Recent Submissions
Affine transformations accelerate the training of physics-informed neural networks of a one-dimensional consolidation problem
(2023) Mandl, Luis; Mielke, André; Seyedpour, Seyed Morteza; Ricken, Tim
Physics-informed neural networks (PINNs) leverage data and knowledge about a problem. They provide a nonnumerical pathway to solving partial differential equations by expressing the field solution as an artificial neural network. This approach has been applied successfully to various types of differential equations. A major area of research on PINNs is the application to coupled partial differential equations in particular, and a general breakthrough is still lacking. In coupled equations, the optimization operates in a critical conflict between boundary conditions and the underlying equations, which often requires either many iterations or complex schemes to avoid trivial solutions and to achieve convergence. We provide empirical evidence for the mitigation of bad initial conditioning in PINNs for solving one-dimensional consolidation problems of porous media through the introduction of affine transformations after the classical output layer of artificial neural network architectures, effectively accelerating the training process. These affine physics-informed neural networks (AfPINNs) then produce nontrivial and accurate field solutions even in parameter spaces with diverging orders of magnitude. On average, AfPINNs show the ability to improve the L2relative error by 64.84%after 25,000 epochs for a one-dimensional consolidation problem based on Biot’s theory, and an average improvement by 58.80%with a transfer approach to the theory of porous media.
Lateral heat flux reduction using a lock-in thermography compensation method
(2023) Rittmann, Johannes; Kreutzbruck, Marc
The naturally diffusive heat flow in solids often results in differences in surface temperatures. Active thermography (AT) exploits such differences to gain information on the internal structure, morphology, or geometry of technical components or biological specimens. In contrast to sound or light waves, thermal waves are lossy; consequently, it is difficult to interpret measured 2D temperature fields. Most AT evaluation methods are based on 1D approaches, and measured 3D heat fluxes are frequently not considered, which is why edges, small features, or gradients are often blurred. Herein, we present a method for reducing the local temperature gradients at feature areas and minimizing the induced lateral heat flux in optical lock-in thermography (LT) measurements through spatial- and temporal-structured heating. The vanishing lateral gradients convert the problem into a 1D problem, which can be adequately solved by the LT approach. The proposed compensation method can bypass the blind frequency of LT and make the inspection largely independent of the excitation frequency. Furthermore, the edge sharpness and separability of features are improved, ultimately improving the feature-detection efficiency.
High-resolution nanoscale NMR for arbitrary magnetic fields
(2023) Meinel, Jonas; Kwon, MinSik; Maier, Rouven; Dasari, Durga; Sumiya, Hitoshi; Onoda, Shinobu; Isoya, Junichi; Vorobyov, Vadim; Wrachtrup, Jörg
Nitrogen vacancy (NV) centers are a major platform for the detection of nuclear magnetic resonance (NMR) signals at the nanoscale. To overcome the intrinsic electron spin lifetime limit in spectral resolution, a heterodyne detection approach is widely used. However, application of this technique at high magnetic fields is yet an unsolved problem. Here, we introduce a heterodyne detection method utilizing a series of phase coherent electron nuclear double resonance sensing blocks, thus eliminating the numerous Rabi microwave pulses required in the detection. Our detection protocol can be extended to high magnetic fields, allowing chemical shift resolution in NMR experiments. We demonstrate this principle on a weakly coupled 13 C nuclear spin in the bath surrounding single NV centers, and compare the results to existing heterodyne protocols. Additionally, we identify the combination of NV-spin-initialization infidelity and strong sensor-target-coupling as linewidth-limiting decoherence source, paving the way towards high-field heterodyne NMR protocols with chemical resolution.
Development of super-specific epigenome editing by targeted allele-specific DNA methylation
(2023) Rajaram, Nivethika; Kouroukli, Alexandra G.; Bens, Susanne; Bashtrykov, Pavel; Jeltsch, Albert
Background. Epigenome editing refers to the targeted reprogramming of genomic loci using an EpiEditor which may consist of an sgRNA/dCas9 complex that recruits DNMT3A/3L to the target locus. Methylation of the locus can lead to a modulation of gene expression. Allele-specific DNA methylation (ASM) refers to the targeted methylation delivery only to one allele of a locus. In the context of diseases caused by a dominant mutation, the selective DNA methylation of the mutant allele could be used to repress its expression but retain the functionality of the normal gene.
Results. To set up allele-specific targeted DNA methylation, target regions were selected from hypomethylated CGIs bearing a heterozygous SNP in their promoters in the HEK293 cell line. We aimed at delivering maximum DNA methylation with highest allelic specificity in the targeted regions. Placing SNPs in the PAM or seed regions of the sgRNA, we designed 24 different sgRNAs targeting single alleles in 14 different gene loci. We achieved efficient ASM in multiple cases, such as ISG15, MSH6, GPD1L, MRPL52, PDE8A, NARF, DAP3, and GSPT1, which in best cases led to five to tenfold stronger average DNA methylation at the on-target allele and absolute differences in the DNA methylation gain at on- and off-target alleles of > 50%. In general, loci with the allele discriminatory SNP positioned in the PAM region showed higher success rate of ASM and better specificity. Highest DNA methylation was observed on day 3 after transfection followed by a gradual decline. In selected cases, ASM was stable up to 11 days in HEK293 cells and it led up to a 3.6-fold change in allelic expression ratios.
