KLUEDO Dokumente/documents https://kluedo.ub.rptu.de/index/index/ Tue, 11 Aug 2026 11:22:30 +0200 Tue, 11 Aug 2026 11:22:30 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13391 Recent research indicates that the use of Multiple External Representations (MERs) has the potential to support learning, especially in complex scientific areas, such as quantum physics. In particular, the provision of informationally redundant external representations can have advantageous effects on learning outcomes. This is of special relevance for quantum education, where various external representations are available, and their effective use is recognized as crucial to student learning. However, research on the effects of informationally redundant external representations in quantum learning is limited. The present study aims to contribute to the development of effective learning materials by investigating the effects of informationally redundant external representations on students’ learning of quantum physics. Using a between-subjects design, 113 students were randomly assigned to one of four conditions. The control group learnt with a traditional multimedia learning unit on the behavior of a single photon in a Mach-Zehnder interferometer. The three intervention groups received redundant essential information in the Dirac formalism, the Bloch sphere, or both. The use of eye tracking enabled insight into the learning process, depending on the external representations provided. While the results indicate no effect of the study condition on learning outcomes (content knowledge and cognitive load), the analysis of visual behavior reveals decreased learning efficiency with the addition of the Bloch sphere to the multimedia learning unit. The results are discussed based on current insights into learning with MERs. The study emphasizes the need for careful instructional design to balance the associated cognitive load when learning with informationally redundant external representations. Eva Rexigel; Linda Qerimi; Jonas Bley; Sarah Malone; Stefan Küchemann; Jochen Kuhn article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13391 Tue, 11 Aug 2026 11:22:30 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13390 A material’s physical behavior depends on its microstructure. In a fibrous material, the fiber directions and inter-fiber contacts have a strong influence on physical properties such as heat flow or stiffness. The material’s microstructure can be quantified via, e.g., analysis of computed tomography (CT) images. To improve the material’s physical behavior regarding such properties of interest, in-silico simulations provide an efficient and sustainable alternative to purely experimental approaches. In this thesis, I study and develop methods for analysis and modeling of fibrous materials on multiple scales. In particular, I work with multiple materials made from fibers: Sheet molding compounds (SMC), which are commonly used in the automotive industry, a wood fiber insulation material for sustainable buildings, and a nonwoven that is studied for a biomedical application. First, I study the accuracy of estimation methods for fiber directions in CT images as well as registration and image similarity measures for multimodal images. Then, I extend the Altendorf-Jeulin model, an established stochastic geometrical method to simulate interacting fibers, so that one can additionally calculate and increase the inter-fiber contacts. Finally, I fit multiple models to the insulation material and compare their results of physics simulations. Alex Keilmann doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13390 Fri, 07 Aug 2026 13:41:42 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13386 Human-AI teams are increasingly deployed across a range of critical applications, including medical diagnosis, recruitment, and credit lending. This widespread adoption is driven by the promise of a human-AI synergy that leverages the strengths of both AI models and human decision makers, while mitigating their respective weaknesses. For instance, such hybrid systems can combine the efficiency and quality of AI-generated decisions with human experience and domain knowledge. Moreover, they can mitigate both AI- and human-specific limitations, for example, statistical biases in AI models and cognitive biases inherent to human decision-making. Despite the excitement around deploying human-AI teams, the AI counterpart is often optimized independently, disregarding the fact that it needs to operate alongside a human, who has their own strengths, weaknesses, and error patterns. This oversight has led to the development of approaches that are mostly heuristic-based and fail to optimize the human-AI team performance. In the first part of this thesis, our goal is to close these gaps and fully unlock the potential in human-AI teams by shifting the focus from model-centric optimization to team-aware learning. We present methods that optimize the AI model jointly with a triage policy that governs the interplay between the human and the AI counterparts. We show that using these methods, human and AI can complement each other, outperforming either in isolation. Integrating AI models into high-stakes decision-making domains, however, requires caution. It is essential to acknowledge that these models exhibit statistical biases which can propagate through decision-making processes and be manifested in the final outcomes, leading to significant societal consequences. For example, when AI models are deployed in domains such as loan approvals or university admissions, their biases can systematically influence who receives opportunities, thereby altering demographic distributions of loan recipients or the educated population over time. The second part of the thesis focuses on such AI biases from a new perspective and studies the scenarios where AI predictions can be unfair within demographic groups. Within-group unfairness can prevent the best candidates from these groups---the candidates who are more likely to be qualified---from moving forward in the selection process and having a chance of being selected. To mitigate this type of unfairness, we present an efficient post-hoc algorithm that can be applied to any black-box model, ensuring its predictions satisfy within-group fairness constraints. Nastaran Okati doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13386 Thu, 06 Aug 2026 09:11:13 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13385 The advancement of cooperative, connected, and automated mobility (CCAM) is of paramount importance, and this objective can be achieved through the implementation of Intelligent Transportation Systems (ITS). The objective of Intelligent Transportation Systems (ITS) is to