Recent Submissions
Development of data visualization plugins for QHAna
(2025) Larionov, Andrey
Visualization techniques are crucial for enhancing the understanding of complex datasets, especially in fields like quantum computing. QHAna is a tool specifically designed to allow users to explore and experiment with datasets generated by quantum computing and quantum machine learning. Built on the RAMP architecture, QHAna integrates additional capabilities through microservices and plugins. The aim of this thesis is to expand QHAna’s functionality by implementing new visualization methods. To ensure the effectiveness of these additions, an evaluation of various visualization techniques will be conducted, focusing solely on those relevant to quantum computing and quantum machine learning. Following this assessment, selected techniques will be implemented, thereby enriching QHAna’s ability to support the interpretation of generated data, especially data generated by quantum computing and quantum machine learning. The micro frontends and generated visualizations for all implemented plugins will be provided to demonstrate their utility. The implementation of these plugins will be discussed briefly, focusing on implementation challenges and plugin architecture.
Numerical analysis of the cell droplet loading process in cell printing
(2024) Wang, Yankun; Pang, Fagui; Lai, Shushan; Cai, Renye; Lai, Chenxiang; Yu, Zexin; Zhu, Yiwei; Wu, Min; Zhang, Heng; Kong, Chunyu
Cell printing is a promising technology in tissue engineering, with which the complex three-dimensional tissue constructs can be formed by sequentially printing the cells layer by layer. Though some cell printing experiments with commercial inkjet printers show the possibility of this idea, there are some problems, such as cell damage due the mechanical impact during cell direct writing, which include two processes of cell ejection and cell landing. Cell damage observed during the bioprinting process is often simply attributed to interactions between cells and substrate. However, in reality, cell damage can also arise from complex mechanical effects caused by collisions between cell droplets during continuous printing processes. The objective of this research is to numerically simulate the collision effects between continuously printed cell droplets within the bioprinting process, with a particular focus on analyzing the consequent cell droplet deformation and stress distribution. The influence of gravity force was ignored, cell droplet landing was divided into four phases, the first phase is cell droplet free falling at a certain velocity; the second phase is the collision between the descending cell droplet and the pre-existing cell droplets that have been previously printed onto the substrate. This collision results in significant deformation of the cell membranes of both cell droplets in contact; the third phase is the cell droplet hitting a rigid body substrate; the fourth phase is the cell droplet being bounced. We conducted a qualitative analysis of the stress and strain of cell droplets during the cell printing process to evaluate the influence of different parameters on the printing effect. The results indicate that an increase in jet velocity leads to an increase in stress on cell droplets, thereby increasing the probability of cell damage. Adding cell droplet layers on the substrate can effectively reduce the impact force caused by collisions. Smaller droplets are more susceptible to rupture at higher velocities. These findings provide a scientific basis for optimizing cell printing parameters.
Centers of Hecke algebras of complex reflection groups
(2024) Chavli, Eirini; Pfeiffer, Götz
We provide a dual version of the Geck-Rouquier Theorem (Geck and Rouquier in Finite Reductive Groups (Luminy, 1994), Progr. Math., vol. 141, Birkhäuser Boston, Boston, pp. 251–272, 1997) on the center of an Iwahori-Hecke algebra, which also covers the complex case. For the eight complex reflection groups of rank 2, for which the symmetrising trace conjecture is known to be true, we provide a new faithful matrix model for their Hecke algebra H . These models enable concrete calculations inside H . For each of the eight groups, we compute an explicit integral basis of the center of H .
Target dependent greedy sampling for Gaussian kernel PDE collocation
(2024) Vogel, Max-Paul
This work develops a target-dependent greedy sampling strategy for solving PDEs using Gaussian kernel collocation within the reproducing kernel Hilbert space framework, achieving efficient and mesh-free approximation with exponential convergence rates.
On properties and applications of Gaussian subordinated Lévy fields
(2023) Merkle, Robin; Barth, Andrea
We consider Gaussian subordinated Lévy fields (GSLFs) that arise by subordinating Lévy processes with positive transformations of Gaussian random fields on some spatial domain. The resulting random fields are distributionally flexible and have in general discontinuous sample paths. Theoretical investigations of the random fields include pointwise distributions, possible approximations and their covariance function. As an application, a random elliptic PDE is considered, where the constructed random fields occur in the diffusion coefficient. Further, we present various numerical examples to illustrate our theoretical findings.
Integration of a Centralized Network Controller into the OMNeT++/INET network simulator
(2024) Tompert, Matthias
The increasing demand for deterministic network transmission in industrial applications, including autonomous driving, automation, and healthcare necessitates highly reliable, time-sensitive data exchanges. In order to meet these needs, the IEEE 802.1 working group's Time-Sensitive Networking (TSN) standards ensure predictable communication with bounded latency and minimal delay variation on Ethernet networks. The TSN standard IEEE 802.1Qbv defines Time-Aware Shaping (TAS) as a mechanism enabling bridges and end stations to transmit frames of different priorities according to specified time schedules. A current area of research is the development of adaptive scheduling algorithms for TAS that can react to varying requirements and topology changes. These algorithms aim to generate synchronized schedules for all bridges within the network during runtime. Typically, these scheduling algorithms are executed on a Centralized Network Controller (CNC). One common method that the CNC may use to implement schedules on the bridges is through the Network Configuration Protocol (NETCONF).
In this work, we present the design and implementation of an extensible NETCONF server interface for TSN bridges, which are simulated within an OMNeT++/INET simulation environment. The primary focus of our implementation is on the reconfiguration of TAS during runtime. We provide two distinct server backends for handling the NETCONF messages. Moreover, we design an additional dummy CNC module to provide a transparent interface to an external CNC, which is not part of the simulation. In order to facilitate communication between the dummy CNC and the external CNC, we implement an additional external component, designated as the CNC bridge server. Furthermore, we extend the NETCONF server interface with a basic Link Layer Discovery Protocol (LLDP) implementation, thereby enabling the external CNC to construct a global network topology.
