Implementing a programming language is a significant undertaking. This is especially problematic for domain-specific languages, whose limited general applicability skews the effort-to-benefit ratio further. On the other hand, a domain-specific language can be very beneficial indeed for a problem in its intended domain. To realize this potential we require more effective implementation approaches. Of course, the technical requirements of a domain-specific language overlap significantly with those of a general purpose language, thus advances for one tend to benefit the other as well.

Programming in general tends to increase productivity through abstraction, various means of either relegating implementation details to more limited portions of a program or omitting them altogether. This dissertation explores abstraction in three areas: reusing work between language implementations, ambiguity in programming language syntax, and automatic representation selection for high-level data structures.

In the first area we provide syncons, short for syntactical construct, where each syncon is a user-defined language construct that includes syntax, binding semantics, and run-time semantics. The run-time semantics are given in terms of macro expansion to another language, which in turn is also defined using syncons. A language is constructed by defining whatever new syncons are required and reusing others from previous languages.

In the second area we formally define resolvable ambiguity, an ambiguity that can be resolved by a programmer by editing an ambiguous program. The class of resolvable languages is larger than the class of unambiguous languages, which gives greater design freedom. For example, a designer need not make arbitrary choices in their grammar to satisfy the needs of most modern parsing methods. This thesis focuses on three cases for resolvable ambiguity: syntactic composition, intuitive grammars, and new language constructs for which no syntactic conventions yet exist. We provide two approaches using resolvable ambiguity; one extending context-free grammars and using its own parser, and one meant for embedding in a conventional parser. The former is more expressive, while the latter admits a proof that all ambiguities are resolvable, which we provide, mechanized in Coq. Both can automatically compute all resolutions of an ambiguity.

In the third area we present an approach where new abstract types, representations, operations, and operation implementations can be introduced by a user in an extensible fashion, e.g., new libraries can add representations, operations, and operation implementations to previously defined abstract types. We also support representation switching, where an abstract type can be represented differently throughout lifetime. We complement the design with a set of solvers prioritizing optimality or quick analysis, and also support transferring a previous solution, even in the presence of minor program changes.

Overall, these contributions allow greater control over the level of abstraction in programming languages as well as their implementations, raising it where beneficial, and enforcing greater specificity where desirable.

Autonomous aerial vehicles have been increasingly used and adapted to cover tasks in multiple areas of our daily lives. This growth is mainly due to the rise in computational power in small form factor processing units, making these more efficient and enabling the deployment of novel algorithms for advanced applications. For the same reasons, the deployment of spacecraft for low-orbit operations has experienced an exponential increase, associated with the lower costs of access to orbit and the need for space-bound security and telecommunication. As these systems become ubiquitous, the need to ensure their safe operation in shared environments is ever more significant. In this thesis, we focus on four core problems of multi-agent autonomous systems: robust control, decentralized control for collaborative transportation, formation control, and reliable validation.

To tackle the first problem, we propose a robust control framework for trajectory tracking. Planners onboard the vehicle generate trajectories focusing on a high-level task. Our solution focuses on safe trajectory tracking under disturbances and provides upper bounds on the error with respect to the trajectory. Given a worst-case tracking error, it is possible to plan collision-free trajectories, ensuring system safety.

For the second problem, we propose controllers for collaborative load transportation using aerial vehicles and spacecraft. We develop centralized and decentralized model predictive controllers and compare their performance regarding tracking error and computational tractability. These comparisons allow us to observe the worst-case scenarios for such control architectures and provide an informed view of possible trade-offs in demanding transportation applications.

On the third problem, we focus on control methods for formation control using imaging sensors and informed neighbor motion prediction. We propose a decentralized model predictive controller based on collaborative behavior among the agents in the formation. Then, we develop a formation controller that uses image features and one-range measurement to coordinate the agents.

The last problem of this thesis introduces a novel architecture for ground space robotics testbeds. We propose modular platforms, driven by open-source firmware and software stacks, capable of mimicking spacecraft operating inside space stations or in orbit conditions. We validate these systems in a new space robotics facility at KTH built during this thesis. Lastly, we conclude this monograph with a short analysis of the achieved objectives and an overview of future research directions.