establish a safe, efficient, and environmentally friendly transportation system through the integration of diverse traffic-related systems and communication technologies. In the past, vehicle systems, traffic infrastructure, and service providers operated independently. However, the advent of Intelligent Transportation Systems (ITS) has necessitated that these systems interact and exchange data in order to support automated driving and connected mobility. The challenges to ITS interoperability can be broadly categorized as follows: heterogeneity in communication protocols; lack of standardization; security and privacy concerns; data quality issues; and the high costs associated with achieving interoperability. Notwithstanding these challenges, a number of initiatives are currently underway with the objective of standardizing safety-related traffic information (SRTI) in Europe. Connected driving is not only vital for individual and community safety but also aligns with broader initiatives such as the European ”Vision Zero,” which aims to eliminate road deaths by 2050, and the European Green Deal, which seeks to make Europe the first climate-neutral continent. The potential for connected mobility to facilitate more efficient resource usage and reduce the necessity for personal vehicle ownership could result in a notable reduction in \(CO_2\) emissions. This is consistent with the European Union’s comprehensive strategy to decarbonize the transportation sector, which includes a mandate for all new passenger cars and light commercial vehicles to have zero \(CO_2\) emissions by 2035. The realization of the ambitious objectives set forth for Intelligent Transport Systems (ITS) hinges on the collaborative engagement of all stakeholders within the mobility sector. Nevertheless, several factors impede advancement, including the heterogeneity of communication protocols, data formats, and software architectures utilized in ITS. The absence of standardization results in complications with regard to seamless communication and data exchange between disparate systems. Furthermore, concerns pertaining to security and privacy emerge from the heterogeneous mechanisms and standards employed across diverse applications, rendering the implementation of consistent regulatory frameworks and the safeguarding against data breaches and cyberattacks a challenging endeavor. In light of these challenges, the thesis introduces the ITS-PIE (Intelligent Transportation System Protocol Interoperability Evaluation), a multi-step process designed to analyze protocol specifications for interoperability. To facilitate interoperability, the thesis proposes the creation of a catalog of criteria for the extraction of pertinent parameters from specifications, with a particular emphasis on the assurance of data accuracy, format consistency, and security. The process entails the collection of meta-information regarding the specifications, the examination of cited references for the purpose of situating them within a broader context of ITS, the comparison of formal data and protocol specifications, the potential for divergent usage of terminology across specifications, the analysis of data elements, and the evaluation of the architectural frameworks within which these protocols operate. By following this structured approach, stakeholders can gain a more nuanced understanding of the discrepancies between system architectures and work towards harmonizing them. To support this approach, the evaluation of LLMs is conducted to ascertain their potential contribution to the application of ITS-PIE. In conclusion, while the challenges to ITS interoperability are significant, the analysis presented in the thesis underscores the importance of continued efforts to standardize protocols and ensure reliable data exchange. Achieving interoperability is essential for realizing the full potential of ITS, including the goals of zero road deaths, optimized traffic flow, and improved environmental outcomes. Despite the difficulties posed by the diverse and fragmented nature of ITS ecosystems, a solution-specific approach, informed by thorough analysis and evaluation, is necessary to overcome these barriers and advance the future of transportation. As a result, this dissertation presents methods and tools that serve to identify inconsistent assumptions in the interaction of multiple standards. Jonas Vogt doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13385 Wed, 05 Aug 2026 12:00:09 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13381 Spin-torque and spin-Hall oscillators (SHOs) have emerged as promising candidates for building blocks in neuromorphic computing due to their ability to synchronize mutually, a process that can be mediated by propagating spin waves. We demonstrate a SHO that takes advantage of a low-damping magnetic garnet with dominant perpendicular magnetic anisotropy (PMA), namely gallium-substituted yttrium-iron- garnet (Ga:YIG). In-plane magnetized Ga:YIG allows for the operation at a high efficiency level while also enabling resonant spin-wave emission. A nonlinear self-localization of the excitation is avoided by exploiting the positive nonlinear frequency shift, which facilitates a current-controlled frequency of the emitted spin waves. Using micro-focused Brillouin light scattering spectroscopy, we investigate the properties of the local auto-oscillation and its spin-wave emission. Multiple modes are excited and compete internally, with two propagating modes detected up to distances larger than 10 μm. Their frequencies combine to an extended frequency bandwidth of ∼1.6 GHz. The experimentally observed two-mode system and its transition to a single mode at higher currents are reproduced via micromagnetic simulations, which account for the spatial variation of the PMA arising due to the microstructures on Ga:YIG. Our results propose a promising platform for hosting SHOs, interconnected via propagating spin waves, with particular relevance to neuromorphic computing. Moritz Bechberger; David Breitbach; Abbas Koujok; Björn Heinz; Carsten Dubs; Abbass Hamadeh; Philipp Pirro article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13381 Wed, 05 Aug 2026 09:45:41 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13384 Die Arbeit beschäftigt sich mit der Frage, welche Faktoren einen Zusammenschluss von Kirchengemeinden eher gelingen lassen. Dabei sind folgende Teilaspekte wichtig: a) Was sagt die Fusions-Literatur aus dem Profit-Bereich über Erfolgsfaktoren aus? b) Gibt es in der Non-Profit-Literatur und aus dem kirchlichen Bereich entsprechende Hinweise und was sagen diese aus? c) Worin decken und unterscheiden sich diese Erkenntnisse und was davon kann insbesondere für kirchengemeindliche Fusionen ertragreich