Universität Stuttgart
Permanent URI for this communityhttps://elib.uni-stuttgart.de/handle/11682/1
Browse
Search Results
Item Open Access Physics-informed transformers for electronic quantum states(2025) Sobral, João Augusto; Perle, Michael; Scheurer, Mathias S.Neural-network-based variational quantum states, particularly autoregressive models, are powerful tools for describing complex many-body wave functions. However, their performance depends on the computational basis chosen and they often lack physical interpretability. We propose a modified variational Monte-Carlo framework which leverages prior physical information to construct a complete computational many-body basis containing a reference state that serves as a rough approximation to the true ground state. A Transformer is used to parametrize and autoregressively sample corrections to this reference state, giving rise to a more interpretable and computationally efficient representation of the ground state. We demonstrate this approach in a fermionic model featuring a metal-insulator transition by employing Hartree-Fock and a strong-coupling limit to define physics-informed bases. We also show that the Transformer’s hidden representation captures the natural energetic order of the different basis states. This work paves the way for more efficient and interpretable neural quantum-state representations.Item Open Access Nanoscale mapping of magnetic auto-oscillations with a single spin sensor(2025) Hache, Toni; Anshu, Anshu; Shalomayeva, Tetyana; Richter, Gunther; Stöhr, Rainer; Kern, Klaus; Wrachtrup, Jörg; Singha, AparajitaSpin Hall nano-oscillators convert DC to magnetic auto-oscillations in the microwave regime. Current research on these devices is dedicated to creating next-generation energy-efficient hardware for communication technologies. Despite intensive research on magnetic auto-oscillations within the past decade, the nanoscale mapping of those dynamics remained a challenge. We image the distribution of free-running magnetic auto-oscillations by driving the electron spin resonance transition of a single spin quantum sensor, enabling fast acquisition (100 ms/pixel). With quantitative magnetometry, we experimentally demonstrate for the first time that the auto-oscillation spots are localized at magnetic field minima acting as local potential wells for confining spin-waves. By comparing the magnitudes of the magnetic stray field at these spots, we decipher the different frequencies of the auto-oscillation modes. The insights gained regarding the interaction between auto-oscillation modes and spin-wave potential wells enable advanced engineering of real devices.Item Open Access Comparison of different strategies to include structural mechanics in the optimization process of an axial turbine’s runner blade(2025) Fraas, Stefan; Tismer, Alexander; Riedelbauch, StefanDifferent strategies to include structural mechanical aspects in the design process of hydraulic machines are compared. Therefore, an axial turbine’s runner blade is optimized using evolutionary algorithms. Four different setups with a scalar objective function are investigated. In the first two setups, structural mechanical aspects are added to the optimization process as a constraint, once with a penalty term and once with a modified selection operator. If structural mechanical aspects are considered as a constraint, the risk of a premature convergence increases. For this reason, additionally, two setups including the minimization of the maximum stress as an objective within a scalar objective function are analyzed. Furthermore, a multi-objective optimization with resolution of the Pareto front is performed. The differences in the results regarding fitness between the setups using a scalar objective function are small. However, the best result is found for a setup where the minimization of the stress is added as an objective. This demonstrates the risk of a premature convergence involved with constraint handling strategies. The worst result is found for the multi-objective optimization with resolution of the Pareto front, most likely due to a less directed search.Item Open Access Temperaturbestimmung von Lithium Ionen Zellen mittels künstlicher neuronaler Netze basierend auf Daten der elektrochemischen Impedanzspektroskopie(2025) Ströbel, Marco; Birke, Kai Peter (Prof. Dr.-Ing.)Item Open Access Defined polymer architectures enabled by yttrium-mediated ring-opening polymerization of renewable lactones(2025) Hornberger, Lea-Sophie; Buchmeiser, Michael R. (Prof. Dr.)Although global plastic production exceeds 410 million tons annually, less than 0.7 % currently originates from bio-based sources. Given the finite fossil resources and the low biodegradability of conventional plastics, the development of polymers from renewable feedstocks offers considerable potential. This highlights the largely unexploited opportunities offered by renewable monomers. Ring-opening polymerization (ROP) enables the synthesis of polyesters with precise control over molar mass, polydispersity, and architecture, while reversible-deactivation radical polymerization (RDRP) techniques such as atom transfer radical polymerization (ATRP) provide complementary strategies for post-polymerization modification of functional polyesters. This dissertation employs aminoalkoxy bis(phenolate) yttrium complexes for the controlled ROP of lactones from renewable resources, systematically expanding the accessible monomer scope from small, strained four-membered rings to unstrained macrolactones and functional seven-membered terpene-derived lactones. In the first part, the entropy-driven ROP of the 16-membered macrolactone ω pentadecalactone (PDL) was achieved under controlled conditions, affording high-molar-mass poly(ω-pentadecalactone) (PPDL) with moderate polydispersities. Its aliphatic backbone makes