This doctoral thesis addresses the rapidly evolving landscape of computer security threats posed by advancements in artificial intelligence (AI), particularly large language models (LLMs). We demonstrate how AI can automate and enhance cyberattacks to identify the most pressing dangers and present feasible mitigation strategies. The study is divided into two main branches: attacks targeting hardware and software systems and attacks focusing on system users, such as phishing. The first paper of the thesis identifies research communities within computer security red teaming. We created a Python tool to scrape and analyze 23,459 articles from Scopus's database, highlighting popular communities such as smart grids and attack graphs and providing a comprehensive overview of prominent authors, institutions, communities, and sub-communities. The second paper conducts red teaming assessments of connected devices commonly found in modern households, such as connected vacuum cleaners and door locks. Our experiments demonstrate how easily attackers can exploit different devices and emphasize the need for improved security measures and public awareness. The third paper explores the use of LLMs to generate phishing emails. The findings demonstrate that while human experts still outperform LLMs, a hybrid approach combining human expertise and AI significantly reduces the cost and time requirements to launch phishing attacks while maintaining high success rates. We further analyze the economic aspects of AI-enhanced phishing to show how LLMs affect the attacker's incentive for various phishing use cases. The fourth study evaluates LLMs' potential to automate and enhance cyberattacks on hardware and software systems. We create a framework for evaluating the capability of LLMs to conduct attacks on hardware and software and evaluate the framework by conducting 31 AI-automated cyberattacks on devices from connected households. The results indicate that while LLMs can reduce attack costs, they do not significantly increase the attacks' damage or scalability. We expect this to change with future LLM versions, but the findings present an opportunity for proactive measures to develop benchmarks and defensive tools to control the misuse of LLMs.

Integrating high-efficiency heat pumps to renewable electricity will significantly accelerate decarbonization progress. Despite the advancements in smart sensors and communication technologies that enhance data generation in heat pump units, much of this data remains underutilized for performance analysis. The primary issue is that the data often exhibits incompleteness, inconsistency, and inaccuracies. Consequently, data collection and storage impose an economic burden on manufacturers and end-users, and they have limited returns. This thesis aims to unlock the potential of various data resources, delivering knowledge-based services and addressing gaps in data availability and utilization.

The dissertation introduces the Data-Information-Knowledge-Service (DIKS) framework as an adaptation of the traditional Data-Information-Knowledge-Wisdom (DIKW) hierarchy, emphasizing the practical application of turning inadequate data into knowledge-based services in heat pump technologies. The transformative process within the DIKS pyramid is illustrated, detailing how each layer converts inadequate raw heat pump data into actionable, knowledge-based services. It begins with the aggregation and integration of various data types, followed by advanced processing techniques to refine data quality and identify significant patterns. This foundation of knowledge is then applied to improve heat pump services, demonstrating the practical benefits of this structured approach.

Five different scenarios are examined utilizing different types of data, including high-quantity, low-quality in-situ measurements, high-quality, low-quantity lab data, and technical specifications. The first scenario, utilizing in-situ field measurement data, develops an artificial neural network (ANN) model to create soft sensors that compensate for the absence of costly physical sensors in heat pump systems. These soft sensors use incomplete data to accurately estimate essential heat pump parameters, supporting functions like operational monitoring, fault detection, smart energy management, and developing digital twins. The second scenario, also utilizing in-situ field measurement data, focuses on models that prioritize minimal input features, enhancing models' usability across various installations. These models estimate power consumption effectively and compensate for the lack of physical power meters, thus facilitating network planning and smart control, etc. In the third scenario, transfer learning techniques are employed to estimate heat pump performance with limited lab data, particularly for natural refrigerants in the context of the phasing out the fluorinated refrigerants. This approach utilizes knowledge from existing refrigerant data to improve model reliability and accuracy, aiding in refrigerant selection. The fourth scenario develops polynomial regression models from technical specifications to evaluate heat pump performance without dynamic measurements. These models assist in tracking the system performance. The final scenario introduces semi-empirical models that use thermodynamic and heat transfer correlations to enhance the understanding of physical meaning. These models are designed with reduced parameters, improving services such as fault diagnosis and system maintenance.

In conclusion, this thesis identifies common status and key issues related to inadequate data from heat pump systems. Furthermore, this thesis proposes solutions to transfer inadequate data to useful structured information and develops data-driven models according to the characteristics of different data types. The models are validated against measurements, demonstrating accurate results. This thesis demonstrates the application of the DIKS framework to effectively harness underutilized inadequate data from heat pump systems, which not only provides insights to alleviate the economic burdens placed on manufacturers and/or users related to data cost but also enable the services of heat pump systems.

This thesis contributes to the field of stochastic bandits by exploring the fundamental limits (information-theoretic lower bounds) of three prevalent objects in various reinforcement learning applications, through a collection of five distinct papers. Each paper presents a novel perspective under a specific structure and scenario. The first paper investigates regret minimization in decentralized multi-agent multi-armed bandits. The second and third papers delve into pure exploration with fixed confidence in a broad spectrum of structured bandits. The last two papers focus on offering new insights into the best arm identification with a fixed budget. 