sein? d) Welche Erfolgsfaktoren können aus der Untersuchung eines Fusionsprozesses zweier Kirchengemeinden herausgearbeitet werden? e) Welche weiterführenden Hypothesen lassen sich aus dem Vergleich der Ergebnisse für eine möglichst erfolgreiche Fusion von Kirchengemeinden formulieren? f) Kann das Ergebnis in eine handhabbare Praxishilfe umgesetzt werden, die den Kirchengemeinden bei ihren Vor-Überlegungen und ggf. auch bei der Planung und Durchführung einer Fusion als Orientierungs- und Reflexionshilfe dienen kann? Auf dem Hintergrund dieser Fragestellungen besteht die Arbeit aus einem Theorieteil, einer empirischen Untersuchung und einem Transfer in die Praxis. Hagen Fried masterthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13384 Wed, 05 Aug 2026 09:26:48 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13379 Im Mittelpunkt der Arbeit stehen Spannungsfelder in der Zusammenarbeit zwischen Haupt- und Ehrenamtlichen in Non-Profit-Organisationen sowie die Frage, welche Gestaltungsansätze sich daraus für die Organisationsentwicklung ableiten lassen. Die Arbeit ist theoretisch-konzeptionell angelegt und stützt sich auf wissenschaftliche Literatur, aktuelle Studien sowie ausgewählte praxisbezogene Fachbeiträge. Im Zentrum der Analyse stehen Fragen der Rollen- und Aufgabenklärung, Kommunikationsstrukturen sowie das Verhältnis von Profession, Freiwilligkeit und Anerkennung. Die Ergebnisse zeigen, dass Spannungen nicht allein auf individuelle Unterschiede zurückzuführen sind, sondern wesentlich durch organisationale Strukturen, Erwartungen, Zuständigkeiten und unterschiedliche Beteiligungslogiken geprägt werden. Zentrale Gestaltungsansätze liegen in der gemeinsamen Klärung von Rollen und Verantwortlichkeiten, transparenten Kommunikations- und Entscheidungswegen, angemessenen Beteiligungsmöglichkeiten, Reflexionsräumen sowie einer strukturell verankerten Anerkennung. Organisationsentwicklung kann Spannungen nicht vollständig auflösen, sie jedoch sichtbar, besprechbar und konstruktiv bearbeitbar machen. Annette Christel Maßem masterthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13379 Mon, 03 Aug 2026 10:49:31 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13378 Electronic safety systems need to guarantee a certain level of dependability. Therefore, their development must consider robust hardware and trustworthy and reliable embedded software such as firmware, drivers and bare metal software. The main purpose of these elements is to interact with the hardware elements such as timers, ADC converters, bus interfaces, peripherals, counters and I/O interfaces. For this reason, their verification also needs to consider these elements, which brings new challenges compared with traditional verification of high-level software, which has mainly been studied in the past 20 years. Additionally, safety-critical systems such as automotive systems are required to ensure high levels of dependability. Thus, the use of formal methods has increased in the industry, especially in the case of hardware. However, the application of formal methods in firmware development lacks a methodology that considers its interaction with hardware elements, its automation and the current verification techniques to be easily adapted. For this reason, this thesis proposes a Compositional Formal Verification of Automotive Firmware Development. This development must comply with standards such as ISO 26262-6 and ISO 21434. Furthermore, it must address the inherent challenges of software verification, including reachability, software weaknesses, and code coverage. The proposed methodology shows how different formal verification techniques can be applied in various development stages to increase the reliability of FW designs of 8, 16, and 32 bits, considering C and Rust-based designs. Furthermore, the use of metamodeling techniques and Model-Driven Engineering (MDE) allows for the automation of the verification process. The results show that the approach is scalable for big designs if we consider modular verification. Most of the properties run in minutes or even seconds, and the results show that complete designs can be verified in hours, which is a significant advantage over simulation methods, where the runtime increases exponentially with the bit width of the function variables. Besides, the HW interaction considers the inclusion of FWHW Co-verification and High-Level Equivalence Checking techniques, which normally rely on constraints that can be reused not only in the FW verification phase, but also in the integration phase. Finally, this thesis provides the main guidelines to be considered during the design process to enable automated verification. These guidelines include recommendations for software verification and new optimization techniques for hardware development, which will be proposed and analyzed during the development of this thesis. Bryan Daniel Olmos Suquillo doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13378 Fri, 31 Jul 2026 12:58:29 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13376 Stateless model checking (SMC) verifies a concurrent program by systematically exploring its state space. To combat the state-space explosion problem, SMC is frequently combined with Dynamic Partial Order Reduction (DPOR), a technique that avoids exploring executions that are deemed equivalent to one another. Still, DPOR’s scalability is limited by the size of the input program. This thesis improves scalability by (i) providing direct support for common coding patterns that would otherwise have to be handled inefficiently, and (ii) combining DPOR with other state-space reduction techniques. Key to our contributions is a DPOR framework that generalizes a state-of-the-art algorithm. In the first part, we extend DPOR to efficiently handle common blocking constructs such as spinloops, which typically lead to the exploration of numerous redundant executions, often exceeding the number of complete program executions. Our approach manages to eliminate the exploration of most of these blocked executions, exploring only those that indicate a liveness violation. Separately, we extend DPOR to support a class of transactional programs where many locations are accessed atomically, a scenario known to be challenging, focusing on programs with mixed-size accesses. In the second part, we explore the combination of DPOR with state- space bounding and symmetry reduction techniques to reduce the number of explored executions. Our first work on bounding combines DPOR with preemption bounding in a way that avoids exploring equivalent executions, a task that had proven to be particularly chal- lenging. Our second work on bounding resolves the remaining issues by employing a different bound definition