Karlsruhe Institute of Technology

Repository KITopen
Not a member yet
    335598 research outputs found

    Molecular dynamics simulation of the oxidation of liquid iron nanoparticle

    No full text
    These videos provide supporting information for PhD thesis of Maliugin Aleksandr. Molecular dynamics trajectories are visualised the oxidation of a liquid iron nanoparticle at 3000 K are also presented. The molecular dynamics trajectory corresponds to a supplementary video, in which regions of immiscible liquids (L1\mathrm{L_1} - green, L2\mathrm{L_2} -purple) are represented as isosurfaces

    Nonlinear Control Design for Neuromorphic Smart Sensing System for Sound Perception

    No full text
    Die technische Sprachverarbeitung ist dem Hörsinn bezüglich der Energieeffizienz und der Verarbeitunggeschwindigkeit unterlegen. Dies liegt daran, dass die Sprachverarbeitung kaskadiert aufgebaut ist, wohingegen beim Hörsinn eine Rückkopplungsschleife zwischen der Hörschnecke und den Gehirn vorliegt. Daher kann letztere adaptiv und schnell auf Veränderungen in der Umgebung reagieren. In der vorliegenden Arbeit wird untersucht inwiefern man den Hörsinn nachahmen kann, indem man einen Regler für biologisch in­s­pi­rie­rte Oszillatoren entwirft und deren nichtlineare Dynamik analysiert. Dafür werden Ergebnisse aus der theoretischen Biologie verwendet, die die Physiologie der Hörschnecke durch ein System mit steuerbaren Andronov-Hopf-Bifurkation modulieren. Solche Systeme können, wie die Hörschnecke, kompressiv und frequenzselektiv auf externe Anregungen reagieren. Für den Entwurf der Regler wird zunächst basierend auf den Hopf-Theorem und den Normalform Theorem ein Benchmark-Oszillatormodell definiert. Mit diesem Modell werden dann die Eigenschaften von verschiedenen Reglern, wie Kopplung zwischen Oszillatoren oder eine steuerbare Totzeit, analysiert. Neben der steuerbaren Andronov-Hopf-Bifurkation, wird bei der Analyse die charakteristische Frequenz der Systeme analysiert. Es wird dabei gezeigt, dass man die charakteristische Frequenz steuern kann, indem man die Regler anpasst. Um die Resultate der Reglerentwürfe mit Hilfe der Benchmark-Oszillatoren zu verifizieren, wird eine technische Realisierung eines biologisch inspirierten Oszillators untersucht. Diese Realisierung besteht aus einem thermisch aktuierten mikro-elektromechanischen System (MEMS), dessen Auslenkung gemesen werden kann. Die Dynamik des MEMS kann daher gesteuern, indem der MEMS mit der gemessene Auslenkung aktuiert wird. Um die Dynamik zu analysieren, wird zunächst ein mathematisches Modell des MEMS hergeleitet, das aus zwei gekoppelten partiellen Differentialgleichungen (PDEs) besteht. Diese PDEs werden approximiert, indem man die Rayleigh-Ritz-Methode und die Galerkin-Methode anwendet. Damit kann man das Modell des MEMS auf ein 3-dimensionales System bestehend aus gewöhnlichen Differentialgleichungen (ODEs) reduzieren. Dieses System aus ODEs wird bezüglich der charakteristischen Frequenz und der Entstehung von steuerbaren Andronov-Hopf Bifurkationen untersucht. Hierbei wird gezeigt, dass die charakteristische Frequenz des MEMS aufgrund dessen Geometrie steuerbar ist, und dass man zwei steuerbare Andronov-Hopf Bifurkationen erzeugen kann, wenn man die Rückkopplung richtig auslegt. Daher kann das MEMS-Modell im Rückkopplungsmodus genutzt werden, um die auf den Benchmark-Oszillator basierenden Regler zu validieren. Es stellt sich heraus, dass die Anzahl der Bifurkationen bei diffusiver Kopplung steuerbare Totzeit vorhergesagt werden können. In Gegensatz dazu, entsteht eine zusätzliche Bifurkation bei zwei injektiv gekoppelten MEMS. Dieser Unterschied wird dadurch hervorgerufen, dass der Bifurkationsparamter des einzelnen MEMS auch durch die diffusive Kopplung und die steuerbare Totzeit in die Umgebung seines kritischen Punktes getrieben wird. Des Weiteren kann die charaketeristische Frequenz durch die drei Regler verstellt werden. Abschließend wird basierend auf der Frequenzverstellbarkeit die in-Sensor-Signalverarbeitung diskutiert. Dabei werden mehrere Benchmark-Oszillatoren benutzt, um die Frequenzantwort eines Signals abzutasten. Es wird gezeigt, dass die Antwort eines Benchmark-Oszillators eindeutig ist, nachdem die Transienten verschwunden sind, sodass man den Frequenzbereich mit diesen Oszillatoren rekonstruieren kann. Die Abtastung des Frequenzbereichs wird dann mit verschieden Algorithmen, wie sequentielles Abtasten und Abtastung basierend auf Reinforcement Learning, diskutiert

    Analytical nuclear second derivatives for frozen-density embedding employing self-consistent field methods

