchmarkprovides a systematic view of how tool set size affects function-callingreliability and offers practical guidance for building agentic systems.

Real-time multimodal AI interaction on mobile devices is constrained not by model capability, but by the data-processing bottleneck at inference time. This thesis investigates how a modular, edge-assisted pipeline can reduce Time To First Token (TTFT) in conversational, multimodal large language model applications on Android, addressing the core question: how do data reduction strategies and the choice between local and remote model execution affect latency? To explore this, the Android application TripleEM was developed using Flutter, and benchmarked across 12 configurations combining three image resolutions (1080p, 480p, 224p), two Voice-to-Text (VTT) execution modes (on-device vs. remote), and Region of Interest (ROI) detection toggled on or off. Tests were conducted on 100 images from the Microsoft COCO dataset, with TTFT as the primary metric and classification accuracy evaluated using an LLM-as-a-judge approach. Results show that the dominant driver of latency is VTT engine placement. On-device transcription via Whisper Tiny averaged 2019 ms, slower than the total TTFT of any remote VTT configuration, making it the single largest bottleneck. Image downscaling meaningfully reduced upload time and backend inference cost with no measurable accuracy degradation, even at 224p. ROI detection added minimal processing overhead (under 50 ms) but did not consistently reduce upload time, and in some cases harmed accuracy by discarding relevant content. The fastest configuration achieved a mean TTFT of 408 ms, meeting the 500 ms real-time threshold, while the slowest exceeded 3800 ms. An unexpected finding was a modality bias in the remote multimodal model: configurations offloading VTT to the backend produced lower visual classification accuracy, suggesting the model may over-attend to audio input during joint inference. The study concludes that downscaling combined with remote VTT offers the best latency, while on-device VTT is preferable where accuracy is critical.

The Swedish electrical production mix is changing from a system dominated by large synchronous generators to a system with a higher share of renewable energy that utilizes power electronic inverters. This transition complicates reactive power balance management, which is critical for maintaining grid voltage stability and reducing losses. Reactive power is managed regionally by the so-called distribution system operator (DSO) by, among other methods, imposing consumption limits on large reactive power users such as electricity intensive large industries.

Industrial plants commonly operate induction machines, which are inductive loads and thus by definition consumers of reactive power. To compensate for these, many industries utilize capacitor banks which provide a capacitive load, which counteracts the induction machines' reactive power consumption. These banks, however, provide only discrete steps in reactive power management which may sometimes over or under-compensate. In the studied plant they are also manually controlled, leading to a limitation in response time. An alternative solution is the use of synchronous machines, which can be magnetized to both produce and consume reactive power independent of their active power output. Furthermore, an opportunity exists to operate a synchronous machine in idle solely for reactive power support, as a so called synchronous condenser.

The objective of this thesis is to investigate how the total reactive power at a Swedish pulp and paper industry can be controlled to stay within the DSO-imp-osed limits. As these threshold over-steps are billed based on 60-minute averages, the control also needs to be fast enough. This leads to the two research questions. The first question relates to which methods can be used to control reactive power control and the potential these methods have, while the second relates to how automatic control strategies can be implemented to improve the response time of the controllers.

The work consists of two main parts, supported by a comprehensive theory chapter. The first part analyzes reactive power measurements from the studied plant and how this data compares to the DSO limits. These measurements are also integral for the second part where simulations are conducted to show how the control of existing capacitor banks and the potential of synchronous machine excitation can reduce the level down to the limits. Furthermore, automatic control strategies are studied to analyze how fast and coordinated control can be achieved.

The findings indicate that automatic, dynamic synchronous machine reactive power control provides a fast and continuous method which improves flexibility through both added capability and a fast response time. This is also the case for the synchronous condenser. The results also show that capacitor banks serve an important role in compensating high steady-state reactive power demand and therefore act as an important complement to synchronous machine excitation.

