The effect of magnetic shear on ballooning-driven plasma edge turbulence is studied through nonlinear simulations complemented by linear numerical and analytical investigations. Nonlinear, 3D, global, flux-driven simulations using the GBS code show that the scale separation between radial, x, and poloidal, y, size of turbulent eddies, k x ≪ k y , considered by Ricci, Rogers, and Brunner 1 and extensively used to predict pressure gradient lengths, SOL width, particle and heat fluxes, is observed with high magnetic shear. In contrast, for low magnetic shear, k x ∼ k y is observed, with fluctuation properties resembling those shown by recent low-shear stellarator simulations reported in Coehlo, Loizu, Ricci, and Tecchiolli 2. Global linear investigations of the ballooning mode qualitatively captures the transition in mode structure with varying magnetic shear, showing that k x ≪ k y is achieved with sufficiently strong poloidal mode coupling enhanced by increasing magnetic shear, resistivity, toroidal mode number, and equilibrium gradient scale length. This is confirmed by an analytical study considering a dominant poloidal mode and its sidebands, which highlights that the poloidal mode structure is determined by curvature and k ∥ effects.

Humans and animals are able to rapidly, and with reasonable accuracy, estimate the number of objects in a visual scene. Visual-evoked potential studies have described a sequence of functionally distinct stages associated with numerosity perception. Recently, a specialized cognitive-perceptual system for the numerosity estimation for human stimuli (NEH), distinct from that for non-social stimuli, was revealed using virtual reality, revealing a stable NEH overestimation bias (ie perceiving more people than were shown). This bias was further increased when a social hallucination characterized by the false perception of another invisible person (presence hallucination) was induced robotically and repeatedly prior to NEH trials. However, little is known about the brain mechanisms of NEH and robot-induced presence hallucination. Here we combined virtual reality and robotically-induced presence hallucination with visual-evoked potentials to investigate the neural mechanisms and processing steps of NEH. We report that NEH induces numerosity-related components as observed for non-human numerosity stimuli and, critically that experimentally-induced presence hallucination selectively modulated the P2p component, whose amplitude correlated with NEH overestimation magnitude. This effect was localized in left extrastriate cortex, showing that robot-induced presence hallucination are integrated with NEH processing during the P2p time period and relying on social numerosity mechanisms in extrastriate cortex.

Glacier retreat in the Alps has progressed dramatically since the 1980s as a result of climate change and will be even more pronounced in the future. Consequently, flow regimes and sediment yield from watersheds will change significantly. Reservoirs are exposed to continuous sedimentation, reducing storage volume and posing exploitation and security risks. This chapter is primarily based on a literature review of the Gries watershed (currently mostly glacier-covered above 2,400 m asl), including the reservoir, glacier, and dam. The length of the Gries Glacier has been recorded since 1847, and annual mass balance data have been available since 1961. Future glacier evolution models, coupled with a hydrological model that considers global warming, predict that the length of the glacier will continue to decline, such that by around 2060, the entire catchment will be virtually glacier-free. Runoff and erosion will shift to a pluvio-nival regime, and associated water resources will become less reliable. Past surveys of sediment yield and reservoir sedimentation allow the characterization of reservoir sedimentation patterns, which are dominated by plunging turbidity currents. Sustainable sedimentation management measures will preserve the reservoir's storage volume, safety, and economic operability beyond the disappearance of the glacier.

Strongly correlated physics has been a central topic in condensed matter physics for decades, as it gives rise to a wide range of emergent phenomena such as unconventional superconductivity, quantum spin liquids and fractionalized excitations. In recent years, twisted van der Waals materials have emerged as an ideal platform for exploring strong correlated physics. These moire systems provide a highly tunable environment in which strong correlations can coexist and interplay with topology, enabling the discovery of novel quantum phases. One of the most fascinating manifestations of such quantum matter is unconventional superconductivity in twisted graphene systems. Since its discovery in 2018, extensive efforts have been made to understand the origin of superconductivity and the nature of the associated order parameters. While significant progress has been achieved such as evidence for nodal order parameters, a complete and unified understanding of the superconducting mechanism remains elusive.

In this thesis, we focus on the investigation of superconductivity in twisted trilayer graphene systems. First, we study magic-angle twisted trilayer graphene (MATTG) and demonstrate the existence of double-dome superconductivity within a narrow displacement field window. Second, we study mirror-symmetry-broken twisted trilayer graphene and demonstrate the existence of a supermoire lattice from the interference of two moire lattice, along with the presence of superconductivity and strong correlation effects on the supermoire length scale. Our results offer new insights into the nature of superconductivity in moire graphene systems and suggest new pathways for designing and discovering novel quantum phases.

