Search

Vast amounts of (open) data are increasingly used to make arguments about crisis topics such as climate change and global pandemics. Data visualizations are central to bringing these viewpoints to broader publics. However, visualizations often conceal the many contexts involved in their production, ranging from decisions made in research labs about collecting and sharing data to choices made in editorial rooms about which data stories to tell. In this paper, we examine how data visualizations about climate change and COVID-19 are produced in popular science magazines, using Scientific American, an established English-language popular science magazine, as a case study. To do this, we apply the analytical concept of data journeys (Leonelli, 2020) in a mixed methods study that centers on interviews with Scientific American staff and is supplemented by a visualization analysis of selected charts. In particular, we discuss the affordances of working with open data, the role of collaborative data practices, and how the magazine works to counter misinformation and increase transparency. This work provides an empirical contribution by providing insight into the data (visualization) practices of science communicators and demonstrating how the concept of data journeys can be used as an analytical framework., Comment: 44 pages, 4 figures, 3 boxes

In contrast to objectively measurable aspects (such as accuracy, reading speed, or memorability), the subjective experience of visualizations has only recently gained importance, and we have less experience how to measure it. We explore how subjective experience is affected by chart design using multiple experimental methods. We measure the effects of changes in color, orientation, and source annotation on the perceived readability and trustworthiness of simple bar charts. Three different experimental designs (single image rating, forced choice comparison, and semi-structured interviews) provide similar but different results. We find that these subjective experiences are different from what prior work on objective dimensions would predict. Seemingly inconsequential choices, like orientation, have large effects for some methods, indicating that study design alters decision-making strategies. Next to insights into the effect of chart design, we provide methodological insights, such as a suggested need to carefully isolate individual elements in charts to study subjective experiences., Comment: 19 pages, 5 figures, 2 tables

Charts are used to communicate data visually, but designing an effective chart that a broad set of people can understand is challenging. Usually, we do not know whether a chart's intended message aligns with the message readers perceive. In this mixed-methods study, we investigate how data journalists encode data and how a broad audience engages with, experiences, and understands these data visualizations. We conducted a series of workshops and interviews with school students, university students, job seekers, designers, and senior citizens to collect perceived messages and subjective feedback on a sample of eight real-world charts. We analyzed these messages and compared them to the intended message of the chart producer. Four of the collected messages from consumers were then provided to data journalists (including the ones that created the original charts) as a starting point to re-design the charts accordingly. The results from our work underline the difficulty of complex charts such as stacked bar charts and Sankey diagrams. Consumers are often overwhelmed with the amount of data provided and are easily confused with terms (as text) not well known. Chart producers tend to be faithful with data but are willing to abstract further when asked to transport particular messages visually. There are strong conventions on how to visually encode particular information that might not be to the benefit of many consumers., Comment: 11 pages, 8 figures, 5 tables

Despite the success of Siamese encoder models such as sentence transformers (ST), little is known about the aspects of inputs they pay attention to. A barrier is that their predictions cannot be attributed to individual features, as they compare two inputs rather than processing a single one. This paper derives a local attribution method for Siamese encoders by generalizing the principle of integrated gradients to models with multiple inputs. The solution takes the form of feature-pair attributions, and can be reduced to a token-token matrix for STs. Our method involves the introduction of integrated Jacobians and inherits the advantageous formal properties of integrated gradients: it accounts for the model's full computation graph and is guaranteed to converge to the actual prediction. A pilot study shows that in an ST few token-pairs can often explain large fractions of predictions, and it focuses on nouns and verbs. For accurate predictions, it however needs to attend to the majority of tokens and parts of speech., Comment: Accepted to EMNLP'23