that is better suited for DPOR. Lastly, our work on symmetry reduction leverages thread-level symmetries, as well as symmetries between equivalent operations performed by different threads, resulting in a significant reduction in the number of executions explored for programs that exhibit such symmetries. Iason Marmanis doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13376 Wed, 29 Jul 2026 15:59:31 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13375 Metabarcoding offers a powerful approach for assessing benthic biodiversity in remote and understudied deep-sea environments. However, the methodological performance of different DNA extraction strategies remains insufficiently evaluated, especially for benthic meiofaunal communities. In this study, we compared two extraction strategies— community DNA (ComDNA) from isolated benthic metazoans and sedimentary environmental DNA (SedDNA) from whole sediment—to assess benthic metazoan diversity in the Clarion-Clipperton Fracture Zone (CCZ), an area targeted for future polymetallic nodule mining. Using the 18S V1–V2 rRNA marker, we analyzed alpha and beta diversity, taxonomic composition, and the number of metazoan operational taxonomic units (OTUs) recovered by each strategy. ComDNA extractions yielded substantially higher benthic metazoan OTU richness (2,145 OTUs) than SedDNA (392 OTUs), with only 1.2% of OTUs shared between them. Community composition also differed significantly, driven by strategy-specific detection biases. To evaluate the effectiveness of SedDNA for biodiversity monitoring, we modeled the sediment volume required to recover OTU richness comparable to ComDNA samples. Depending on sequencing depth and statistical approach, we estimated that the processing of 27–82 mL of sediment is necessary to match ComDNA-derived richness. Our findings underscore the superior taxonomic resolution of ComDNA extractions but also highlight the potential of optimized SedDNA protocols for scalable biodiversity monitoring. We recommend sediment homogenization, increased sample volume, and a higher sequencing depth of at least 100,000 reads per sample for improving SedDNA-based assessments, particularly in heterogeneous deep-sea environments. These results provide critical methodological guidance for the development of standardized, efficient monitoring strategies in the context of deep-sea mining impact assessments. Lukas Damm; Sahar Khodami; Verena Rubel; Massimiliano Molari; Annemiek Vink; Thorsten Stoeck; Pedro Martinez Arbizu article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13375 Wed, 29 Jul 2026 10:06:39 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13374 This study addresses a critical challenge in surface engineering: developing robust nanocomposite layers on high-performance thermoplastics without inducing macroscopic thermal degradation. While cold gas dynamic spraying (CGDS) of polymers is often described in the literature as a deposition process based on purely mechanical anchoring of particles into polymer surface, our work establishes a multi-scale, hybrid physical–chemical adhesion framework. Using a coupled 3D thermoplasticity finite element model with a Mie–Grüneisen equation of state and Johnson–Cook criteria, we evaluate the supersonic impact dynamics (𝑉0=1000 m/s) of single-walled (5,0) CNTs impacting a PEEK substrate at oblique angles (0–20°). The core scientific lies in bridging continuum mechanics with quantum-chemical statistics. By applying Weibull weakest-link theory to a 37-bond monomer model, we demonstrate that compliant 𝐶𝑎𝑟𝑜𝑚−𝑂 ether bonds selectively absorb impact energy, covering 8.37% of their dissociation barrier. This non-uniform energy sharing yields a 0.23% monomer activation probability, generating a high free-radical density of ~1100 μm−2 beneath the particle plume. This localized “chemical nano-welding” network provides exceptional chemical adhesion, while the remaining 99.77% of intact chains ensure structural rigidity, reinforced by a mechanical “self-locking” field (residual compressive stresses up to 0.9 GPa). This study provides a scientific foundation for designing functional coatings tailored for engineering, aerospace, and biomedical applications. Yankovskyi Illia; Antonyuk Sergiy article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13374 Tue, 28 Jul 2026 16:01:14 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13372 Furan occurs as heat-induced contaminant in a variety of thermally treated foods and was categorized as hepatotoxic and possibly carcinogenic to humans (IARC 2B). The formation of furan in food is frequently accompanied by co-occurrence of various furan analogues that include, for example, 2-methylfuran, 3-methylfuran, 2,5-dimethylfuran, 2-ethylfuran or 2-pentylfuran. In contrast to furan, far less is known regarding the metabolism or potential toxicity of these alkylfurans. However, due to their structural similarity and frequent co- occurrence, alkylfurans may contribute significantly to the overall dietary furan exposure and to an increased risk for consumers. Analogue to furan, cytochrome-P450 (CYP)-catalyzed metabolic activation to highly reactive cis-enedial intermediates is expected to play a central role for the potential toxicity of alkylfurans. In the context of this project, we investigated the metabolic activation of 2-methylfuran in metabolically competent cells. It was demonstrated that 2-methylfuran is metabolically activated in vitro to the respective reactive intermediate 3acetylacrolein by quantification of corresponding Nα-acetyl-L-lysine adducts in cell supernatants. In addition, CYP2E1 was identified as the main contributor to the metabolic activation of 2-methylfuran. Although, 3-acetylacrolein has shown to be capable of reacting with DNA bases in silico, corresponding DNA adducts could not be detected in in vitro experiments, indicating that formed 3-acetylacrolein reacts primarily with cytosolic proteins. Within the scope of these in vitro experiments, no direct genotoxicity of 2-methylfuran could be confirmed. In the case of 2,5-dimethylfuran, experiments with human liver microsomes confirmed a previous hypothesized side-chain hydroxylation of 2,5-dimethylfuran to 5-methyl- 2-furfuryl alcohol for the first time. However, a comparable biotransformation of 2-methylfuran and 3-methylfuran to the corresponding primary alcohols could not be observed. Subsequently, CYP2E1 was identified as the CYP-isoform that catalyzes the hydroxylation of 2,5-dimethylfuran. Besides the CYP-mediated hydroxylation it could be shown that 2,5-dimethylfuran can be metabolized to the reactive cis-enedial intermediate cis-3-hexene-2,5-dione by CYP2E1 but also CYP3A4 and CYP2D6, respectively. By parallel quantification of 5-methyl-2-furfuryl