Neurological disorders such as spinal cord injury and Parkinsonian syndromes often result in sustained autonomic dysfunction, leading to hemodynamic instabilities that threaten neurological recovery and impact quality of life. Among these populations, hemodynamic instabilities are gaining more attention as they deter individuals from partaking in daily life activities, rehabilitation, and regaining more appreciated functions such as mobility. One such instability is orthostatic hypotension (OH), or a severe drop in blood pressure upon standing. The underlying pathophysiology often involves a failing baroreflex, a homeostatic mechanism that spans the brain stem, spine, and heart. When communication from the brain to the spine and periphery is compromised via an injury or neurodegeneration, and the baroreflex is compromised, individuals face increased risk of chronic fatigue, fainting, and long-term cardiovascular and cerebral risks such as heart attacks and stroke.

Hence, this thesis aims to develop a neuroprosthetic-based solution combining spinal cord stimulation and hemodynamic monitoring to help individuals with neurological conditions manage their blood pressure instabilities in an autonomous fashion.

This thesis first quantifies the clinical burden of OH in SCI in 1,479 individuals and exposes the ineffective treatment of autonomic complications despite the use of conservative measures such as pharmacology and supportive garments. To address this clinical burden, we developed a purpose-built implantable system based on biomimetic epidural electrical stimulation (EES) of the spinal cord that immediately triggered robust pressor responses. The system durably reduced the severity of hypotensive complications in 14 individuals with SCI, removed the necessity for conservative treatments, improved quality of life, and enabled superior engagement in activities of daily living. Detailed spatial and temporal mapping identified the lower thoracic spinal cord as the optimal stimulation target, and a purpose-built stimulation platform, the ARC\textsuperscript{IM}, enabled flexible, programmable delivery of open- and closed-loop stimulation. Longitudinal assessments confirmed the safety, efficacy, and durability of this therapy and established the foundation for a pivotal regulatory trial.

Although the etiology of orthostatic hypotension in Parkinsonian syndromes differs from that of spinal cord injury, both conditions disrupt descending commands from the brainstem vasomotor centers that modulate sympathetic neurons in the thoracic spinal cord. We hypothesized that individuals with Parkinsonism with spared sympathetic pathways could benefit from spinal cord stimulation. To understand these mechanisms, we developed an experimental framework to induce orthostatic hypotension after neurodegeneration in rodent and non-human primate models. These preclinical findings were translated into a series of compassionate-use clinical cases, where EES mitigated hypotension, improved mobility, and enhanced patient-reported outcomes in individuals with Parkinson's disease and multiple system atrophy. These results support the broader applicability of spinal cord neuromodulation to restore autonomic stability across neurological conditions characterized by baroreflex dysfunction.

Inquiry is a foundational skill for lifelong learning, critical thinking, and democratic participation. Open-ended learning environments (OELEs) offer a powerful way to foster such skills through authentic, self-directed exploration. Yet for these tools to be truly impactful, it is not always sufficient for learners to simply engage with them as they are often complex to navigate; educators and researchers must also be able to make sense of learner behaviour, determine when and how to intervene, and design support that accommodates the varied ways students approach inquiry.

This thesis aims to address key limitations in the modelling of inquiry learning by centring diversity, generalisability, and interpretability as core design principles. It responds to five persistent challenges in the field: the poor scalability of existing models across tasks and domains; the difficulty of modelling in low-data, high-variance contexts; the predominance of post-hoc analyses that preclude timely support; the tendency to overlook learner diversity in behaviour and background; and the limited integration between computational modelling and educational insight. Across these challenges, the thesis develops methods that are theoretically grounded, empirically robust, and practically applicable, advancing both the technical modelling of inquiry and the human contexts in which such models are deployed.

We begin with a generalisable framework for predicting students' conceptual understanding early in the learning process using only their interaction logs. Evaluated across two different user studies, our models achieve high predictive accuracy after just a few actions while maintaining interpretability. We show how inquiry behaviours such as experimental setups relate to different levels of conceptual mastery, enabling real-time support.

We then explore how inquiry strategies transfer across simulations and domains. Using clustering and representation learning, we find that learners tend to maintain consistent inquiry profiles across tasks. However, when strategy shifts do occur, especially when transitioning from simpler to more complex simulations, they are often linked to improved learning outcomes. This insight opens the door to generalisable modelling of inquiry as a transferable cognitive skill.

This thesis also addresses algorithmic biases head-on by investigating how demographic differences, shaped by broader societal structures, can lead to biased algorithmic outcomes. We develop a diagnostic framework to quantify how these behavioural differences affect model performance and show that signal variation, not just data imbalance, can drive unfairness. In response, we propose mitigation strategies based not on demographic attributes, but on behavioural profiles. These approaches improve model fairness without sacrificing accuracy, and generalise to multiple learning paradigms including simulations and educational games.

This thesis places a strong emphasis on real-world applicability and empirical validation. Across 12 user studies involving hundreds of students in Switzerland, the U.S., Canada, and Colombia, we evaluate the technical contributions under conditions of data scarcity and curricular diversity.

By combining algorithmic design with cross-context behavioural analysis and fairness-aware modelling, this work lays the foundation for scalable, and fair timely educational AI systems.