PPDL a promising sustainable analogue to polyolefins. Sequential block copolymerization with the four-membered racemic β-butyrolactone (BBL) yielded semi-crystalline materials that integrate the crystalline domains of both homopolymers, enabling tunable material properties. The second part investigated the effect of substitution pattern and stereochemistry on the polymerization kinetics and mechanism of the seven-membered terpene-based (-)-menthide and (+)-carvomenthide, which differ only in the relative positions of their substituents. Kinetic analysis combined with density functional theory (DFT) calculations revealed that subtle stereoelectronic differences strongly impact activation parameters, propagation rates, and susceptibility to side reactions. In (-)-menthide, the isopropyl group adjacent to the reactive ester moiety introduces steric hindrance and increases the activation, whereas the reduced steric demand near the ester in (+)-carvomenthide enables faster propagation but also promotes side reactions. These findings provide valuable guidelines for the rational design of terpene-based lactones. The third part focused on trans (+)-dihydrocarvide (DHC), a seven-membered lactone bearing a pendant isopropenyl group. ROP of DHC produced amorphous poly(dihydrocarvide) (PDHC) with full retention of the double bond. Block copolymerization with semi-crystalline PPDL or syndiotactic poly(3-hydroxybutyrate) (PHB) introduced crystallinity and phase separation. The pendant double bonds in PDHC were further functionalized via thiol-ene chemistry to generate ATRP macroinitiators, enabling orthogonal grafting-from polymerizations of ethyl acrylate that afforded high-density polyester-based brush architectures. Overall, the combination of yttrium-mediated ROP with orthogonal post-polymerization techniques enables the construction of renewable polyester architectures such as block and graft copolymers that integrate amorphous, semi-crystalline, and functional segments. This modular approach offers a versatile platform to tailor thermal, mechanical, and functional properties, providing new opportunities for advanced biomedical and high-performance materials.Item Open Access The impact of domain models on energy consumption of classical planners(2025) Tekin, SerhatThe increasing integration of artificial intelligence into real-world systems has intensified concerns about the ecological footprint of computational processes. As the capabilities of AI expand and their applications spread into diverse areas of society, questions of efficiency are no longer confined to algorithmic performance alone but extend to the broader impact of computation on energy usage. Within this context classical planning provides a particularly relevant case since it is a core technique in automated planning. Research in this field has traditionally emphasized runtime efficiency, plan quality and algorithmic design while the energetic dimension has remained largely neglected. This thesis examines that omission by shifting the focus from planners to the domain models that constitute their input. Through systematic modifications of syntactic, semantic and solvability related features it demonstrates that modeling decisions can exert a measurable influence on energy consumption. Rather than viewing energy use as an inherent property of planners, the study shows it to emerge from the interaction between algorithmic behavior and representational form. The work introduces a replicable framework that combines controlled domain transformations with fine grained energy measurements, thereby enabling systematic evaluations of energy usage in symbolic AI. The empirical analysis indicates that syntactic variations usually result in only minor fluctuations, whereas modeling inefficiencies can increase energy demand, with operator arity standing out as a recurring factor. The most pronounced effects arise from solvability constraints which, depending on the planner and the domain, can lead to substantial increases in energy usage or in some cases reductions. Taken together the results highlight that domain modeling is not only a matter of syntactic correctness or semantic adequacy but also of energetic efficiency. The contribution of this thesis is twofold. It establishes a framework for investigating the energy implications of domain features and provides empirical evidence that modeling choices shape the energy profile of planning systems. These findings offer a foundation for further research and provide practical guidance for approaching domain modeling with energy consumption in mind.Item Open Access Separation of short-chain fatty acids from primary sludge into a particle-fee permeate by coupling chamber filter-press and cross-flow microfiltration : optimization, semi-continuous operation, and evaluation(2025) Shylaja Prakash, Nikhil; Maurer, Peter; Horn, Harald; Saravia, Florencia; Hille-Reichel, AndreaShort-chain fatty acids (SCFAs) are valuable metabolic intermediates that are produced during dark fermentation of sludge, which, when capitalized on, can be used as chemical precursors for biotechnological applications. However, high concentrations of solids with SCFAs in hydrolyzed sludge can be highly detrimental to downstream recovery processes. This pilot-scale study addresses this limitation and explores the recovery of SCFAs from primary sludge into a particle-free permeate through a combination of chamber filter-press (material: polyester; mesh size: 100 µm) and cross-flow microfiltration (material: α-Al2O3; pore size: 0.2 µm; cross-flow velocity: 3 m∙s-1; pressure = 2.2 bars). Firstly, primary sludge underwent dark fermentation yielding a hydrolyzate with a significant concentration of SCFAs along with total solids (TS) concentration in the range of 20 to 30 g∙L-1. The hydrolyzate was conditioned with hydroxypropyl