It is a well known fact that physical activity and sports provide significant health benefits. While many different assistive tools have been developed for visually impaired individuals throughout the years, they still often lack the assistive technology to participate in sports independently. In this thesis, the gap is addressed through the development of a prototypical platform that utilizes computer vision and machine learning to create directional perception for the sonification of objects in real-time. The system, built up of a Raspberry Pi 5 and an AI Camera, employs the pinhole camera model to translate 2D pixel coordinates of an object into a 3D target vector consisting of an azimuth, elevation and distance. These vectors are then rendered into spatial audio with Head-Related Impulse Responses (HRIR), and a sound is played inside a pair of headphones that corresponds to the direction of said object, along with a frequency attenuation dependent on the distance to it.

A comparative analysis was conducted between two object detection algorithms in this context: MobileNet SSD V2 320x320 and Color Segmentation (ColSeg). The results demonstrated that MobileNet SSD maintained exceptional hardware stability, averaging a 20% CPU load with zero dropped targets over a one-hour stress test. This was possible due to the AI Cameras internal IMX500 sensor handling the image processing. However, MobileNet's object detection provided inconsistent bounding-box dimensions around human targets under movement, which led to unreliable distance estimates.

Conversely, while ColSeg provided stable distance metrics for rigid objects like balloons, its processing demands over a one-hour stress test caused the Raspberry Pi to reach its 85°C thermal throttling limit. This resulted in a cascading hardware failure: HRIR calculation latency spiked over 30 ms, causing buffer backlogs, target drops, and severe audio stuttering.

The study concludes that while the developed platform establishes a robust foundation for a non-invasive directional perception tool, relying on 2D bounding boxes for depth estimation is methodologically insufficient for dynamic sports environments. To ensure system survivability and user safety in future iterations of the system, the inclusion of hardware depth sensors, such as LiDAR or stereo cameras, and the use of pre-computed HRIR filters are recommended.

Kandidatarbetet behandlar utvecklingen av ett verktyg för visualisering av VHDL-kod i mjuk realtid genom blockdiagram. Projektet gjordes på beställning av Institutionen för systemteknik på Linköpings Universitet. Bakgrunden till projektet är att hård-varubeskrivande språk som VHDL oftast är svåra att visualisera, eftersom språket beskriver parallella processer utan en tydlig visuell representation. Syftet med rapporten är att beskriva hur projektgruppen har arbetat under kandidatprojektet och hur den slutgiltiga produkten har skapat värde för kunden. Utöver arbetet dokumenterades arbetsprocesserna och deras utveckling under projektets gång. Projektet resulterade i ett användarvänligt system för realtidsgenerering av diagram med funktioner för bland annat navigering mellan diagram och kod, markering av kodavsnitt i diagrammet samt inställningar för exempelvis diagramdjup, tema och visning av mellansignaler. Rapporten kom även fram till att tidig utveckling av en systemanatomi ger en överblick över systemet och stödjer projektets arkitektoniska beslut. Slutligen diskuteras hållbarhetsaspekter på arbetet samt förslag på vidare utveckling av projektet.

In large technology companies, maintaining and verifying compliance across interconnected systems is a significant challenge. This project investigates how the OSCAL framework can be used to track, visualise, and structure requirements management within compliance.

The purpose is to assess whether OSCAL is a suitable framework for identifying deficiencies in a system landscape and whether it can serve as an initial step toward automated compliance analyses.

To evaluate the framework, the prototype PRISM was developed, using OSCAL to model the user’s system. The prototype was iteratively developed in collaboration with compliance experts to validate functionality and continuously adjust the development direction.

The results show that OSCAL can be applied in a real scenario to identify deficiencies and visualise compliance. PRISM demonstrates that an OSCAL-based solution enables structured visualisation and traceability of compliance.

The conclusion is that OSCAL is a useful framework for software development within compliance. PRISM indicates that an OSCAL-based tool can support organisations in tracking security requirements and enable earlier detection of compliance deficiencies.

The purpose of this thesis is to determine whether there is a group action of A₅ on a Riemann surface of genus six. We have calculated possible candidates of orbifolds with signature (g; m₁, m₂, m₃, m₄) and considered if we are able to define epimorphisms to A₅ from the related Fuchsian groups. Lastly we examined if A₅ was acting as a subgroup of a bigger automorphism group.

