This paper proposes a novel data-driven approach for feedback linearization of nonlinear control-affine systems by leveraging the Koopman operator framework. We establish theoretical connections between feedback linearization on the original state manifold and the higher-dimensional Koopman observable manifold using concepts from system immersion. For systems with exact Koopman bilinear representations, we provide closed-form solutions to the feedback linearization problem without solving partial differential equations. When exact bilinear representations are not available, we develop an approximate method based on singular value decomposition that converges to the exact solution as the observables are enriched. The simulation results and numerical examples demonstrate the effectiveness of the approach.

Define the codewords of the Tensor Reed-Muller code TRM(r1,m1;r2,m2;...;rt, mt) to be the evaluation vectors of all multivariate polynomials in the variables {xij}j=1,…mii=1,…,t with degree at most ri in the variables xi1,xi2,…,ximi. The generator matrix of TRM(r1,m1;...;rt, mt) is thus simply the tensor product of the generator matrices of the Reed-Muller codes RM(r1,m1),...,RM(rt, mt).We show that on the binary symmetric channel, for any constant rate R below capacity, one can construct a tensor Reed-Muller code TRM(r1,m1;... ;rt, mt) of rate R that is decodable in quasilinear time. For any blocklength n, we provide two constructions of such codes•Our first construction (with t = 3) has error probability n−ω(log n) and decoding time O(n log log n).•Our second construction, for any t ≥ 4, has error probability 2−n12−12(t−2)−o(1) and decoding time O(n log n).One of our main tools is a polynomial-time algorithm for decoding an arbitrary tensor code C = C1 ⊗ ... ⊗ Ct from dmin (C)2max{dmin (C1),…,dmin (Ct)}−1 adversarial errors. Crucially, this algorithm does not require the codes C1,...,Ct to themselves be decodable in polynomial time.

An vielen Tankstellen kostet der Liter Benzin aktuell zwei Franken. Doch seinen Tank füllen, kostet in der Schweiz im Vergleich zu vielen anderen Staaten wenig. Auch verglichen mit der Vergangenheit ist Benzin heute viel günstiger, wenn man die Entwicklung der Löhne und die höhere Effizienz der Fahrzeuge betrachtet. Interview durch Matthias Heim

Advances in genetic engineering and metabolic imaging are transforming our ability to probe brain function with unprecedented precision. This chapter explores how the integration of optogenetics, chemogenetics, and genetically encoded biosensors with functional magnetic resonance spectroscopy (fMRS) enables dynamic, cell-type-specific interrogation of brain metabolism in animal models. Building on foundational concepts, such as the astrocyte–neuron lactate shuttle, the malate–aspartate shuttle, and the glutamate–glutamine–GABA cycle, we highlight the limitations of classical frameworks in explaining fast, stimulus-locked metabolic changes captured by modern fMRS techniques. We discuss emerging analysis methods inspired by functional MRI—including glutamate response functions and general linear modeling—that are redefining how fMRS data are interpreted. Finally, we argue that the synergy between genetic control, real-time metabolic monitoring, and computational modeling is ushering in a new era of mechanistic, quantitative neuroenergetics, with profound implications for understanding both normal brain function and metabolic pathology.

Adaptive composition theorems quantify how information leakage accumulates across multiple stages of interaction. In this work, we study the "degenerate"regime of adaptive composition, in which a finite sequence of leakage-limited mechanisms enables full disclosure of the sensitive input. We show that degeneration occurs if and only if full disclosure is possible through a finite collection of feasible binary deterministic maps. Remarkably, this condition is unaffected by adaptivity. We specialize this result to mutual information and maximal leakage, obtaining exact thresholds for degeneration. We further study two-stage adaptive composition for maximal leakage in the non-degenerate regime, derive exact closed-form bounds, and identify a phase transition for ternary inputs.

Electrons in the fractional quantum Hall effect collectively give rise to quasiparticles known as anyons. These excitations carry a fractional charge and obey neither fermionic nor bosonic exchange statistics. Among them, non-Abelian anyons are predicted to encode information non-locally in their braiding history, making them attractive candidates for fault-tolerant topological quantum computation. However, to fully identify these quasiparticles, both their exchange statistics and their fractional charge must be measured. While recent advances in anyonic interference have revealed braiding and phase jumps consistent with fractional exchange statistics, a direct measurement of the fractional charges remains essential to establish their fundamental properties. In GaAs, the historical platform for these studies, fractional charges have been measured through shot noise techniques, but newer platforms remain comparatively unexplored. Bilayer graphene is particularly promising because its electrically tunable band structure and exceptional cleanliness host fractional states inaccessible in GaAs, including even-denominator states that are candidates to host non-Abelian anyons. Nonetheless, poor-quality ohmic contacts have largely limited charge detection in bilayer graphene so far. Quantum Hall antidots, tunable hills in the electrostatic potential landscape, address this gap: they act as controlled impurities around which single quasiparticles localize, and their charge can be extracted from the periodic tunneling events.

