• An extended r -adaptive IGA framework is proposed for weak-discontinuous problems. • A Gaussian monitor redistributes geometric resolution using level-set interfaces. • Control-point relocation enhances interface resolution without mesh refinement. • Up to 65.7% error reduction is achieved with negligible computational overhead.

Noisy intermediate-scale quantum computers (NISQ) are computing hardware in their childhood, but they are showing high promise and growing quickly. They are based on so-called qubits, which are the quantum equivalents of bits. Any given qubit state results in a given probability of observing a value of zero or one in the readout process. One of the main concerns for NISQ machines is the inherent noisiness of qubits, i.e., the observable frequencies of zeros and ones do not correspond to the theoretically expected probability, as the qubit states are subject to random disturbances over time and with each additional algorithmic operation applied to them. Models to describe the influence of this noise exist. In this study, we conduct extensive experiments on quantum noise. Based on our data, we show that existing noise models lack important aspects. Specifically, they fail to properly capture the aggregation of noise effects over time (or over an algorithm’s runtime), and they are underdispersed. With underdispersion, we refer to the fact that observable frequencies scatter much more between repeated experiments than what the standard assumptions of the binomial distribution would allow for. Based on these shortcomings, we develop an extended noise model for the probability distribution of observable frequencies as a function of the number of gate operations. The model roots back to a known continuous random walk on the (Bloch) sphere, where the angular diffusion coefficient can be used to characterize the standard noisiness of gate operations. Here, we superimpose a second random walk at the scale of multiple readouts to account for overdispersion. Further, our model has known, explicit components for noise during state preparation and measurement. The interaction of these two random walks predicts theoretical, runtime-dependent bounds for probabilities. Overall, it is a three-parameter distributional model that fits the data much better than the corresponding one-scale model (without overdispersion), and we demonstrate the better fit and the plausibility of the predicted bounds via Bayesian data-model analysis.

Abstract In recent research, in the domain of speech processing, large End-to-End (E2E) systems for Automatic Speech Recognition (ASR) have reported state-of-the-art performance on various benchmarks. These systems intrinsically learn how to handle and remove noise conditions from speech. Previous research has shown, that it is possible to extract the denoising capabilities of these models into a preprocessor network, which can be used as a frontend for downstream ASR models. However, the proposed methods were limited to specific fully convolutional architectures. In this work, we propose a novel method to extract the denoising capabilities, that can be applied to any encoder-decoder architecture. We propose the Cleancoder preprocessor architecture that extracts hidden activations from the Conformer ASR model and feeds them to a decoder to predict denoised spectrograms. We train our preprocessor on the Noisy Speech Database (NSD) to reconstruct denoised spectrograms from noisy inputs. Then, we evaluate our model as a frontend to a pretrained Conformer ASR model as well as a frontend to train smaller Conformer ASR models from scratch. We show that the Cleancoder is able to filter noise from speech and that it improves the total Word Error Rate (WER) of the downstream model in noisy conditions for both applications.

Various computational fluid dynamic simulations in engineering, such as external aerodynamics, only need the silhouette of an input geometry. Often, it is a laborious process that can take up many human hours. In addition, the CAD geometries are too complex and contain intricate features and topological holes. We showcase an effortless way to shrink-wrap triangulated surfaces with the sole intent of topology and surface simplification. Building upon the concepts of mathematical morphology and newer advancements in geometry processing, we present a straightforward and robust algorithm that can guarantee genus-zero surfaces. Our techniques are equally applicable to general polyhedral meshes and well-suited for handling both oriented and unoriented point clouds. We provide examples using unoriented point clouds to demonstrate the versatility of our algorithms. We have designed our algorithms with a wide variety of applications in mind. However, we specifically highlight their capability for aerodynamic simulations, fluid volume extraction, and surface simplification. Additionally, we emphasize the practicality and ease of implementing the proposed algorithms, and we chain additional algorithms to develop variants of our wrap algorithm.

In this work, we propose a novel training scheme to modularize end-to-end systems. Our training scheme aims at altering the flow of information in an end-to-end system to use the kernels of this system for another system that fulfills another task. We apply this scheme to extract the noise reduction capabilities from a noise-robust automatic speech recognition (ASR) system and implement a speech enhancer from it. This enhancer receives spectral representations from unfiltered audio and outputs cleaned spectral representations. Our enhancer can be integrated into an ASR system as front-end, is trainable, and reduces background noise. Our front-end uses a decoder to clean speech based on the hidden activations of the ASR system Jasper. While training, we exclusively adapt the weights in our decoder and the batch normalization in Jasper. The resulting spectral representations show less background noise. Further, areas in the spectral features are not reconstructed if they do not contribute to speech recognition. We demonstrate that our front-end can be combined with a pre-trained ASR system as back-end and supports speech recognition in noisy conditions. Further, we show that training another ASR system with our front-end results in an increased performance of the ASR system in noisy as well as noiseless conditions. The ASR system's performance is especially improved on challenging speech datasets.

Modelling nonlinear phenomena in thin rubber shells calls for stretch-based material models, such as the Ogden model which conveniently utilizes eigenvalues of the deformation tensor. Derivation and implementation of such models have been already made in Finite Element Methods. This is, however, still lacking in shell formulations based on Isogeometric Analysis, where higher-order continuity of the spline basis is employed for improved accuracy. This paper fills this gap by presenting formulations of stretch-based material models for isogeometric Kirchhoff–Love shells. We derive general formulations based on explicit treatment in terms of derivatives of the strain energy density functions with respect to principal stretches for (in)compressible material models where determination of eigenvalues as well as the spectral basis transformations is required. Using several numerical benchmarks, we verify our formulations on invariant-based Neo-Hookean and Mooney–Rivlin models and with a stretch-based Ogden model. In addition, the model is applied to simulate collapsing behaviour of a truncated cone and it is used to simulate tension wrinkling of a thin sheet.

Computational neuroscience uses models to study the brain. The Hodgkin-Huxley (HH) model, and its extensions, is one of the most powerful, biophysically meaningful models currently used. The high experimental value of the (extended) Hodgkin-Huxley (eHH) models comes at the cost of steep computational requirements. Consequently, for larger networks, neuroscientists either opt for simpler models, losing neuro-computational features, or use high-performance computing systems. The eHH models can be efficiently implemented as a dataflow application on a FPGA-based architecture. The state-of-the-art FPGA-based implementations have proven to be time-consuming because of the long-duration synthesis requirements. We have developed flexHH, a flexible hardware library, compatible with a widely used neuron-model description format, implementing five FPGA-accelerated and parameterizable variants of eHH models (standard HH with optional extensions: custom ion-gates, gap junctions, and/or multiple cell compartments). Therefore, flexHH is a crucial step towards high-flexibility and high-performance FPGA-based simulations, eschewing the penalty of re-engineering and re-synthesis, dismissing the need for an engineer. In terms of performance, flexHH achieves a speedup of 1,065x against NEURON, the simulator standard in computational neuroscience, and speedups between 8x-20x against sequential C. Furthermore, flexHH is faster per simulation step compared to other HPC technologies, provides 65% or better performance density (in FLOPS/LUT) compared to related works, and only shows a marginal performance drop in real-time simulations.