alcohol and cis-3-hexene-2,5-dione after scavenging with glutathione it was possible to deduce the respective formation kinetics in human liver microsomes and CYP-isoenzyme systems. The kinetic data indicated that the metabolic activation of 2,5-dimethylfuran to the reactive metabolite cis-3-hexene-2,5-dione represents the main route of biotransformation, whereas the parallel side-chain hydroxylation is less pronounced. Elke Richling report https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13372 Mon, 27 Jul 2026 11:35:33 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13371 Tensor and fusion categories arise throughout modern algebra and mathe- matical physics. They appear as representation categories of groups, Lie alge- bras and Hopf algebras, as well as in the theory of subfactors and even through topological phases of matter. In particular, fusion categories arising from sub- factors are of special interest due to their close connection to conformal field theories and their construction. This thesis develops computational methods for determining the Drinfeld center of a fusion category and for explicitly computing its structural data. These methods are applied to the Haagerup subfactor, the smallest known exotic subfactor for which no conformal field theory has been found. The unitary 𝐹-symbols of its center are computed numerically and non-unitary 𝐹- symbols symbolically. This yields complete associator data for this example. Furthermore, the centers of all multiplicity-free fusion categories of rank at most five are determined. A further algorithmic contribution concerns the construction of module categories over fusion categories. Module categories are realized explicitly via algebra objects, and an algorithm for computing modules over separable algebras is provided. This construction also produces new fusion categories in the form of categories of bimodules for which explicit 𝐹-symbols can be computed. Finally, the algorithms presented are implemented in the open-source soft- ware package TensorCategories.jl, which made available as part of this thesis. The package is based on the open-source project Oscar.jl which is developed in the scope of the SFB-TRR 195. This implementation allows for the practical application of the developed methods to a wide range of examples of fusion categories, as well as their centers and module categories. Fabian Mäurer doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13371 Mon, 27 Jul 2026 09:04:38 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/9590 The use of fungicides in agriculture can reduce animal biodiversity. Such non-target effects could be mitigated through fungus-resistant cultivars that reduce the need for fungicide applications. Hence, we conducted a study in 32 commercial vineyards in southwest Germany. We investigated the effect of fungicide spraying intensities (susceptible vs resistant cultivars) and management (conventional vs organic) on ground beetles as a widely used indicator group in agroecological studies. In addition, we examined the effects of local habitat conditions (e.g., microclimate, vegetation) and landscape characteristics. Fungicide applications were reduced by half in conventional, and almost by two-thirds in organic vineyards with fungus-resistant cultivars, compared to susceptible grape varieties in the same management regime. While there was no evidence that fungus-resistant cultivars positively affect the diversity of ground beetles, organic management doubled the number of individuals of conservation-relevant species per vineyard. Additionally, the proportion of semi-natural habitats in the surrounding landscape, and the vegetation height significantly affected the species composition across vineyards. We conclude that local habitat conditions and landscape characteristics influence ground beetles more than the fungicide spraying intensity. Additional measures such as organic management, moderate ground vegetation management, and landscape diversification are needed for the conservation of ground beetles in viticulture. Julia T. Gieser; Sebastian Kolb; Jo Marie Reiff; Kai Riess; Maura Hunke; Martin H. Entling; Jens Schirmel article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/9590 Tue, 21 Jul 2026 15:29:58 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13367 Automatic differentiation (AD) is an established tool in scientific computing for the efficient retrieval of accurate derivatives. The reverse mode of AD, also known as backpropagation in machine learning, enables, for example, sensitivity analysis, parameter identification, and discrete adjoint procedures for simulation codes from all application domains. To handle large-scale problems, these codes are typically run on high performance computing clusters and written according to parallel paradigms such as MPI or OpenMP, which require dedicated handling in terms of AD. We present OpDiLib, an add-on for operator overloading AD tools that enables the reverse mode differentiation of OpenMP parallel codes. It relies on OpenMP's tools interface (OMPT) to detect and augment OpenMP constructs in a fully automatic fashion. OpDiLib deduces a corresponding reverse parallelism and embeds it into the tape of the underlying operator overloading AD tool. For the sake of generality and applicability, OpDiLib addresses the inherent data races of shared-memory parallel reverse mode AD by defaulting to atomic updates on adjoint variables. To improve the performance, this can be lifted according to user knowledge in a fine-grained manner. Our benchmarks showcase the achievable speedup and scalability and demonstrate the impact of optimizations such as the elimination of atomic updates on adjoint variables, both in OpenMP parallel and in MPI-OpenMP hybrid parallel applications. Johannes Blühdorn; Nicolas R. Gauger conferenceobject https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13367 Tue, 21 Jul 2026 14:53:49 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/9588 The parametric excitation of spin waves by coherent surface acoustic waves in metallic magnetic thin film structures is demonstrated experimentally using Brillouin light scattering spectroscopy. Complementary micromagnetic simulations and analytical modeling reveal that, depending on the experimental conditions, the spin-wave instabilities originate from different phonon-magnon and magnon-magnon scattering processes. This opens novel ways to create micro-scaled nonlinear magnonic systems for logic and data processing that profit from the efficient excitation of phonons using piezoelectricity. Moritz Geilen; Roman Verba; Alexandre Abbass Hamadeh; Alexandra Nicoloiu; Daniele Narducci; Adrian Dinescu; Milan Ender; Morteza Mohseni; Florin