trimethyl ammonium starch (HPAS), and then dewatered using a filter press, reducing TS by at least 60%, resulting in a filtrate with a suspended solids concentration ranging from 100 to 1300 mg∙L-1. Despite the lower suspended solids concentration, the microfiltration membrane underwent severe fouling due to HPAS’s electrostatic interaction. Two methods were optimized for microfiltration: (1) increased backwashing frequency to sustain a permeate flux of 20 L∙m-2∙h-1 (LMH), and (2) surface charge modification to maintain the flux between 70 and 80 LMH. With backwashing, microfiltration can filter around 900 L∙meff-2 (without chemical cleaning), with the flux between 50 and 60 LMH under semi-continuous operation. Evaluating the particle-free permeate obtained from the treatment chain, around 4 gCSCFAs∙capita-1∙d-1 can be recovered from primary sludge with a purity of 0.85 to 0.97 CSCFAs∙DOC-1.Item Open Access 3D microprinting of structures with lanthanide‐based fluorophores on optical fibers for multiplexed sensing(2025) Aslani, Valese; Baghapour, Shaghayegh; Warren‐Smith, Stephen C.; Zhang, Wenqi; Ebadati, Esmat; Plush, Sally E.; Herkommer, Alois; Toulouse, Andrea; Afshar V., ShahraamFemtosecond direct laser writing (fs‐DLW) has revolutionized the fabrication of micro‐optical elements, yet its potential in multiplexed sensing has remained constrained by material limitations and fluorescence crosstalk. Here, a novel platform that integrates lanthanide‐based fluorophores-specifically europium complexes-into commercial fs‐DLW resists (OrmoComp and IP‐Visio) to directly print nano/microstructures on the tips of optical fibers is reported. This strategy exploits the exceptional photostability, narrow emission lines, and long luminescence lifetimes to overcome spectral overlap and photobleaching commonly seen with organic fluorophores. By enabling spectral, temporal, and spatial multiplexing, this approach allows simultaneous detection of distinct biochemical and physical parameters. Five distinct structures are fabricated: two woodpile structures for temperature and redox sensing, a Fabry‐Pérot cavity for refractive index detection, and disc and annular geometries for spatially selective excitation. The results show that combining sub‐micron 3D microfabrication with lanthanide photophysics significantly enhances sensing fidelity, opening new avenues for compact, multi‐analyte fiber‐based diagnostics in biomedical applications.Item Open Access Enhancing HTN planning with deep reinforcement learning for method selection(2025) Bahrami, SepidehAutomated planning is a central area within Artificial Intelligence (AI), enabling intelligent behavior in domains such as cloud computing, autonomous systems, context-aware activity recognition, and smart environments. Hierarchical Task Network (HTN) planning, which decomposes complex tasks into simpler subtasks using predefined methods, has proven effective in such structured domains. However, its performance is often constrained by static method selection strategies that lack adaptability to varying planning contexts. To address this limitation, this thesis proposes a neuro-symbolic framework that integrates HTN planning with Deep Reinforcement Learning (DRL), combining the strengths of symbolic reasoning and data-driven learning. Among the available DRL algorithms, Deep Q-Learning (DQL) is particularly suitable due to its off-policy nature, batch-efficient learning, and robust generalization across symbolic planning states. These characteristics align well with deterministic and hierarchical planners, enabling offline learning from curated datasets without requiring interactive exploration. The proposed integration introduces a learning-based decision layer that improves adaptability while preserving the reproducibility and determinism of the underlying planner. The effectiveness of this approach is demonstrated through a comprehensive evaluation across planning efficiency, memory consumption, and plan quality. Results highlight the potential of reinforcement learning to enhance classical HTN systems and support intelligent decision-making in complex, structured environments.Item Open Access Zur optimalen Standortplatzierung von leistungselektronischen Kompensationsvorrichtungen : ein Beitrag zum Problem der Spannungsstützung in ausgedehnten Verbundsystemen(2025) Lisin, Wladimir; Scheffknecht, Günter (Prof. Dr. techn.)Behandelt wird ein NP-hartes Zuordnungsproblem. Die Lösung dieses Problems versteht sich als Antwort auf die Frage nach der optimalen Platzierung von leistungselektronischen Kompensationsvorrichtungen - den sog. FACTS. Die Güte einer Platzierungswahl bemisst sich dann am dynamischen Antwortverhalten infolge ausgelöster Netzfehler. Eine adäquate Abwägung zwischen Geräteanzahl und den dazu erforderlichen Aufwendungen überführt die Aufgabe in eine Pareto-optimale Mehrzielsuche. Des Weiteren ist die Frage nach einer optimalen Standortwahl zugleich auch ein fallvariables Problem, da die individuellen Netznutzungsfälle auch jeweils individuelle Probleminstanzen definieren. Zu ermitteln sind schließlich Ort, Art, Anzahl und die Auslegung von Parametern der Dynamikmodelle der dabei platzierten Anlagen. Konstruiert wurde hierzu ein modulares Bestimmungsverfahren. Die Gütebewertung ermittelter Konfigurationen erfolgt mithilfe von Lastflussberechnungen im detaillierten Netzmodell des europäischen Stromverbundsystems. Die ermittelte Lösung ist schließlich ein aus optimal-korrespondierenden Ort-Geräte-Paaren erweiterter Netz-Anlagen-Park, der bei optimal bestimmten Installationsorten, der optimal bestimmten Anzahl jeder verwendeten Geräteart sowie der optimalen, ortsgebundenen Parametrierung ihrer jeweiligen Dynamikmodelle das anfängliche Systemverhalten eines ausgewählten Netzgebiets bzgl. Stabilität und Robustheit in optimaler Weise verbessert.