In the first paper, two popular scenarios in a decentralized multi-agent setting are addressed, one involving collision and the other not. In each of them, we propose an instance-specific optimal algorithm. Interestingly, our results show that the fundamental limits match the ones in the centralized analogue.The second paper introduces a simple but versatile algorithm, Frank-Wolfe Sampling, which achieves instance-specific optimality across a wide collection of pure explorations in structured bandits. Meanwhile, the numerical results and current studies demonstrate the strong numerical performance of our algorithm in various pure exploration problems. However, Frank-Wolfe Sampling is not computationally efficient when the number of arms is extremely large. To address this issue, the third paper introduces Perturbed Frank-Wolfe Sampling, which can be implemented in polynomial time while maintaining the instance-specific minimal sample complexity in combinatorial semi-bandits.

Unlike the sample complexity or regret minimization discussed above, characterizing the fundamental limit of the error probability in best arm identification with a fixed budget remains a challenge. The fourth paper addresses this challenge in two-armed bandits, introducing a new class of algorithms, stable algorithms, which encompass a broad range of reasonable algorithms. We demonstrate that no consistent and stable algorithm surpasses the algorithm that samples each arm evenly, answering the open problems formulated in prior work. In general multi-armed bandits, the final paper in this thesis presents, to our knowledge, the first large deviation theorem for the generic adaptive algorithm. Based on this, we provide the exact analysis of the celebrated algorithm, Successive Rejects. Furthermore, this new large deviation technique allows us to devise and analyze a new adaptive algorithm, which is the current state-of-the-art to the best of our knowledge. This thesis provides new insight for deriving fundamental limits in various online stochastic learning problems. This understanding guides us to develop more efficient algorithms and systems.

In this thesis, the production of bio-based and biodegradable foams from wheat proteins (wheat gluten) is presented. Wheat gluten, a side stream from the ethanol/starch industry, was processed using both batch and continuous foaming methods. To produce the foams, glycerol, an effective protein plasticiser, was used, with two foaming agents common in food: sodium bicarbonate (SBC) and ammonium bicarbonate (ABC), to generate soft foams with both closed-and open-cell structures. ABC proved to be a better foaming agent in foam extrusion than SBC, but SBC performed well in the batch methods. Due to ABC’s low decomposition temperature, extrusion could take place at a temperature as low as 70 °C.

     Three different multifunctional additives (citric acid, gallic acid and genipin) were also used to influence and improve the foam properties. The mechanical properties showed that some of the materials could be potentially useful in cushioning and sealing applications. The foams also showed a high absorption of saline (model substance for body fluid) and blood (in the form of sheep’s blood), even under mechanical pressure. Based on these results, a wheat gluten-based product was manufactured as a proof-of-concept.

     The degradability of the foam in various relevant environments was studied. It was found that some foams degraded almost completely in soil after 8 weeks and in alkaline water after 5 weeks. It was also demonstrated that the foam also worked as a good fertilizer. As an alternative to direct composting when the foam is no longer used, the possibility of reusing the foam in a different form was evaluated. In this context, it was possible to produce plastic films from the foam.

A growing demand has emerged for new point-of-care (POC) platforms capable of delivering reliable clinical data in real-time through minimally invasive procedures. Currently, the majority of clinical data is derived from analyzing collected biological samples, primarily blood or plant sap. Unfortunately, these methods cost discomfort to patients, and are even destructive to plants. For example, conventional sap collection requires sacrifice the plants. The lack of portable tools for fast, on-site patient/plant monitoring has driven research into alternative strategies using biosensors.

The glucometer (i.e. blood glucose meter) stands out as one of the most successful examples of a POC device. It reflects the key features we strive for in such a biosensing platform: minimal limitations on who and where it can be used, combined with high reliability and affordability. Electrochemical readouts are advantageous in this case due to its fast response, wide detection range, and ease of integration into portable devices. This doctoral thesis introduces advancements of electrochemical biosensing platforms for detecting various analytes both in humans and plants. The key findings are summarized in the Results and Discussion section based on the four published research articles.

Briefly, the first type of electrochemical biosensor was developed for the determination of glycine in various human biofluids (e.g., blood, sweat, and urine). Considering the increasing importance of amino acid detection for clinical applications, we then created a new biosensing platform based on microneedles (MN) that aims to measure in dermal interstitial fluid. This minimally invasive strategy highlights the novelty of our second work. Third, we extended MN-based biosensors to another important analyte, lactate, which is previously widely analyzed in sweat. Finally, we demonstrated the first example of applying the MN sensors for continuous and real-time plant monitoring.

This thesis presents an innovative approach to the development of a fully integrated multi-channel neuromuscular electrical stimulator (NMES) system and a multi-channel surface electromyography (sEMG) acquisition system for musculoskeletal (MSK) healthcare applications. The main objective is to integrate therapeutic and diagnostic tools into a compact wearable device, enabling closed-loop electrical therapy. By leveraging advancements in semiconductor technology, this thesis explores the implementation of application-specific integrated circuits (ASIC) to combine high-voltage (HV) NMES and low-voltage sEMG signal acquisition circuits on a single chip using a 180 nm bipolar-CMOS-DMOS technology.  