Alzheimer's disease (AD) is a progressive, neurodegenerative disease and the most common form of dementia. Despite more than a hundred years of research, there is no cure for AD. The pathological aggregation of the protein tau to form intraneuronal neurofibrillary tangles (NFTs) and of Aß into extracellular plaques is a major hallmark of AD. NFTs have been classified into pretangles, mature tangles, and ghost tangles based on staining profiles for different amyloid dyes and tau markers detected by immunohistochemistry, which, however, provides no ultrastructural information. Electron microscopy (EM) has long been the gold standard for visualizing ultrastructure and was instrumental in the initial identification of tau paired helical filaments in AD. Correlative light and electron microscopy (CLEM) bridges this gap by preserving both molecular specificity, via immunolabeling, and ultrastructural detail, via EM, in the same specimen. In Chapter 1, I present a room-temperature CLEM workflow enabling precise mapping of disease-relevant protein aggregates within post-mortem human brain, setting the stage for the ultrastructural studies that follow in Chapters 2 and 3. In Chapter 2, I apply this CLEM workflow to characterize the ultrastructure of different levels of NFT maturation. Mature tangles contained highly aligned, clustered twisting tau fibrils that were mostly devoid of membranous material, while pretangles showed no ultrastructural differences to control cells in the same tissue. Multilamellar bodies could be occasionally found in mature tangles but to a much lesser degree than in dystrophic neurites. Dystrophic neurites were swollen and showed a continuum of pathology, which was either predominantly membranous or fibrillar. Ghost tangles were highly compartmentalized, most likely due to astrocytic infiltration. Fibrils were highly aligned and densely packed. Interestingly, they lacked a twist and were thinner than fibrils in mature tangles. Based on the staining profile and the morphology of these ghost tangle fibrils, it seems likely tau fibrils are cleaved and the two protofilament subunits fall apart as NFTs progress from mature tangle to ghost tangle. More recently, the atomic structures of patient-derived tau fibrils have been solved. However, due to the extraction of fibrils from tissue, all spatial context is lost in these studies. Cryo-CLEM coupled with cryo-electron tomography allows for sub-nanometer studies in situ. Chapter 3 presents preliminary results of a cryo-CLEM pipeline adapted to chemically-fixed, post-mortem AD brain. This pipeline allows the visualization of tau fibrils. This opens the door to future studies that link molecular epitope identity, ultrastructural morphology, and atomic-level structure in post-mortem human AD tissue. Together, this thesis provides an adapted technical platform for correlative light and electron microscopy and applies it directly to a central biological question in Alzheimer's disease: By correlating molecular identity with subcellular structure across pathology stages at the nanometer scale, this work contributes to a more spatially and temporally resolved understanding of tau aggregation. The findings underscore the importance of spatial context for interpreting molecular and structural data, and they advocate for a shift toward in situ analysis in human neuropathology and open new avenues for mechanistic and translational studies.

This thesis investigates the fundamental physics and computational modeling of quantum thermal transport in crystalline systems, focusing on regimes where established theoretical approaches face intrinsic limitations. Thermal transport, studied for decades, remains an active research area due to the shrinking size and increasing complexity of materials and devices, making microscopic and mesoscopic heat propagation both a theoretical challenge and a practical necessity.

Central to this work is the study of phonons, quantized vibrational modes representing atomic oscillations in crystal lattices and primary heat carriers in insulating and semiconducting materials. Phonon scattering under temperature gradients directly impacts thermal conductivity. While first-principles simulations and semiclassical methods based on the Boltzmann transport equation and Fermi's golden rule have predictive success, they depend on approximations that fail where quantum effects and strong anharmonicity dominate.

To overcome these limitations, this thesis develops a theoretical and computational framework extending beyond conventional thermal transport models. A key focus is the breakdown of the combined Boltzmann transport equation and Fermi's golden rule in describing phonon lifetimes. By relaxing strict energy conservation in individual phonon-phonon scattering processes through collisional broadening, the approach better captures phonon interactions, especially where computational smearing creates ambiguities in thermal conductivity calculations. The work demonstrates that an accurate, consistent description requires moving beyond these approximations using Green's function theory and the full quantum Kadanoff-Baym equation. This framework inherently includes virtual phonon-mediated scattering, a quantum phenomenon always present in the quantum regime but neglected in semiclassical and current first-principles approaches, marking a significant novel contribution to computational thermal transport.

The thesis further explores heat hydrodynamics, a regime dominated by phonon collisions that conserve crystal quasi-momentum, causing phonons to flow collectively rather than diffusively, similar to real fluids. This hydrodynamic transport produces complex phenomena such as thermal backflow and phonon vortices, especially in spatially confined systems like two-dimensional materials. Analytical tools inspired by fluid dynamics are developed and applied to describe these effects, providing new insights into controlling and directing heat in mesoscale materials.

This research advances the understanding of thermal transport from both microscopic quantum dynamics and macroscopic heat flow perspectives. It enhances computational models' predictive capabilities across diverse physical conditions, offering deeper insights into phonon-mediated heat conduction and suggesting new strategies for designing materials and devices with tailored thermal properties.