    No full text
    We report the derivation and implementation of analytical nuclear ground-state second derivatives for uncoupled frozen-density embedding (FDEu) within Hartree–Fock and Kohn–Sham density functional theory. Exchange integrals are evaluated using the chain-of-spheres exchange approach, for which exact nuclear second derivatives are derived and implemented, including all grid derivative contributions, to assess the accuracy of new approximate derivatives. The accuracy of the FDEu Hessian is assessed for selected sample systems by comparison with corresponding supermolecule calculations, encompassing both weakly and strongly coupled subsystems to investigate strengths and limitations of the approach. Finally, the vibrational frequencies of a naphthalene dimer embedded in a crystalline naphthalene environment—comprising up to 44 molecules and a total of 792 atoms—are presented, demonstrating the method’s applicability to extended molecular systems and its potential for the conceptual analysis of intermolecular dimer vibrations relevant to charge transport

    Towards FAIR fundamental physics: PUNCH4NFDI approach

    No full text
    Research data management is becoming increasingly important in data-intensive fundamental physics, including astroparticle, particle, nuclear and astrophysics. Growing data volumes and complexity, increasing interdisciplinarity, high costs of data acquisition and processing, and rising expectations for open science and AI-ready digital research outputs are driving these communities toward FAIR (Findable, Accessible, Interoperable, Reusable) data management practices. Key efforts in this direction include reuse and practical harmonisation of semantic artefacts such as metadata schemas, thesauri and controlled vocabularies; making explicit expert knowledge embedded in collaboration-internal data transformations, specialised software tools, and internal data formats (e.g., FITS, ROOT, HDF5); and defining minimal requirements for reusable digital research objects. This talk presents the multi-layer metadata model and the concept and the use-case-based implementations of Digital Research Product, developed within the PUNCH4NFDI (Particles, Universe, NuClei and Hadrons for the German National Research Data Infrastructure, punch4nfdi.de) consortium in the context of a modern FAIR Data landscape and as core elements of the PUNCH4NFDI Science Data Platform

    Absolute Energy Calibration of the Auger Engineering Radio Array

    No full text
    The Pierre Auger Observatory detects ultra-high-energy cosmic rays over a surface area of more than 3000 km2, requiring a precise and stable energy-scale calibration. Traditionally, this calibration relied solely on fluorescence detectors (FD). In recent years, radio detection has become an increasingly valuable technique, offering complementary insights through an independent reconstruction. Previous comparisons revealed that the energy scales determined with the FD and the Auger Engineering Array (AERA) are consistent within systematic uncertainties, with energies reconstructed with AERA being 12% higher. We present how the AERA standard event reconstruction has been adapted to provide electromagnetic and cosmic-ray energies consistent with the energy scale determined with AERA. We also outline how this approach will be extended to inclined air showers in future work

    IFC Schema Extension for Navigation Models of Humanoid Robots in the Built Environment

    No full text

    Semi-explicit discretization schemes for weakly coupled elliptic-parabolic problems

    No full text

    Path-constrained trajectory planning for multi-robot manufacturing systems using null-space descent optimization with reduced Hessian

    No full text
    In a multi-robot manufacturing system, several robotic manipulators can be physically coupled and collaborate on a process along a given path. Due to the high degrees of freedom and the high-dimensional constraints for path tracking and physical coupling, it remains challenging to plan the motion trajectories for the robots toward the maximization of their collective capabilities, such as load capacity and stiffness. The state-of-the-art methods face conflicts between the optimality and computational complexity, as well as between constraint satisfaction and convergence speed. To resolve the conflicts, this paper proposes a single-stage optimization method set in order to solve for the optimal robot placement and the joint motion simultaneously under a given manufacturing path on the work-piece. This method set is centered upon a reduced Hessian optimization method that descends orthogonally to all the path and coupling constraints in a robot-specific null-space of reduced dimension, while also considering the motion limits. The convergence of the reduced Hessian method is proved mathematically. Furthermore, we verify the reduced Hessian method statistically against existing methods in numerical experiments of pulling a spring. To make the algorithm applicable to real-world robot-driven manufacturing processes with paths of arbitrary geometric complexity, the method set is extended with a parallel computing technique that deals with multiple connected path segments. Through real-world multi-segment milling experiments, we validate the concept of capability enhancement for the multi-robot manufacturing system under the usage of the proposed method set

    Robust Distance Estimation with Out-of-distribution Detection in Ophthalmic Surgery

    No full text
    Objective: Micrometer-scale precision is vital for patient safety in ophthalmic surgery. Recent advancements in instrument-integrated optical sensors aim to accurately measure instrument-to-tissue distances. However, the reliability of these measurements is often hindered by segmentation errors caused by artifacts in the signal. Methods: We propose a deep learning framework to identify optical coherence tomography (OCT) M-scans that fall outside the expected distribution. Our approach incorporates adaptive remote center of motion (RCM)-informed retinal modeling along with time series analysis to effectively detect and rectify segmentation errors. This method estimates retinal distances and their associated confidence levels by leveraging retinal models, instrument positions, and validated distance data. Results: Validation tests conducted on ex vivo human eyes reveal that our pipeline achieves an 88.8% accuracy in identifying out-of-distribution (OOD) measurements. Furthermore, distance estimation improved by 89% and 93% when compared to two existing methods, resulting in an overall mean absolute error (MAE) of less than 40 μm across diverse conditions, including scans with blood and obstructions. Conclusion: This research enhances the accuracy of instrument-to-retina distance estimation, thereby contributing to improved patient safety in ophthalmic surgical procedures. Significance: The proposed method has potential applications beyond ophthalmic surgery, offering benefits to a variety of surgical disciplines and sensorequipped instruments

    0

    full texts

    335,598

    metadata records
    Updated in last 30 days.
    Repository KITopen
    Access Repository Dashboard
    Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