The research addresses several key challenges in existing NMES and sEMG systems: the need for a compact, multi-channel NMES device; the need for safe electrical muscular stimulation; the need for spatiotemporal information through multi-channel acquisition; and the need for high channel counts and efficient chip area utilization. To overcome these challenges and advance the NMES technology, this thesis proposes several innovative circuit solutions, including a configurable HV-tolerant multi-channel stimulator, an integrated fail-safe protection circuit, and an inductorless on-chip HV generator. Additionally, channel-sharing techniques for multi-channel biopotential acquisition are comprehensively explored, and a novel frequency-division multiplexed architecture is proposed, featuring low noise, low power consumption, and minimized system complexity. 

A significant contribution of this thesis work is the integration of multi-channel NMES and sEMG systems in an ASIC, leading to the development of a real-time embedded system for wearable medical applications. This embedded system incorporates the proposed ASIC for bidirectional interfacing with muscles and an off-the-shelf microcontroller for data acquisition, signal processing, and stimulation pattern control. The proposed system facilitates the continuous collection of vital physiological conditions (e.g., motion intention, contraction force, and fatigue level) of the human muscular system, enabling timely adjustments and interventions via electrical stimulation. In-vivo experimental results showcase its potential to enhance electrical therapy outcomes through closed-loop control and pave the way for improved patient care.

The global shift towards sustainable development requires the replacement of fossil-based materials with renewable alternatives. Lignin, a complex aromatic biopolymer derived from lignocellulosic biomass, represents one of the most abundant sources of renewable carbon. Retrieved as a byproduct from the pulp and paper industry, lignin has thus far been underutilized despite its potential. Its unique chemical structure, characterized by phenolic units linked through various interunit linkages, makes it a strong candidate for creating durable resistant materials. However, lignin's complex and heterogeneous structure, as well as its limited reactivity, present challenges for its use. In this work, lignin has been investigated for thermosetting applications through different methodologies, including extraction, fractionation, chemical modification, and direct utilization of unmodified lignin. This present research demonstrates different ways to develop materials with enhanced mechanical, thermal, and chemical properties, suitable for industrial applications. To achieve this, various methodologies and lignin sources were employed. Lignin was extracted through a mild extraction from wheat straw in order to valorize agricultural products. Microwave-assisted fractionation was employedin order to isolate lignin fractions with more tunable properties and increased reactivity. Chemical modifications, including epoxidation and allylation, were performed to enhance the reactivity of lignin and improve its compatibility in thermosetting formulations. These modified lignins were incorporated into epoxy-based coatings and thiol-ene systems, demonstrating their potential in producing durable and high-performance materials. In addition to modified lignin, unmodified lignin was directly utilized in coating formulations. This thesis demonstrates that both modified and unmodified lignin can be successfully integrated into thermosetting systems, furnishing materials that meet or exceed the performance of conventional fossil-based counterparts. The work emphasizes the advantages and limitations of each method, highlighting the importance of optimizing processing efficiency, material performance, and environmental sustainability.

Machine Learning (ML) models have achieved significant success in representation learning across domains like vision, language, graphs, and tabular data. Constructing effective ML models hinges on several critical considerations: (1) data representation: how to represent the input data in a meaningful and effective way; (2) learning objectives: how to define desired prediction target in a specific downstream task; (3) model architecture: which representation learning model architecture, i.e., the type of neural network, is the most appropriate for the given downstream task; (4) training strategy: how to effectively train the selected ML model for better feature extraction and representation quality.

This thesis explores representation learning and parallelization in machine learning, addressing how to boost model accuracy and reduce training time. Our research explores several innovative approaches to improve the efficiency and effectiveness of ML applications on graph, tabular, and time-series data, with contributions to areas such as combinatorial optimization, parallel training, and ML methods across these data types. First, we explore representation learning in combinatorial optimization and integrate a constraint-based exact solver with the predictive ML model to enhance problem-solving efficiency. We demonstrate that combining an exact solver with a predictive model that estimates optimal solution costs significantly reduces the search space and accelerates solution times. Second, we employ graph Transformer models to leverage topological and semantic node similarities in the input data, resulting in superior node representations and improved downstream task performance. Third, we empirically study the choice of model architecture for learning from tabular data. We showcase the application of tabular Transformer models to large datasets, revealing their ability to create high predictive power features. Fourth, we utilize Transformer models for detailed user behavior modeling from time-series data, illustrating their effectiveness in capturing fine-grained patterns. Finally, we dive into the training strategy and investigate graph traversal strategies to improve device placement in deep learning model parallelization, showing that optimized traversal order enhances parallel training speed. Collectively, these findings advance the understanding and application of representation learning and parallelization in diverse ML contexts.