Ciubotaru; Mathias Weiler; Alexandru Müller; Burkard Hillebrands; Christoph Adelmann; Philipp Pirro article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/9588 Tue, 21 Jul 2026 13:30:56 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/9589 Bioelectrochemical systems (BESs) offer a sustainable method for chemical production, including the enhanced production of succinic acid. By combining fermentation with BES, it could be possible to achieve sustainable succinic acid production and CO2 fixation using Actinobacillus succinogenes. In literature, the potential application of BES is commonly associated with increased succinate yields, as it is expected to enhance the availability of NADH, thereby influencing the intracellular nicotinamide adenine dinucleotide (NADH/NAD+) balance. However, it remains unclear whether BES can improve NADH regeneration and achieve higher NADH/NAD+ ratios across all growth phases of A. succinogenes. This study investigates the impact of an applied electrical potential on the intracellular NADH/NAD+ ratio during an electrochemical-assisted fermentation process. Using an adapted high-performance liquid chromatography method with a Supelcosil LC-18-T column, it was demonstrated that NADH availability in BES, particularly during the stationary growth phase, improved by up to 1.98-fold compared to the control. This enhancement in reducing power led to a succinate yield of 0.747 ± 0.01 g g−1, representing a 15.65% increase compared to a fermentation without electrochemical assistance. These findings support the expectation that the use of BES could enhance the competitiveness of bio-based succinate production. Jan-Niklas Hengsbach; Marcel Cwienczek; Wolfgang Laudensack; Judith Stiefelmaier; Nils Tippkötter; Roland Ulber article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/9589 Tue, 21 Jul 2026 13:25:37 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13366 Speckle-interferometrische Verfahren wie die Holografie und die Shearografie ermöglichen die flächenhafte und hochsensitive Erfassung von Oberflächenverformungen und werden in der zerstörungsfreien Prüfung zur Detektion charakteristischer Defektmechanismen eingesetzt. Die Defektdetektion erfolgt dabei indirekt über die mechanische Reaktion eines Bauteils auf eine definierte Belastung. Aus der hohen Sensitivität der Verfahren sowie der Vielzahl an Einflussgrößen bei der Messkonfiguration ergibt sich eine starke Abhängigkeit der Messergebnisse von der gewählten Parametrierung und der anschließenden Auswertung. Trotz ihres Potentials ist der industrielle Einsatz speckle-interferometrischer Verfahren bislang auf spezialisierte Anwendungen beschränkt. Als wesentliche Ursache wird die fehlende systematische und reproduzierbare Parametereinstellung sowie die komplexe Interpretation der hochdimensionalen interferometrischen Messdaten angesehen. Die vorliegende Arbeit untersucht diese Problemstellung durch eine Analyse der zugrunde liegenden physikalischen Zusammenhänge sowie durch die Betrachtung datenbasierter Auswertungsverfahren. Zunächst wird das Verfahren in seine Prozessschritte zerlegt und hinsichtlich der Abhängigkeit von Erfahrungswissen analysiert. Darauf aufbauend wird eine Simulationskette formuliert, die die Abbildung von Bauteil, Belastung und resultierendem Messsignal beschreibt und zur systematischen Parametrierung genutzt wird. Für die Untersuchung datenbasierter Auswertungsverfahren wird mit dem Datensatz SADD eine Datengrundlage geschaffen, die die charakteristischen Eigenschaften speckle-interferometrischer Messdaten abbildet. Der Datensatz umfasst definierte Prüfkörper, kontrollierte Belastungssituationen sowie eine strukturierte Annotation der resultierenden Signalstrukturen. Auf dieser Grundlage werden verschiedene Ansätze zur Merkmalsextraktion und Defektdetektion untersucht, darunter überwachte und unüberwachte Verfahren sowie Zero-Shot-Ansätze. Die Modelle werden hinsichtlich ihrer Fähigkeit bewertet, charakteristische Signalstrukturen zu erkennen und unterschiedliche Defekttypen zu differenzieren. Die Untersuchungen zeigen, dass sowohl modellbasierte Ansätze zur Parametrierung als auch datengetriebene Verfahren zur Auswertung geeignet sind, zentrale Teilaspekte der bisher erfahrungsbasierten Vorgehensweise zu adressieren. Die Parametrierung wird dabei auf eine physikalisch begründete Beschreibung der Zusammenhänge zwischen Bauteil, Belastung und Messsignal zurückgeführt, während die Auswertung durch datengetriebene Methoden hinsichtlich der Erkennung charakteristischer Signalstrukturen untersucht wird. Insgesamt wird ein Ansatz aufgezeigt, mit dem sich die Abhängigkeit von domänenspezifischem Erfahrungswissen in einzelnen Prozessschritten reduzieren lässt und der damit eine Grundlage für eine weitergehende Automatisierung speckle-interferometrischer Messverfahren bildet. Jessica Plassmann doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13366 Tue, 21 Jul 2026 08:11:49 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13343 Ever more powerful artificial intelligence (AI) methods have found their way into our everyday lives over the past decade, from Face ID to ChatGPT. Whereas a high test accuracy is often sufficient for the deployment of AI in convenience applications, high-risk applications require more than just good performance. When deploying AI in high-risk domains like healthcare, every single failure case or data loss can have considerable consequences. This is further emphasised by regulations like the General Data Protection Regulation (GDPR) and the upcoming European AI Act. Trustworthy AI (TAI) describes AI systems that are characterised by various properties, but particularly explainability, privacy, and ethical behaviour. This thesis proposes EPiX, a general-purpose framework built on three pillars of TAI, providing trustworthiness even in complex domains like dermatology. The explainability pillar of EPiX consists of 7 eXplainable AI (XAI) components that are built with a strong focus on user-centricity and diversity. The components are specifically tailored to the users' mental models by providing explanations related to user-defined and hitherto unknown concepts. Moreover, they share the ability to highlight complex and distributed medical concepts while improving the interpretability through the diversity of their presentation. The privacy pillar consists of 5 components. The first two components allow for privacy-preserving training, introducing a novel privacy method and proving its advantages through benchmarking. The remaining 3 components investigate the interplay between privacy, accuracy, and explainability, revealing valuable insights for practitioners. The ethical pillar provides a first basis for ethical evaluation of the framework. The