In this thesis, we study gate-defined quantum Hall antidots in bilayer graphene. By tuning the antidot into the Coulomb-dominated regime, we extract the charge of the tunneling quasiparticles through conductance oscillations. In the integer quantum Hall regime, we find the expected electron charge and further show that the oscillation regime can be tuned, shifting from a Coulomb-dominated to an Aharonov-Bohm-dominated regime. In the odd-denominator electron-like fractional states, we observe oscillations consistent with the minimal fractional excitation charge, whereas in their hole-conjugate states, we observe a distinct tunneling charge, which we believe depends on the parity of the integer edge modes. Finally, in the even-denominator fractional states, we resolve both the minimal excitation e/4, and its double e/2, revealing a crossover between two regimes that is likely governed by the slope of the antidot potential. These results open the way toward the control and manipulation of individual anyons and establish gate-defined quantum Hall antidots as a simple, versatile, and broadly applicable route to probing fractional charge in van der Waals materials.

Centrioles are evolutionarily conserved cylindrical structures characterized by 9-fold radial symmetry. SAS-6 is a conserved and key molecular player, which forms the basis of the cartwheel, critical for centriole assembly and establishing 9-fold symmetry. While SAS-6 self assembly, ring formation, and stacking are well characterized in vitro, these processes remain insufficently explored in the cellular context. This thesis addresses these questions using two approaches. First, a mechanistic approach where we probed human SAS-6 (HsSAS-6) function by modulation with binding monobodies and monitoring its integration into growing cartwheels in human cells. Second, a descriptive and structural approach where we leveraged the centrioles of Trichonympha agilis, a protist with thousands of centrioles featuring elongated cartwheel bearing proximal regions amenable to structural analysis. Addressing how SAS-6 integrates into the growing cartwheel during centriole duplication, pulse overexpression experiments using GFP tagged HsSAS-6 revealed complex integration patterns, suggesting both proximal and distal incorporation early in the assembly process and more distal incorporation later, though technical limitations prevented definitive conclusions. To further probe function of HsSAS-6 in the cellular context, we investigated modulating HsSAS-6 assembly using monobodies that bind HsSAS-6 and affect centriole duplication. Under-duplicating monobodies NMB5 and NMB571 reduced cartwheel length without altering overall procentriole length, suggesting these molecules modulate HsSAS-6 recruitment. In T. agilis, follow up studies on mass spectrometry candidates for proximal region components aimed to identify potential A-C linker, Pinhead, and cartwheel inner density (CID) components. Microtubule binding assays yielded inconclusive results, whereas co-IP assays for T. agilis SAS-6 (TaSAS-6) interactors suggested two potential binders. Ultra-expansion microscopy unexpectedly revealed asymmetric centrin distribution along T. agilis centriole length and width, with this asymmetry maintained across neighboring centrioles along the cell membrane. Furthermore, in situ cryo-electron tomography leveraging Serial Lift-Out and TEM tomography unveiled structural features never before observed in cryogenic conditions for centrioles, including curved spokes emanating from the cartwheel hub and previously unreported densities outside centrioles. A specific population of these densities was found to be radially asymmetrically distributed around the centriole. Together, this work provides an avenue to advance our understanding of in situ cartwheel architecture and paths to investigate how SAS-6 assembles and stacks into the highly organized structure that templates centriole formation.

Gene expression regulation is a fundamental biological process governing protein abundance and cellular function. Translation, particularly elongation, is a key regulatory step whose dysregulation is implicated in metabolic disorders and cancer. Despite its importance, the sequence, structural, and contextual determinants of elongation dynamics remain incompletely understood. In this thesis, we develop three complementary interpretable machine learning frameworks to systematically elucidate translation elongation. First, we propose Riboclette, a conditional transformer-based model that achieves state-of-the-art performance in predicting ribosome density profiles across multiple amino acid deprivation conditions. Leveraging a novel double-head architecture, Riboclette isolates condition-specific effects. Through attribution and perturbation analyses, we uncover codon- and motif-level drivers of ribosome stalling, including a previously uncharacterized deprivation codon ramp effect. Second, we present RiboGL, a graph-based framework that integrates mRNA sequence and secondary structure by modeling transcripts as graphs. By combining graph neural networks with sequence models, RiboGL captures long-range structural interactions and explains stalling phenomena missed by sequence-only approaches. Finally, we develop BIO-Distiller, a model-agnostic knowledge distillation framework that transfers representations from biological foundation models into efficient and interpretable predictors. Applied to translation-related tasks, it enables motif-level insights while maintaining high predictive performance. Together, these frameworks provide a unified perspective on translation elongation, bridging local sequence features, global structural context, and scalable representation learning. By uncovering mechanistic drivers of ribosome dynamics across conditions and scales, this work advances our understanding of translational regulation and establishes a foundation for identifying regulatory elements, characterizing disease-associated dysregulation, and enabling the design of targeted therapeutic strategies.