This thesis enhances representation learning and parallelization in ML models, addressing key challenges in representation quality. Our methods advance combinatorial optimization, parallel training, and ML on graph, tabular, and time-series data. Additionally, our findings contribute to understanding Transformer models, leading to more accurate predictions and improved performance across various domains.

Proteins are one of the fundamental building blocks of life as we know it. They are central to various biological processes and pivotal parts of essential procedures in the healthcare, food and sustainability industries. Over the past few years, significant research efforts have been made to employ bio-based strategies as alternative pathways to shift away from human dependency on petroleum-based polymers, placing protein as central building blocks for sustainable material development. In this framework, this thesis explores a specific class of material building blocks derived from the food we eat, referred to as protein nanofibrils (PNFs). Particular attention is paid to PNFs from soy protein isolate (SPI) and whey protein isolate (WPI) formed at low pH and high temperature, mirroring processes commonly found in food processing and cooking. The presented work focuses explicitly on how these fibrils assemble from acid hydrolysis of the initial food proteins into smaller aggregation-prone species, how aggregation occurs and how we can potentially process fibrils as valuable materials for society. 

A critical step in forming fibrils at low pH and high temperature is the heat-induced acid hydrolysis of protein chains into shorter peptides. The work first uses SPI as a model protein source to understand PNF formation. By determining the activation energy for SPI hydrolysis and comparing it with the lower activation energies for the aggregation processes of one of the peptide building blocks, BB2, we establish that hydrolysis is the with the highest energy of activation. Computer simulations are then employed to replicate the hydrolysis of proteins to model peptide production and degradation from rates of protein hydrolysis. With this, a connection between monomer production and fibrillation kinetics can be established. With this, numerical simulations of BB2 at 90 °C and pH two are then attempted, demonstrating that aggregation of this peptide alone does not replicate the fibrillation behaviour of SPI.

The thesis continues focusing on the fibrillation of food proteins and their potential safety for humans, now exploring how food fibrils affect the aggregation behaviour of Aβ42 aggregation, a hallmark of Alzheimer’s disease. The results show that seeds from food amyloids do not accelerate Aβ42 aggregation, with lysozyme- and oat-derived fibrils even delaying the aggregation of Aβ42. Further kinetic analysis reveals that aggregation inhibition is likely due to interactions between Aβ42 aggregates and food-derived fibrils, likely affecting the pathway for secondary nucleation in Aβ42 assembly. 

In contrast, seeding accelerates fibrillation in WPI, leading to the formation of longer fibrils that can reach the percolation threshold necessary for gelation and material formation. However, the results also indicate that percolation theories might need to be refined to better frame and predict the behaviour of PNFs and how they interact to form macromolecular structures.

Finally, the usage of food-derived WPI fibrils in material applications is also investigated. Fibril-based aerogels exhibit distinctive microstructural differences from aerogels without fibrils and enhanced pollutant adsorption capabilities, represented using the model molecule ibuprofen. Additionally, preliminary results show that fibrillar hydrogels can be used as electrolyte matrices for energy storage, displaying a broad operational potential window and excellent rate performance. 

Combined, the results provide important insights regarding the formation of fibrils at low pH and high temperature, their potential non-hazardous nature for human consumption, and their application in areas such as water purification and energy storage.

Understanding the complex relationship between urban environments and crime is crucial for effective urban planning and crime prevention strategies. Spatial analytical methods have provided valuable knowledge into crime patterns, enabling the detection of crime-concentrated environments and informing law enforcement operations and urban planning interventions. The international literature highlights the increasing use of remote sensing in crime analysis, driven by improved data availability and accuracy. Given the potential of this approach, this thesis investigates the use of spatio-temporal data analyses, particularly the incorporation of remote sensing data along with traditional socio-demographic and land use indicators in understanding the dynamics of crime in urban environments. Four crime categories—cannabis-related crimes, street theft, residential burglaries, and sexual crimes—are investigated using Stockholm City in Sweden as a case study. Remote sensing data, particularly very high-resolution imagery, combined with machine learning algorithms, such as the Random Forest classifier, facilitate the prediction of crime risk areas and the identification of environmental factors associated with crime occurrences. While the thesis reflects upon the advantages and disadvantages of using remote sensing in crime analyses, findings offer practical insights for policymakers, urban planners, and law enforcement agencies, enabling the development of data-informed strategies to foster safer and more resilient urban environments.

The global shift towards renewable energy sources and the electrification of transportation necessitates advanced energy storage solutions, with lithium-ion batteries (LIBs) at the forefront. However, conventional batteries with liquid electrolytes in LIBs pose several limitations such as flammability, poor chemical stability, leakage risks, overall safety concerns, and limited processability. This thesis investigates hybrid polymer-liquid electrolytes (HEs) as an alternative to address these issues in LIBs as well as a way to obtain additional functionalities e.g. improved structural integrity.