holistic Z-Inspection® co-design process is applied to evaluate the development of EPiX in an interdisciplinary team of experts, focusing on ethical and societal implications of the system. With its set of diverse explanation, novel privacy, and holistic ethics components, EPiX is a framework for trustworthy AI with an enormous value to all stakeholders in complex automated decision-making. exAID and SkinDoc are presented as two practical applications of EPiX, ensuring good accessibility of the framework through user-friendly, web-based, and mobile interfaces. Not only does EPiX help conform to regulations such as GDPR or the AI Act, but it also empowers users to learn from the AI system, identify its weaknesses in real-world use, and overcome them. Adriano Lucieri doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13343 Mon, 20 Jul 2026 14:20:37 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13342 In recent years, the throughput and flexibility requirements of communication systems have grown dramatically to support increased data traffic and versatile use cases. In this context, Neural Network (NN)-based algorithms have emerged as a promising approach to satisfy the stringent requirements of future communication systems. In particular, state-of-the-art research has shown remarkable results for various applications such as demapping and equalization, where the adaptability and nonlinearity of NNs can provide advantages as compared to classical solutions. While related work typically focuses on the algorithmic aspects of NN-based communications, it is crucial to also consider the implementation complexity of the proposed methods for practical deployment. Specifically, the hardware design of sophisticated NNs on resource-constrained edge devices like Field-Programmable Gate Arrays (FPGAs) imposes major challenges, ranging from huge design spaces and cross-layer dependencies to high computational complexity combined with limited hardware resources. In contrast to NNs, classical signal processing algorithms have been optimized for decades for efficient implementation. Thus, for fair comparison, it is crucial to evaluate and compare the implementation aspects of NN-based communication systems to classical ones. Therefore, in this work, we bridge the gap between algorithm and implementation by providing efficient FPGA hardware architectures of NNs for communications and edge intelligence. In particular, we analyze and evaluate the implementation aspects of NN-based demapping and equalization. Further, we consider broader application scenarios such as arrhythmia and anomaly detection. In this context, we perform extensive design space explorations, analyze interdependencies between the design layers, provide efficient FPGA architectures, and benchmark our solutions against classical counterparts and conventional hardware platforms. As a result, we demonstrate the adaptability of NN-based demapping on FPGA and develop a hybrid approach combining our solution with a classical one. Moreover, we present a high-performance NN-based equalizer achieving a data rate of more than 40 GBd and show how NN training can be incorporated for high-throughput equalization on FPGA. Additionally, we propose hardware-aware Neural Architecture Search (NAS) for arrhythmia detection, and show how anomaly detection can be performed at the edge. To summarize, we believe that this work provides a crucial step towards the practical deployment of NN-based communication algorithms on resource-constrained devices, while also contributing to the broader field of NN processing at the edge. Jonas Ney doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13342 Mon, 20 Jul 2026 10:16:24 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13345 This thesis explores massively parallel methods for algebraic computations, focusing on parallel algorithms for computing syzygies and free resolutions of finitely generated modules over polynomial rings, relying on ideas from Schreyer's method.Further topics are modular methods for free resolutions and higher syzygy modules over polynomial rings, with applications to computing rational parametrizations of rational normal curves, as well as plane (singular) rational curves. Massively parallel computations are a transformative approach in the field of computer algebra, enabling the handling of complex and large-scale algebraic problems, with applications on the geometric side, that were previously infeasible. In this thesis, we focus on Schreyer's resolution algorithm and demonstrate how this algorithm can be restructured to accommodate parallel computations by modeling the algorithm in the form of a Petri net. This facilitates the coordination of tasks and data structures within a parallel computing environment and allows us to exploit the inherent parallelism of the algorithm. In order to formulate the algorithm as a Petri net, we in particular restructure it by translating the recursive algorithm for computing the tail terms of syzygies into an iterative form. Moreover, we develop a parallel approach for summing up the syzygy terms. For our implementations, we utilize the SINGULAR/GPI-SPACE framework, which combines the computer algebra system SINGULAR [DGPS24] with the workflow management system GPI-SPACE [Fra25], which natively supports an enriched variant of Petri nets as a coordination language for parallel computations. We observe that in our setting with a large number of small computations, executing SINGULAR interpreter code in the computational backend becomes a serious bottleneck. We thus extend the capabilities of the SINGULAR/GPI-SPACE framework to run kernel level computations based on the C++ library version of SINGULAR, and implement our computational building blocks in this way. We expect that this technique will have numerous further applications in the future. A key challenge in algebraic computations over the rational numbers, such as those relying on Buchberger’s algorithm for computing Gröbner bases, is the issue of intermediate coefficient growth [Arn03]. This phenomenon can significantly affect the efficiency and feasibility of computations. Modular and p-adic methods have proven to be successful in addressing intermediate coefficient growth in various applications, see, for example, [Arn03],[IPS11], [BDFP15]. Building on generic massively parallel modular methods, we present a novel approach to parametrizing rational normal curves using massively parallel modular methods. We implement a modular technique for determining free resolutions as well as specific higher syzygy matrices to avoid the computational cost associated with lifting all syzygy matrices. For example, the last syzygy matrix can be used to compute certain Ext modules, which