While recently developed NLP explainability methods let us open the black box in various ways (Madsen et al., 2022), a missing ingredient in this endeavor is an interactive tool offering a conversational interface. Such a dialogue system can help users explore datasets and models with explanations in a contextualized manner, e.g. via clarification or follow-up questions, and through a natural language interface. We adapt the conversational explanation framework TalkToModel (Slack et al., 2022) to the NLP domain, add new NLP-specific operations such as free-text rationalization, and illustrate its generalizability on three NLP tasks (dialogue act classification, question answering, hate speech detection). To recognize user queries for explanations, we evaluate fine-tuned and few-shot prompting models and implement a novel Adapter-based approach. We then conduct two user studies on (1) the perceived correctness and helpfulness of the dialogues, and (2) the simulatability, i.e. how objectively helpful dialogical explanations are for humans in figuring out the model's predicted label when it's not shown. We found rationalization and feature attribution were helpful in explaining the model behavior. Moreover, users could more reliably predict the model outcome based on an explanation dialogue rather than one-off explanations., Comment: EMNLP 2023 Findings. Camera-ready version

In this report, we present a versatile and efficient preconditioned Anderson acceleration (PAA) method for fixed-point iterations. The proposed framework offers flexibility in balancing convergence rates (linear, super-linear, or quadratic) and computational costs related to the Jacobian matrix. Our approach recovers various fixed-point iteration techniques, including Picard, Newton, and quasi-Newton iterations. The PAA method can be interpreted as employing Anderson acceleration (AA) as its own preconditioner or as an accelerator for quasi-Newton methods when their convergence is insufficient. Adaptable to a wide range of problems with differing degrees of nonlinearity and complexity, the method achieves improved convergence rates and robustness by incorporating suitable preconditioners. We test multiple preconditioning strategies on various problems and investigate a delayed update strategy for preconditioners to further reduce the computational costs.

In the context of stochastic portfolio theory we introduce a novel class of portfolios which we call linear path-functional portfolios. These are portfolios which are determined by certain transformations of linear functions of a collections of feature maps that are non-anticipative path functionals of an underlying semimartingale. As main example for such feature maps we consider the signature of the (ranked) market weights. We prove that these portfolios are universal in the sense that every continuous, possibly path-dependent, portfolio function of the market weights can be uniformly approximated by signature portfolios. We also show that signature portfolios can approximate the growth-optimal portfolio in several classes of non-Markovian market models arbitrarily well and illustrate numerically that the trained signature portfolios are remarkably close to the theoretical growth-optimal portfolios. Besides these universality features, the main numerical advantage lies in the fact that several optimization tasks like maximizing (expected) logarithmic wealth or mean-variance optimization within the class of linear path-functional portfolios reduce to a convex quadratic optimization problem, thus making it computationally highly tractable. We apply our method also to real market data based on several indices. Our results point towards out-performance on the considered out-of-sample data, also in the presence of transaction costs.

Casual data visualizations play a vital role in communicating data to lay audiences. Despite this, little is known about how data visualization practitioners make design decisions based on their envisioned target audiences using different media channels. We draw on the findings of a semi-structured interview study to explore how data visualization practitioners working in various settings conceptualize and design for lay audiences and how they evaluate their visualization designs. Our findings suggest that practitioners often use broad definitions of their target audience, yet they stress the importance of 'knowing the readers' for their design decisions. At the same time, commonly used evaluation and feedback mechanisms do not allow a deep knowledge of their readers but rely instead on tacit knowledge, simple usage metrics, or testing with colleagues. We conclude by calling for different forms of visualization evaluation that are feasible for practitioners to implement in their daily workflows., Comment: 11 pages, 2 tables