The research is organized into four main studies. Paper I focuses on the three-dimensional (3D) reconstruction and analysis of HE structures. Using focused ion beam-scanning electron microscopy (FIB-SEM), the study reveals the complex, interconnected pore networks within HEs that are critical for ionic conductivity and mechanical stability.

Paper II explores the impact of porosity on the ionic and molecular mobility within HEs. By varying the liquid electrolyte content, the study demonstrates how increased porosity enhances ion mobility, directly correlating with improved electrochemical performance. Nuclear magnetic resonance (NMR) diffusion experiments further elucidate the transport mechanisms within the polymer matrix, showing a significant increase in ion diffusion rates with higher electrolyte content.

Paper III examines the role of nanosized carbon black (CB) particles in the polymerization-induced phase separation (PIPS) process used to synthesize HEs. The addition of CB improves the conductivity of HEs without compromising their morphological integrity. The study finds that even small amounts of CB can substantially enhance the overall conductivity, making CB-rich HEs potential candidates for multifunctional roles within battery electrodes, such as conductive binders.

Paper IV evaluates the practical application of HEs by integrating them into commercial LIB electrodes. The HE-infused electrodes maintain their structural and electrochemical properties even after multiple charge-discharge cycles, proving their potential for use in commercial applications.

This thesis contributes to the development of multifunctional electrolytes that not only address the safety issues associated with liquid electrolytes but also advance multifunctionality in LIBs. The methodologies and findings presented provide a foundation for future research in high-performance, safer, and more sustainable battery technologies.

This thesis contains a collection of efficient quantum algorithms and classical pre- and post-processing algorithms for various number- and group-theoretical problems with concrete cryptanalytical applications. More specifically, the focus is on the integer factoring problem (IFP), the discrete logarithm problem (DLP), and the order-finding problem (OFP), and on certain important variations of these problems. It is for example explained

• how the OFP and complete IFP may be solved efficiently in a single run of a reduced version of Shor's algorithm for the OFP with very high success probability, as evidenced by a lower bound that is 1 − o(1),
• how the short DLP in groups of unknown order may be solved efficiently and with an advantage over Shor's algorithm for the DLP (that furthermore requires the order to be known), as evidenced by a lower bound on the single-run success probability that is 1 − o(1),
• how the RSA IFP may be solved efficiently and with an advantage over Shor's algorithm for the IFP, by reducing the RSA IFP to a short DLP,
• how the DLP may heuristically be solved efficiently, and with high success probability that is 1 − o(1), in a single run of a reduced version of Shor’s algorithm for the DLP that admits efficient implementation,
• how the DLP may be solved in groups of unknown order, such as Schnorr groups, by jointly solving an OFP and a DLP at lower overall cost but higher per-run cost compared to first solving the OFP and then the DLP,
• how tradeoffs — between the number of group operations that are evaluated quantumly and the number of runs that are required — may be achieved in all of the aforementioned algorithms, and
• how all of the aforementioned algorithms may be simulated efficiently classically, for arbitrary large cryptographically relevant problem instances, when the solution to the instance is known, enabling the number of runs required when performing tradeoffs to be tightly estimated.

It is furthermore for example explained

• how Regev's recent quantum algorithm for the IFP may be extended to the DLP and OFP, yielding i) a reduction in the circuit size and depth by a factor compared to Shor for these problems when working in , for N an n-bit integer, at the expense of performing runs, and ii) slight improvements of Regev's orginal algorithm for the IFP,
• how the above algorithms may be simulated efficiently classically, for large special-form problem instances that are classically tractable, and
• that Regev’s post-processing, and the post-processing for our extensions, are robust to an arbitrarily large fraction < 1 of the runs being bad.

The above has implications for cryptography based on the intractability of the IFP and DLP — including but not limited to for widely used schemes such as RSA, DSA and Diffie–Hellman. The algorithms in this thesis are all derived from, and make notable improvements to, the seminal works of Shor. Some of the contributions are joint works with other authors.

This thesis investigates the development and characterization of advanced materials derived from renewable sources, with a focus on cellulose nanofibrils (CNFs), lignocellulose nanofibrils (LCNFs), and protein nanofibrils (PNFs). The research aims to understand the intricate dynamics and interactions at the nano-scale, which are essential for enhancing the mechanical properties, sustainability, and practical applications of these materials.

The study begins with the exploration of CNFs in the presence of Helux, a dendritic polyampholyte. By examining the spinnability, alignment, and mechanical properties of CNF-composite filaments, the research demonstrates how Helux influences the assembly process. While Helux reduces fibril alignment due to increased rotary diffusion, it simultaneously creates a robust 3D network through ionic interactions, resulting in a trade-off that enhances the toughness and strength of the filaments.