To validate our approach, we reconfigure TAS and request LLDP data of a simulated TSN bridge from an external CNC during runtime. Subsequently, we employ the result analysis feature of OMNeT++ to confirm the successful implementation of the changes. Our evaluation confirms the successful implementation of the new schedule at the desired point in time. Furthermore, we can confirm, that the retrieved LLDP data contains the requisite information for the computation of a global network topology.
External tracking for head-mounted displays
(2024) Motzer, Simon
In many applications based on head-mounted displays (HMD), external devices are employed to identify and track specific objects within the surrounding environment. One such application is the tracking of faces within the surrounding environment. One potential benefit of this application is the development of user interfaces (UI) that adapt to the recognized faces. As most HMDs are equipped with a camera, it is feasible to integrate face recognition into the system. The Company Meta imposed restrictions on the functionality of the Quest 3’s camera for external developers, thereby precluding the use of the internal camera for facial recognition. In this work, a system was developed that utilizes a smartphone (Samsung Galaxy S24 Ultra), mounted on the Quest via a 3D-printed holder, to recognize faces within the surrounding environment. Subsequently, the positional data is transformed for the Quest to facilitate the optimal marking of the faces. In regard to data transmission, users may select either an Internet or Bluetooth connection. The results of the system are encouraging, demonstrating the successful recognition and transmission of faces. Nevertheless, the insufficient accuracy in certain instances when identifying faces represents a compelling motivation for the ongoing advancement of the system.
Accelerated 2D visualization using adaptive resolution scaling and temporal reconstruction
(2023) Becher, Michael; Heinemann, Moritz; Marmann, Thomas; Reina, Guido; Weiskopf, Daniel; Ertl, Thomas
Data visualization relies on efficient rendering to allow users to interactively explore and understand their data. However, achieving interactive frame rates is often challenging, especially for high-resolution displays or large datasets. In computer graphics, several methods temporally reconstruct full-resolution images from multiple consecutive lower-resolution frames. Besides providing temporal image stability, they amortize the rendering costs over multiple frames and thus improve the minimum frame rate. We present a method that adopts this idea to accelerate 2D information visualization, without requiring any changes to the rendering itself. By exploiting properties of orthographic projection, our method significantly improves rendering performance while minimizing the loss of image quality during camera manipulation. For static scenes, it quickly converges to the full-resolution image. We discuss the characteristics and different modes of our method concerning rendering performance and image quality and the corresponding trade-offs. To improve ease of use, we provide automatic resolution scaling in our method to adapt to user-defined target frame rate. Finally, we present extensive rendering benchmarks to examine real-world performance for examples of parallel coordinates and scatterplot matrix visualizations, and discuss appropriate application scenarios and contraindications for usage.
Multi-material blind beam hardening correction in near real-time based on non-linearity adjustment of projections
(2023) Alsaffar, Ammar; Sun, Kaicong; Simon, Sven
Beam hardening (BH) is one of the major artifacts that severely reduces the quality of computed tomography (CT) imaging. This BH artifact arises due to the polychromatic nature of the X-ray source and causes cupping and streak artifacts. This work aims to propose a fast and accurate BH correction method that requires no prior knowledge of the materials and corrects first and higher-order BH artifacts. This is achieved by performing a wide sweep of the material based on an experimentally measured look-up table to obtain the closest estimate of the material. Then, the non-linearity effect of the BH is corrected by adding the difference between the estimated monochromatic and the polychromatic simulated projections of the segmented image. The estimated polychromatic projection is accurately derived using the least square estimation (LSE) method by minimizing the difference between the experimental projection and the linear combination of simulated polychromatic projections. As a result, an accurate non-linearity correction term is derived that leads to an accurate BH correction result. The simulated projections in this work are performed using a multi-GPU-accelerated forward projection model which ensures a fast BH correction in near real-time. To evaluate the proposed BH correction method, we have conducted extensive experiments on real-world CT data. It is shown that the proposed method results in images with improved contrast-to-noise ratio (CNR) in comparison to the images corrected from only the scatter artifacts and the BH-corrected images using the state-of-the-art empirical BH correction method.
Eccentric muscle contractions : from single muscle fibre to whole muscle mechanics
(2023) Tomalka, André
Eccentric muscle loading encompasses several unique features compared to other types of contractions. These features include increased force, work, and performance at decreased oxygen consumption, reduced metabolic cost, improved energy efficiency, as well as decreased muscle activity. This review summarises explanatory approaches to long-standing questions in terms of muscular contraction dynamics and molecular and cellular mechanisms underlying eccentric muscle loading. Moreover, this article intends to underscore the functional link between sarcomeric components, emphasising the fundamental role of titin in skeletal muscle. The giant filament titin reveals versatile functions ranging from sarcomere organisation and maintenance, providing passive tension and elasticity, and operates as a mechanosensory and signalling platform. Structurally, titin consists of a viscoelastic spring segment that allows activation-dependent coupling to actin. This titin-actin interaction can explain linear force increases in active lengthening experiments in biological systems. A three-filament model of skeletal muscle force production (mediated by titin) is supposed to overcome significant deviations between experimental observations and predictions by the classic sliding-filament and cross-bridge theories. Taken together, this review intends to contribute to a more detailed understanding of overall muscle behaviour and force generation - from a microscopic sarcomere level to a macroscopic multi-joint muscle level - impacting muscle modelling, the understanding of muscle function, and disease.