Purpose: Field-to-susceptibility inversion in quantitative susceptibility mapping (QSM) is ill-posed and needs numerical stabilization through either regularization or oversampling by acquiring data at three or more object orientations. Calculation Of Susceptibility through Multiple Orientations Sampling (COSMOS) is an established oversampling approach and regarded as QSM gold standard. It achieves a well-conditioned inverse problem, requiring rotations by 0{\deg}, 60{\deg} and 120{\deg} in the yz-plane. However, this is impractical in vivo, where head rotations are typically restricted to a range of +-25{\deg}. Non-ideal sampling degrades the conditioning with residual streaking artifacts whose mitigation needs further regularization. Moreover, susceptibility anisotropy in white matter is not considered in the COSMOS model, which may introduce additional bias. The current work presents a thorough investigation of these effects in primate brain. Methods: Gradient-recalled echo (GRE) data of an entire fixed chimpanzee brain were acquired at 7 T (350 microns resolution, 10 orientations) including ideal COSMOS sampling and realistic rotations in vivo. Comparisons of the results included ideal COSMOS, in-vivo feasible acquisitions with 3-8 orientations and single-orientation iLSQR QSM. Results: In-vivo feasible and optimal COSMOS yielded high-quality susceptibility maps with increased SNR resulting from averaging multiple acquisitions. COSMOS reconstructions from non-ideal rotations about a single axis required additional L2-regularization to mitigate residual streaking artifacts. Conclusion: In view of unconsidered anisotropy effects, added complexity of the reconstruction, and the general challenge of multi-orientation acquisitions, advantages of sub-optimal COSMOS schemes over regularized single-orientation QSM appear limited in in-vivo settings., Comment: Text: 2593 words (without legends, references and statements) Abstract: 239 words References: 33 Figures: 4 Tables: 1

To each complex reflection group $\Gamma$ one can attach a canonical symplectic singularity $\mathcal{M}_\Gamma$ arXiv:math/9903070. Motivated by the 4D/2D duality arXiv:1312.5344, arXiv:1707.07679, Bonetti, Meneghelli and Rastelli arXiv:1810.03612 conjectured the existence of a supersymmetric vertex operator superalgebra $\mathsf{W}_\Gamma$ whose associated variety is isomorphic to $\mathcal{M}_\Gamma$. We prove this conjecture when the complex reflection group $\Gamma$ is the symmetric group $S_N$ by constructing a sheaf of $\hbar$-adic vertex operator superalgebras on the Hilbert scheme of $N$ points in the plane. For that case, we also show the free-field realisation of $\mathsf{W}_\Gamma$ in terms of $\operatorname{rk}(\Gamma)$ many $\beta\gamma bc$-systems proposed in arXiv:1810.03612, and identify the character of $\mathsf{W}_\Gamma$ as a certain quasimodular form of mixed weight and multiple $q$-zeta value. In physical terms, the vertex operator superalgebra $\mathsf{W}_{S_N}$ constructed in this article corresponds via the 4D/2D duality arXiv:1312.5344 to the four-dimensional $\mathcal{N}=4$ supersymmetric Yang-Mills theory with gauge group $\operatorname{SL}_N$., Comment: 58 pages, LaTeX; improvements to the exposition, added many references, added subsection on multiple $q$-zeta values

We perform the energy minimization of the paired composite fermion (CF) wave functions, proposed by M\"oller and Simon (MS) [PRB 77, 075319 (2008)] and extended by Yutushui and Mross (YM) [PRB 102, 195153 (2020)], where the energy is minimized by varying the CF pairing function, in the case of an approximate model of the Coulomb interaction in the second Landau level for pairing channels $\ell = -1, 3, 1$ which are expected to be in the Pfaffian, anti-Pfaffian and particle-hole symmetric (PH) Pfaffian phases respectively. It is found that the energy of the $\ell = -1$ MS wave function can be reduced substantially below that of the Moore-Read wave function at small system sizes, however, in the $\ell = 3$ case the energy cannot be reduced much below that of the YM trial wavefunction. Nonetheless, both our optimized and unoptimized wavefunctions with $\ell=-1,3$ extrapolate to roughly the same energy per particle in the thermodynamic limit. For the $\ell = 1$ case, the optimization makes no qualitative difference and these PH-Pfaffian wave functions are still energetically unfavourable. The effective CF pairing is analyzed in the resulting wave functions, where the effective pairing for the $\ell = -1, 3$ channels is found to be well approximated by a weak-pairing BCS ansatz and the $\ell = 1$ wave functions show no sign of emergent CF pairing., Comment: 14 pages, 4 figures, 7 tables