Translation elongation is a critical, well-regulated layer in the control of protein abundance and cellular homeostasis. This study focused on the eukaryotic initiation factor 5A (eIF5A), a highly conserved translation elongation factor. Most research on eIF5A has been conducted in yeast, where it was found to promote efficient peptide-bond formation at difficult-to-translate motifs, particularly polyproline stretches. This process requires the unique hypusine modification of eIF5A by the enzymes DHPS and DOHH. eIF5A, DHPS and DOHH are upregulated in several cancers, including hepatocellular carcinoma (HCC). Metabolic HCC develops in a background of chronic inflammation and fibrosis, in which activation of hepatic stellate cells (HSCs) drives the deposition of extracellular matrix (ECM) proteins. ECM proteins, including collagens, are enriched in proline-rich motifs, which have previously been associated with eIF5A-dependent translation elongation. Inhibition of eIF5A hypusination has been shown to reduce collagen expression, although underlying mechanisms remain incompletely understood. In this thesis, the role of eIF5A and its hypusine modification was investigated in human HSC and HCC models using RNA sequencing, ribosome profiling and proteomics. TGF-β1-activated LX-2 cells were used as a fibrogenic model characterised by increased expression of ECM and ECM-associated proteins. Inhibition of hypusination with the DHPS inhibitor GC7 reduced fibrotic proteins, including collagens, despite relatively limited changes in mRNA levels. Additionally, GC7 induced a mitochondrial phenotype characterised by reduced proteins involved in mitochondrial translation and decreased mitochondrial DNA content. Responses to GC7 treatment differed between LX-2 and the HCC cell line SNU-182, supporting context-dependent regulation across HSC- and HCC-derived systems, potentially reflecting differences in TGF-β1 responsiveness. To distinguish pharmacological GC7 effects from genetic perturbation of the hypusine pathway, inducible knockdown systems targeting DHPS and eIF5A were generated. In DHPS and EIF5A knockdown, a partially overlapping ECM downregulation was observed, but it did not fully recapitulate the mitochondrial phenotype induced by GC7 treatment. The divergence between protein and RNA levels across perturbations supported predominantly post-transcriptional regulation of ECM-associated proteins. Tripeptide enrichment and ribosome profiling analyses revealed selective enrichment of motifs previously reported to require eIF5A for efficient translation, suggesting some parallels to canonical yeast models. However, no strong ribosome-stalling signatures were detected in regulated genes. Still, increased dwell times of prolines at the E and P-sites of the ribosome were observed upon eIF5A perturbation. In parallel, many collagen transcripts showed reduced translation efficiency, supporting a model in which impaired elongation feeds back on translation initiation. These findings identify proline-rich ECM-associated translation as particularly sensitive to eIF5A perturbations. Rather than causing strong, yeast-like ribosome stalling, eIF5A perturbation in mammalian cells appears to trigger translational adaptation, with modest local elongation defects but substantial effects on the translation efficiency of collagens and other proline-rich transcripts.

In the field of computational chemistry, the construction of databases of transition metal complexes (TMCs) from experimental crystal structures is often hindered by the intrinsic complexity of their electronic structure. Although crystallographic databases provide a vast collection of structures exhibiting broad chemical diversity, they generally lack certain key pieces of information required for quantum chemical computations, such as the total molecular charge or the ground-state spin. Beyond these challenges related to the curation and interpretation of experimental data, the statistical prediction of TMC properties remains limited, both by the range of machine learning (ML) models explored and by the diversity of coordination environments and electronic characteristics considered.

This thesis addresses these challenges by developing automated workflows that leverage crystallographic data for computational chemistry purposes. We first present cell2mol, a Python software package designed for the interpretation of molecular crystal structures, with particular emphasis on systems containing TMCs. This tool enables the extraction of molecular connectivity, the assignment of metal oxidation states, and the determination of total molecular charges and bond orders. It has thus enabled the curation of a chemically diverse database comprising more than 31,000 mononuclear TMCs. Building upon this dataset, we develop a statistical model capable of predicting the ground-state spin of 3d TMCs directly from crystal structures with high accuracy. An enhanced version, cell2mol 2.0, further improves performance, extends functionality, and introduces a web interface to facilitate accessibility.

Leveraging an expanded and carefully curated dataset generated with cell2mol, this thesis subsequently trains physics-inspired ML models to predict a variety of TMC properties across a broad range of coordination environments, charge states, and spin states. Finally, by constructing a dataset of problematic multiconfigurational 3d TMCs inspired by previous studies, we analyze multireference diagnostics by reassessing commonly used thresholds, with the goal of training geometric deep learning models capable of predicting these diagnostics.

Overall, this thesis establishes an automated and integrated framework that strengthens electronic structure workflows for TMCs derived from crystal structures while accelerating their large-scale exploration and design. More broadly, it addresses a practical limitation of previous data-driven approaches and significantly expands the chemical space of TMCs accessible to computational investigation.