in turn allow one to iteratively find a parametrization of a rational normal curve. This, in particular, illustrates how classical algebraic geometry techniques can be transferred to a parallel context. Santosh Gnawali doctoralthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13345 Mon, 20 Jul 2026 09:49:32 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13341 Amtliche Bekanntmachung der RPTU Nr.7/17.07.2026 periodicalpart https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13341 Fri, 17 Jul 2026 12:11:41 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13340 Background/Objectives. Inflammation, comprising many complex and finely coordinated immunological processes, represents a vital protective mechanism of the human body. By regulating inflammatory processes, cytokines play a key role in the modulation of the immune system. Secondary plant compounds such as polyphenols influence cellular immunological processes which might contribute to ensuring a physiologically healthy immune status. Methods/Results. This study investigated eleven polyphenol-rich extracts from red fruit beverages in terms of potential inhibitory effects on pro-inflammatory cytokine secretion of leukemic monocyte THP-1 cells. Extracts originating from fruit juice (apple), fruit juice concentrate (red grape, black currant, pomegranate, elderberry, aronia), fruit juice puree (cranberry, blueberry) or fruit juice pulp (strawberry, sour cherry) were obtained by adsorption onto Amberlite® XAD-7 resin. The Folin–Ciocalteu assay showed a high content of phenolic compounds in the eleven extracts and HPLC-DADESI- MSn analysis revealed that the extracts contained various anthocyanins in addition to copigments and polymers. Further screening using Lumit® Immunoassay showed that all tested extracts caused a reduction in pro-inflammatory cytokine secretion (interleukins (IL): IL-1β, IL-6, IL-8 and tumor necrosis factor TNF-α). The extracts from red grape and black currant were the most active ones. Conclusions. Overall, our results showed that polyphenol-rich fruit extracts can inhibit inflammatory processes in vitro. In vivo studies on the anti-inflammatory effect of fruit juice will be a promising approach to determine the fruit juice-dependent, health-promoting effects in humans. Lea S. Edrich; Mats Kiene; Leona Heinke; Christian Tesch; Simone Stegmüller; Peter Winterhalter; Elke Richling article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13340 Fri, 17 Jul 2026 12:00:17 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13339 The activation of CO\(_2\) at isolated transition-metal centers represents a prototypical problem for understanding elementary steps relevant to single-atom catalysis. Fundamental knowledge of such systems can be acquired by investigating gas phase reactions between transition-metal ions and molecules. Because open-shell transition-metal species often possess multiple accessible spin states, their reactions with CO\(_2\) can proceed along competing spin-changing and spin-conserving pathways. Understanding how spin-orbit coupling influences these competing pathways therefore calls for a direct comparison between differential experiments and multi-state dynamical simulations. In this Perspective, we summarize our combined experimental and theoretical investigations of the Ta\(^+\)+, Nb\(^+\), and Zr\(^+\) + CO\(_2\) reactions. Crossed-beam velocity map imaging provides energy- and angle-resolved differential cross sections, while trajectory surface-hopping simulations on first-principles based full-dimensional multi-spin potential energy surfaces enable a dynamical treatment of intersystem crossing and spin-conserving channels on an equal footing. In all three systems, intersystem crossing competes with the spin-conserving channel for the control of the overall reaction dynamics and kinetics. These reactions all proceed predominantly via an indirect mechanism, as evidenced by the nearly isotropic differential cross sections consistent with long-lived complex formation, and the respective product energy distributions indicate substantial energy deposition into internal modes. Despite their similar potential-energy topographies, Ta\(^+\) and Nb\(^+\) + CO\(_2\) reactions are dominated by spin-changing pathways in all energies investigated, whereas in the Zr\(^+\) system the spin-conserving channel becomes competitive. This difference arises from the markedly different magnitudes of the spin-orbit coupling, which determines the efficiency of intersystem crossing and thereby the balance between the two pathways. Marcel Meta; Yang Liu; Hua Guo; Jennifer Meyer article https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13339 Fri, 17 Jul 2026 09:15:40 +0200 https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13285 Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by differences in social communication and interaction, as well as restricted and repetitive behaviors, interests, or activities. ASD is frequently associated with abnormalities in language production and comprehension. Consequently, research aimed at identifying the neural mechanisms underlying ASD has often focused on the language system. Given that language processing is typically characterized by leftward hemispheric specialization, it has been proposed that ASD may be associated with atypical patterns of hemispheric asymmetry. Despite extensive research investigating functional hemispheric lateralization for language in ASD, no quantitative synthesis has comprehensively examined this phenomenon across neuroimaging, electrophysiological, and behavioral evidence in ASD.Consequently, the extent to which hemispheric specialization differs from that of neurotypical (NT) individuals remains unclear. A systematic review and meta-analysis of 32 studies, including 35 independent cohorts with 629 individuals with ASD and 673 NT controls, integrating findings from multiple modalities (i.e., hemodynamic imaging, electrophysiology, and behavioral methods) can help address these inconsistencies and clarify whether atypical brain lateralization for language is a consistent feature of ASD or varies according to measurement technique, cognitive domain, or participant characteristics. The meta-analysis revealed a significant reduction in hemispheric lateralization in individuals with ASD compared to NT controls. Overall, these findings suggest that altered hemispheric specialization may represent an important neurobiological characteristic of ASD and highlight the relevance of considering language lateralization in future research on autism and communication functioning. Silvia Angelica Puertas Cespedes masterthesis https://kluedo.ub.rptu.de/frontdoor/index/index/docId/13285 Thu, 16 Jul 2026 07:07:55 +0200