Femtosecond laser excitation of materials that exhibit magnetic spin textures promises advanced magnetic control via the generation of ultrafast and non-equilibrium spin dynamics. We explore such possibilities in ferrimagnetic [Fe(0.35 nm)/Gd(0.40 nm)]$_{160}$ multilayers, which host a rich diversity of magnetic textures from stripe domains at low magnetic fields, a dense bubble/skyrmion lattice at intermediate fields, and a single domain state for high magnetic fields. Using femtosecond magneto-optics, we observe distinct coherent spin wave dynamics in response to a weak laser excitation allowing us to unambiguously identify the different magnetic spin textures. Moreover, employing strong laser excitation we show that we achieve versatile control of the coherent spin dynamics via non-equilibrium and ultrafast transformation of magnetic spin textures by both creating and annihilating bubbles/skyrmions. We corroborate our findings by micromagnetic simulations and by Lorentz transmission electron microscopy before and after laser exposure., Comment: 19 article pages, 12 supplementary

Lifted probabilistic inference exploits symmetries in a probabilistic model to allow for tractable probabilistic inference with respect to domain sizes. To apply lifted inference, a lifted representation has to be obtained, and to do so, the so-called colour passing algorithm is the state of the art. The colour passing algorithm, however, is bound to a specific inference algorithm and we found that it ignores commutativity of factors while constructing a lifted representation. We contribute a modified version of the colour passing algorithm that uses logical variables to construct a lifted representation independent of a specific inference algorithm while at the same time exploiting commutativity of factors during an offline-step. Our proposed algorithm efficiently detects more symmetries than the state of the art and thereby drastically increases compression, yielding significantly faster online query times for probabilistic inference when the resulting model is applied., Comment: Extended version of paper accepted to the Proceedings of the 38th AAAI Conference on Artificial Intelligence (AAAI-2024)

In the framework of statistical optics, a Wigner function represents partially coherent radiation. A Gaussian Wigner function, which is an equivalent representation of the more commonly used Gaussian Schell-model cross-spectral density, may be defined in terms of its covariance matrix and centroid. Starting from the relationship between Gaussian Wigner functions and the Gaussian Schell model, we derive coherence properties of the Gaussian Wigner function, including coherence length and degree of coherence. We define a simplified beamline called a matrix-aperture beamline composed of linear transport sections separated by physical apertures. This is an idealized form for a transport beamline in a synchrotron light source or X-ray free electron laser. An envelope model provides a basic foundation for understanding the optics of a given beamline, in a manner analogous to how linear optics are treated in particle beam dynamics, with corresponding definitions of emittance and Twiss parameters. One major challenge to such an envelope model lies in the hard-edge apertures which break the Gaussian condition, raising the question as to the adequacy of a Gaussian model. We present a consistent way to construct a Gaussian approximation of the far-field Wigner function following the hard edge aperture. To this end, we introduce the concept of a Gaussian aperture and analyze its effects on the radiation Wigner function. A software implementation of this model is described as well.

We prove that every right-angled Coxeter group (RACG) is profinitely rigid amongst all Coxeter groups. On the other hand we exhibit RACGs which have infinite profinite genus amongst all finitely generated residually finite groups. We also establish profinite rigidity results for graph products of finite groups. Along the way we prove that the Higman-Thompson groups $V_{n}$ are generated by $4$ involutions, generalising a classical result of Higman for Thompson's group $V$., Comment: 24 pages

Temporal, spatial or spatio-temporal probabilistic models are frequently used for weather forecasting. The D-vine (drawable vine) copula quantile regression (DVQR) is a powerful tool for this application field, as it can automatically select important predictor variables from a large set and is able to model complex nonlinear relationships among them. However, the current DVQR does not always explicitly and economically allow to account for additional covariate effects, e.g. temporal or spatio-temporal information. Consequently, we propose an extension of the current DVQR, where we parametrize the bivariate copulas in the D-vine copula through Kendall's Tau which can be linked to additional covariates. The parametrization of the correlation parameter allows generalized additive models (GAMs) and spline smoothing to detect potentially hidden covariate effects. The new method is called GAM-DVQR, and its performance is illustrated in a case study for the postprocessing of 2m surface temperature forecasts. We investigate a constant as well as a time-dependent Kendall's Tau. The GAM-DVQR models are compared to the benchmark methods Ensemble Model Output Statistics (EMOS), its gradient-boosted extension (EMOS-GB) and basic DVQR. The results indicate that the GAM-DVQR models are able to identify time-dependent correlations as well as relevant predictor variables and significantly outperform the state-of-the-art methods EMOS and EMOS-GB. Furthermore, the introduced parameterization allows using a static training period for GAM-DVQR, yielding a more sustainable model estimation in comparison to DVQR using a sliding training window. Finally, we give an outlook of further applications and extensions of the GAM-DVQR model. To complement this article, our method is accompanied by an R-package called gamvinereg.