Conclusions. We successfully delivered ASM at multiple genomic loci with high specificity, efficiency and stability. This form of super-specific epigenome editing could find applications in the treatment of diseases caused by dominant mutations, because it allows silencing of the mutant allele without repression of the expression of the normal allele thereby minimizing potential side-effects of the treatment.
Unobtrusive movement in XR
(2025) Arab, Metin
Both virtual and augmented interfaces are classic depictions of our technical advancement in futuristic movies. For good reasons, because there is an exponential rise of them in our modern day life. These interfaces could be beneficial for fighting the issues of sedentary lifestyles, in environments like the workplace. One way of achieving this is by moving users unobtrusively. This includes making them turn, reach their hand out, lean backward, etc. In this study, I explored, compared and discussed, at which speeds/offsets, different movements in XR (VR and AR), become unobtrusive. For that I made participants read, slowly moving, XR-displays, for which they had to guess the direction of movement. Afterwards, I plotted the results, for every participant. It turns out that VR and AR brought forward similar results with no major differences.
Procedural generation via operator graphs using GPU work graphs
(2025) Ryan, Robert
Procedural generation is an important tool in a multitude of industries like games, film, and computer aided design. Several approaches for procedural content generation exist and are implemented in varying tools, such as digital content creation software or game engines. Synthesizing large 3D-objects can take a relatively large amount of time which results in slower iteration times. Operator graphs provide a common intermediate representation for a variety of procedural generation approaches as well as a scheduling for executing such a procedural generation on GPUs, lowering time spent generating. By representing operator graphs using a declarative domain specific language (DSL) and compiling them into GPU work graphs, procedural generations can both be executed on the GPU and seamlessly be integrated into 3D-applications. Additionally, GPU work graphs offer a direct path to mesh shaders, allowing for immediate rendering of a procedural generation whilst staying within a fixed memory budget in addition to maintaining interactive frame rates and practically response to a change in parameters.
Nodal band-off-diagonal superconductivity in twisted graphene superlattices
(2023) Christos, Maine; Sachdev, Subir; Scheurer, Mathias S.
The superconducting state and mechanism are among the least understood phenomena in twisted graphene systems. Recent tunneling experiments indicate a transition between nodal and gapped pairing with electron filling, which is not naturally understood within current theory. We demonstrate that the coexistence of superconductivity and flavor polarization leads to pairing channels that are guaranteed by symmetry to be entirely band-off-diagonal, with a variety of consequences: most notably, the pairing invariant under all symmetries can have Bogoliubov Fermi surfaces in the superconducting state with protected nodal lines, or may be fully gapped, depending on parameters, and the band-off-diagonal chiral p -wave state exhibits transitions between gapped and nodal regions upon varying the doping. We demonstrate that band-off-diagonal pairing can be the leading state when only phonons are considered, and is also uniquely favored by fluctuations of a time-reversal-symmetric intervalley coherent order motivated by recent experiments. Consequently, band-off-diagonal superconductivity allows for the reconciliation of several key experimental observations in graphene moiré systems.
Dopant-assisted stabilization of negatively charged single nitrogen-vacancy centers in phosphorus-doped diamond at low temperatures
(2023) Geng, Jianpei; Shalomayeva, Tetyana; Gryzlova, Mariia; Mukherjee, Amlan; Santonocito, Santo; Dzhavadzade, Dzhavid; Dasari, Durga Bhaktavatsala Rao; Kato, Hiromitsu; Stöhr, Rainer; Denisenko, Andrej; Mizuochi, Norikazu; Wrachtrup, Jörg
Charge state instabilities have been a bottleneck for the implementation of solid-state spin systems and pose a major challenge to the development of spin-based quantum technologies. Here we investigate the stabilization of negatively charged nitrogen-vacancy (NV - ) centers in phosphorus-doped diamond at liquid helium temperatures. Photoionization of phosphorous donors in conjunction with charge diffusion at the nanoscale enhances NV 0 to NV - conversion and stabilizes the NV - charge state without the need for an additional repump laser. The phosphorus-assisted stabilization is explored and confirmed both with experiments and our theoretical model. Stable photoluminescence-excitation spectra are obtained for NV - centers created during the growth. The fluorescence is continuously recorded under resonant excitation to real-time monitor the charge state and the ionization and recombination rates are extracted from time traces. We find a linear laser power dependence of the recombination rate as opposed to the conventional quadratic dependence, which is attributed to the photo-ionization of phosphorus atoms.
Efficient joint broadband radar and communication for airborne SAR imaging with small platforms
(2024) Johannes, Winfried; Kallfass, Ingmar (Prof. Dr.-Ing.)
Technologie-Startups : Karrierewege von Gründerinnen und Gründern
(2024) Brem, Alexander; Hess, Sophia; Maguire, Darcy
Über welche Bildungsabschlüsse und Berufserfahrung verfügen Startup-Gründerinnen und -Gründer? Welche Unterschiede zeigen sich in deren Karriereverläufen? Wie ausgeprägt ist der Gender Gap im Bereich der Gründungen? Ergebnisse einer aktuellen Studie zu technologieorientierten Startups in Baden-Württemberg.