The work then extends to LCNFs derived from unbleached softwood kraft pulps with varying lignin content. Lignin adds complexity to the alignment and mechanical properties of LCNFs, acting as a natural adhesive that enhances interfibrillar interactions. The study shows that LCNF-filaments exhibit higher tensile strength and modulus compared to CNF-filaments, particularly when lignin content is optimized. Additionally, LCNF-based foams are evaluated for their mechanical properties and lower cumulative energy demand, highlighting their potential for sustainable material applications.

In the final section, the thesis examines PNFs, focusing on how their morphology affects alignment and assembly under different flow conditions. Using microfluidic techniques and in situ small-angle X-ray scattering (SAXS), the research reveals the significant role of nanofibril morphology in forming hierarchical structures. The study also explores the use of genipin as a cross-linker to enhance the mechanical properties of PNF-based microfibers, demonstrating how cross-linking can improve fiber strength and ductility.

This thesis advances the field of sustainable material development by offering insights into the factors that influence the alignment, assembly, and mechanical performance of nanofibril-based materials, contributing to the creation of high-performance, environmentally friendly materials.

Point-of-need devices perform analytical tests that help inform decisions where they are needed, away from modern lab infrastructure, be it in-field or in resource-poor settings. They have many applications, including veterinary medicine, agriculture, food safety, environmental monitoring, and forensics. In medical diagnostics, such devices are called point-of-care tests, and they could help combat societal challenges such as the spread of epidemic diseases and providing adequate healthcare in developing countries. Point-of-care devices could also be wearable to non-invasively monitor body fluids such as sweat or urine from the patient. Ideal point-of-care devices conform with the REASSURED criteria, that they should be Real-time connected, Easy to collect samples, Affordable, Sensitive, Specific, User-friendly, Rapid, robust, Equipment-free, environmentally friendly, and Deliverable to the end user.

We have here developed Point-of-need devices based on textiles and Printed Circuit Boards (PCBs); both well-established technologies that could offer low-cost mass production using existing industrial resources. Specifically, we have made electrochemical biosensors based on gold-coated yarn in a rolling architecture, as well as combined with wicking Coolmax® yarn acting as microfluidic channels in wearable systems, enabling advanced textile-based diagnostic devices suitable for automation or machine-stitching into fabrics. We also showed biosensors based on gold-coated PCBs that can connect to portable potentiostats for electrochemical detection and have integrated heating for isothermal nucleic acid amplification.

A multitude of physical phenomena are accurately modeled by partial differential equations (PDEs). These equations are complicated to solve in general, and when an analytical solution is not able to be found, a numerical method can give an approximate solution. This can be very useful in many applications. This thesis explores the development and analysis of cut finite element methods (CutFEM) for discretising PDEs with a focus on preserving divergence conditions essential in applications such as fluid dynamics and electromagnetism. CutFEM has been developed with the aim to simplify distretising PDEs in domains with complicated geometries, by allowing the geometry to be positioned arbitrarily relative to the computational mesh. Traditional CutFEM have failed to maintain the divergence conditions, leading to numerical inaccuracies. Following the mixed finite element method (FEM) framework, the research contained herein introduces novel strategies that preserve the divergence at the discrete level and addresses other key challenges when discretizing PDEs in geometries unfitted to the computational mesh. For example, the techniques are also able to control the condition number of the linear systems. The virtual element method (VEM) is another method able to handle complicated geometries. It does this by allowing for a mesh to be constructed from general polytopal elements, not just triangles or rectangles. One work of the thesis investigates the spectral condition number of the mixed VEM, demonstrating the effectiveness of auxiliary space preconditioning in bounding spectral condition numbers independently of mesh element aspect ratios.

Surface science techniques applied to simplified model systems have long been used to study catalytic mechanisms and aspects of surface reactions that aredifficult to isolate in real catalytic reactors. Experimental techniques are usually combined to obtain information on structure, surface kinetics, surfacedynamics, and reaction chemistry. Recently, a main focus in surface science has been to attempt to ”close the gap” to real catalysis: Pushing experimental techniques to higher pressure than typical ultra-high vacuum (UHV) conditions, working with structurally more complex catalysts and introducing some of the complexity of real reaction conditions.

In this thesis, experimental studies on two model systems are presented. In the first part, the reaction of naphthalene on Ni(111) and Fe(110) is examined as a model for catalytic tar decomposition used in biomass gasification. The effect of sulfur, a typical impurity in biomass, on the dehydrogenation ofnaphthalene on Ni(111) is elucidated with XPS and STM, and shown to lead to an inhibition of carbon bulk dissolution. The decomposition of naphthalene on Fe(110) was studied on the clean surface and in the presence of oxygen with XPS, TPD and SFG to enable a direct comparison to Ni(111). A very similar activity towards naphthalene decomposition is observed, as well as key differences in carbon-carbon bond cleavage, carbon formation and reactivityof ”dirty” surfaces.