We managed to show that there was a Fuchsian group with signature (0; 2, 2, 2, 3) with an epimorphism to A₅ and that A₅ could act as subgroup of S₅. 

This thesis examines how children involved in gang-related criminality are represented in Swedish newspaper articles. The thesis is based on a critical discourse analysis of 18 newspaper articles published in Expressen, Aftonbladet, Dagens Nyheter and Svenska Dagbladet between October 2025 and March 2026. Drawing on theories of childhood, particularly constructions of the innocent and dangerous child, as well as research on media representation, the study uses Critical Discourse Analysis to examine how children involved in gang-related criminality are represented in Swedish news media.

The thesis examines how linguistic choices, emotional framing and institutional perspectives construct meanings related to children, vulnerability, criminality and responsibility in Swedish news media.The analysis identified three dominant and recurring representations in the material: the child positioned as a victim, the child constructed as being at the center of criminality, and gendered representations of children in relation to gang-related criminality. Together, these representations produced an ambivalent understanding of the child. Children were represented in the examined material as innocent subjects exposed to manipulation, fear, exclusion and structural inequality, while at the same time being portrayed as active participants capable of serious violence and criminal behaviors.

The findings therefore demonstrate that children were not constructed in one coherent or stable way. Instead, they were represented through contradictory and overlapping meanings in which innocence, agency, dependency and dangerousness coexisted at the same time. This also illustrates that news media is not entirely neutral but uses different tones and emphases when representing social phenomena such as gang-related criminality, thereby shaping how such issues are interpreted and understood. The analysis highlights how newspaper representations both reflects and reproduces competing understandings of children within contemporary discussions surrounding gang criminality in Sweden.

This paper investigates parents’ perspectives on their children’s sport commitments and the reasons and ways of supporting this involvement. The study is guided by four research questions: 1) What reasoning do parents provide for allowing and supporting their child’s sport commitment? 2) What benefits and potential challenges do parents observe in relation to their child’s sport commitment, and how do they perceive and reflect on these effects? 3) How do parents describe the potential sacrifices their family may face in relation to their child’s sport commitment, what these potential sacrifices consist of, and how they navigate them? 4) How do parents reflect on the various actors whose potential demands may contradict or interfere with their child’s sport commitment, what these demands consist of, and how the family navigates them? Previous research highlights the importance of parental support and shows that a child’s own motivation strongly influences continued participation. Studies also point to challenges such as time management and financial demands. This qualitative study is based on six semi-structured interviews with parents of girls aged 12–15 who played football or ice hockey. Three themes were identified: “See the Child,” “I am a Parent,” and “The Importance of Social Relations.” The findings show that parents primarily supported their children’s sport participation because the children themselves wanted to be involved. Parents also observed several benefits, including social, physical, and mental development. Challenges included financial costs and the pressure children might experience. Most parents reported making personal sacrifices, such as reduced free time, but emphasized that these sacrifices were accepted as part of being a parent and motivated by love. 

This thesis investigates algorithms for sonifying objects into spatial audio. The primary focus has been to examine how Head-Related Transfer Functions (HRTFs) could be interpolated while maintaining accurate spatial localization cues. More specifically, how different interpolation methods compare when interpolating spatial localization cues. The findings from the experiments showed that using a neural network to predict the spectrum of the HRTFs provide high spectral accuracy and is superior in capturing the localization cues. Furthermore, because human hearing cues are highly individual and depend on anatomical features such as ear and head shape, this thesis explores how a personalized neural network can improve HRTF interpolation. Utilizing the HUTUBS database, a model was trained to predict the individual spectrum of the HRTF. The results of the thesis showed that the model improved the shape of the magnitude interpolation. However, th etiming differences between filters were inaccurate. This thesis also investigates how the input layer of the anatomical features could be optimized to both increase prediction accuracy, but also to ease measurements for data collection. Overall, the findings demonstrate the potential of the proposed personalized model. However, further work is required to improve phase and timing prediction, as well as to expand the training dataset.