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 pre-processor 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., Comment: Submitted and accepted for ICANN 2023 (32nd International Conference on Artificial Neural Networks)

The size of the effect of the difference in two groups with respect to a variable of interest may be estimated by the classical Cohen's $d$. A recently proposed generalized estimator allows conditioning on further independent variables within the framework of a linear regression model. In this note, it is demonstrated how unbiased estimation of the effect size parameter together with a corresponding standard error may be obtained based on the non-central $t$ distribution. The portrayed estimator may be considered as a natural generalization of the unbiased Hedges' $g$. In addition, confidence interval estimation for the unknown parameter is demonstrated by applying the so-called inversion confidence interval principle. The regarded properties collapse to already known ones in case of absence of any additional independent variables. The stated remarks are illustrated with a publicly available data set.

We report the first extragalactic detection of the higher-order millimeter hydrogen recombination lines ($\Delta n>2$). The $\gamma$-, $\epsilon$-, and $\eta$-transitions have been detected toward the millimeter continuum source N105-1A in the star-forming region N105 in the Large Magellanic Cloud (LMC) with the Atacama Large Millimeter/submillimeter Array (ALMA). We use the H40$\alpha$ line, the brightest of the detected recombination lines (H40$\alpha$, H36$\beta$, H50$\beta$, H41$\gamma$, H57$\gamma$, H49$\epsilon$, H53$\eta$, and H54$\eta$), and/or the 3 mm free-free continuum emission to determine the physical parameters of N105-1A (the electron temperature, emission measure, electron density, and size) and study ionized gas kinematics. We compare the physical properties of N105-1A to a large sample of Galactic compact and ultracompact (UC) H II regions and conclude that N105-1A is similar to the most luminous ($L>10^5$ $L_{\odot}$) UC H II regions in the Galaxy. N105-1A is ionized by an O5.5 V star, it is deeply embedded in its natal molecular clump, and likely associated with a (proto)cluster. We incorporate high-resolution molecular line data including CS, SO, SO$_2$, and CH$_3$OH ($\sim$0.12 pc), and HCO$^{+}$ and CO ($\sim$0.087 pc) to explore the molecular environment of N105-1A. Based on the CO data, we find evidence for a cloud-cloud collision that likely triggered star formation in the region. We find no clear outflow signatures, but the presence of filaments and streamers indicates on-going accretion onto the clump hosting the UC H II region. Sulfur chemistry in N105-1A is consistent with the accretion shock model predictions., Comment: 51 pages, 30 figures, 2 tables (including appendices); accepted for publication in The Astrophysical Journal (ApJ)

We report on time-resolved coherent diffraction imaging of gas-phase silver nanoparticles, strongly heated via their plasmon resonance. The x-ray diffraction images reveal a broad range of phenomena for different excitation strengths, from simple melting over strong cavitation to explosive disintegration. Molecular dynamics simulations fully reproduce this behavior and show that the heating induces rather similar trajectories through the phase diagram in all cases, with the very different outcomes being due only to whether and where the stability limit of the metastable superheated liquid is crossed., Comment: 17 pages, 8 figures (including supplemental material)