In the second part, CO oxidation on Pd(110) is studied as a model system for palladium catalysts and a good example surface for the effect of surfacereconstruction and increase of pressure above UHV. The reaction was examined using the recently developed near-ambient pressure velocity map imaging (NAP-VMI) technique, which enables the simulatenous measurement of kinetic constants and dynamic information at pressures up to 10⁻³  mbar. Using the unique capabilities of VMI, two reaction channels with fast diffusion could be attributed to CO adsorption sites and an effective activation energy extracted.

The research presented here demonstrates the usefulness of these surfacescience methods in understanding catalytic mechanisms. It also illustrates some key limitations and opportunities for future developments in the field.

In the context of rising global health challenges, the mechanistic investigation and

treatment of complex diseases, including cancer, liver diseases, has emerged as a

vital focus in scientific research. A thorough understanding of basic biological

processes is crucial for the development of tools that aid in diagnosing, monitoring,

and treating human diseases. This doctoral thesis investigates the molecular

mechanisms underlying complex human diseases, with an emphasis on discovering

novel therapeutic targets and compounds though systems biology approaches. By

leveraging large-scale transcriptomic data, this work aims to uncover novel insights

into disease biology that can drive drug repositioning and precision medicine. The

thesis integrates various computational strategies and biological frameworks to

connect gene expression patterns with disease progression and therapeutic

opportunities, focusing primarily on cancer and metabolic disorders.

The studies compiled in this thesis contribute to the understanding of human disease

biology through the systematic analysis of gene expression profiles and the

application of network-based methodologies. Paper I introduces the Human

Pathology Atlas, providing an in-depth analysis of gene expression prognostic

features across different cancer types, which improves our understanding of

relationships between gene expression and disease outcomes. Paper II and Paper

III employ gene co-expression network analysis combined with drug repositioning

strategies, identifying promising therapeutic candidates for hepatocellular

carcinoma and pancreatic ductal adenocarcinoma, respectively. These studies

illustrate how network-based approaches can locate key molecular targets and

potential repurposable drugs for various cancer types.

In Paper IV, we apply a network-based approach to investigate the dysregulated

transcriptional regulation in non-alcoholic fatty liver disease (NAFLD). This study

identifies critical genes and pathways involved in the disease progression, providing

new insights into the pathophysiology of NAFLD. Lastly, Paper V presents

comprehensive review on the emerging role of PKLR in liver diseases, highlighting

its connection to metabolic diseases. This review discusses PKLR’s potential as a

therapeutic target, providing a foundation for future studies in metabolic disease

research.

In summary, this thesis contributes to the field of systems biology by integrating

gene expression and network methodologies, offering innovative strategies for

therapeutic development and personalized medicine across complex diseases.

The design of modern embedded systems is increasingly complex. Different applications must share a heterogeneous embedded platform while satisfying demanding design requirements. The financial cost and engineering effort to implement such modern embedded systems are proportional to this increasing complexity. Model-driven-engineering (MDE) approaches mitigate this complexity by using well-defined models as the central elements of the design process. Namely, these models can be used to automate design process activities, such as design space exploration (DSE).

This thesis brings two contributions within this context. First, a novel meta-modelling approach for MDE with its implementation ForSyDe IO. Second, the design space identification (DSI) approach with its implementation IDeSyDe. ForSyDe IO is a language-agnostic MDE framework that promotes cooperative development through its underlying common model and capabilities. IDeSyDe is a modular and tuneable MDE DSE framework that enables the composable construction of DSE solutions via DSI. Of greater practical value, DSI and its implementation IDeSyDe seamlessly combine different DSE solutions to improve the overall exploration performance.

To ensure these contributions have immediate practical value, this thesis also presents four different MDE DSE scenarios originating from industrial cooperation incorporated into IDeSyDe. The industrial cooperation includes periodic workloads from avionics and automotive contexts and digital signal processing applications from academic contexts. ForSyDe IO was used to express each case study’s applications, platforms and requirements, which shows how it can be incrementally adapted for different scenarios. At the same time, IDeSyDe is used to construct DSE solutions for each case study in a fashion that displays IDeSyDe’s modularity, tuneability and capabilityfor synergizing different DSE solutions.

The case studies show qualitatively how the contributions, DSI in particular, aid in providing a cooperative and modular environment for developing MDE DSE solutions. At the same time, the numeric results of case studies show quantitatively that the overhead of the DSI automated proceduresis negligible compared to the overall DSE process and that the transparent combination of explorers improves the overall exploration performance without additional development effort.