We report on further observations of the field of the quasar Q1218+0832. Geier et al. 2019 presented the discovery of the quasar resulting from a search for quasars reddened and dimmed by dust in foreground damped Lyman-alpha absorbers (DLAs). The DLA is remarkable by having a very large HI column density close to 10^22 cm^-2 . Its dust extinction curve shows the 2175 AA bump known from the Local Group. It also shows absorption from cold gas exemplified by CI and CO molecules. For this paper, we present narrow-band observations of the field of Q1218+0832 and also use an archival Hubble Space Telescope (HST) image to search for the galaxy counterpart of the DLA. No emission from the DLA galaxy is found in either the narrow-band imaging or in the HST image. In the HST image, we could probe down to an impact parameter of 0.3 arcsec and a 3-sigma detection limit of 26.8 mag per arcsec^2. In the narrow-band image, we probed down to a 0 arcsec impact parameter and detected nothing down to a 3-sigma detection limit of about 3x10-17 erg s^-1 cm^-2 . We did detect a bright Lyman-alpha emitter 59 arcsec south of Q1218+0832 with a flux of 3x10^-16 erg s^-1 cm^-2 . We conclude that the DLA galaxy must be located at a very small impact parameter (<0.3 arcsec, 2.5 kpc) or it is optically dark. Also, the DLA galaxy most likely is part of a galaxy group., Comment: 9 pages, 10 figures. Accepted for publication in A&A. Corrections from the proofs have been added

Linking biomedical entities is an essential aspect in biomedical natural language processing tasks, such as text mining and question answering. However, a difficulty of linking the biomedical entities using current large language models (LLM) trained on a general corpus is that biomedical entities are scarcely distributed in texts and therefore have been rarely seen during training by the LLM. At the same time, those LLMs are not aware of high level semantic connection between different biomedical entities, which are useful in identifying similar concepts in different textual contexts. To cope with aforementioned problems, some recent works focused on injecting knowledge graph information into LLMs. However, former methods either ignore the relational knowledge of the entities or lead to catastrophic forgetting. Therefore, we propose a novel framework to pre-train the powerful generative LLM by a corpus synthesized from a KG. In the evaluations we are unable to confirm the benefit of including synonym, description or relational information.

Permutation matrices play a key role in matching and assignment problems across the fields, especially in computer vision and robotics. However, memory for explicitly representing permutation matrices grows quadratically with the size of the problem, prohibiting large problem instances. In this work, we propose to tackle the curse of dimensionality of large permutation matrices by approximating them using low-rank matrix factorization, followed by a nonlinearity. To this end, we rely on the Kissing number theory to infer the minimal rank required for representing a permutation matrix of a given size, which is significantly smaller than the problem size. This leads to a drastic reduction in computation and memory costs, e.g., up to $3$ orders of magnitude less memory for a problem of size $n=20000$, represented using $8.4\times10^5$ elements in two small matrices instead of using a single huge matrix with $4\times 10^8$ elements. The proposed representation allows for accurate representations of large permutation matrices, which in turn enables handling large problems that would have been infeasible otherwise. We demonstrate the applicability and merits of the proposed approach through a series of experiments on a range of problems that involve predicting permutation matrices, from linear and quadratic assignment to shape matching problems., Comment: 13 pages, 6 figures

Background: Automated segmentation of spinal MR images plays a vital role both scientifically and clinically. However, accurately delineating posterior spine structures presents challenges. Methods: This retrospective study, approved by the ethical committee, involved translating T1w and T2w MR image series into CT images in a total of n=263 pairs of CT/MR series. Landmark-based registration was performed to align image pairs. We compared 2D paired (Pix2Pix, denoising diffusion implicit models (DDIM) image mode, DDIM noise mode) and unpaired (contrastive unpaired translation, SynDiff) image-to-image translation using "peak signal to noise ratio" (PSNR) as quality measure. A publicly available segmentation network segmented the synthesized CT datasets, and Dice scores were evaluated on in-house test sets and the "MRSpineSeg Challenge" volumes. The 2D findings were extended to 3D Pix2Pix and DDIM. Results: 2D paired methods and SynDiff exhibited similar translation performance and Dice scores on paired data. DDIM image mode achieved the highest image quality. SynDiff, Pix2Pix, and DDIM image mode demonstrated similar Dice scores (0.77). For craniocaudal axis rotations, at least two landmarks per vertebra were required for registration. The 3D translation outperformed the 2D approach, resulting in improved Dice scores (0.80) and anatomically accurate segmentations in a higher resolution than the original MR image. Conclusion: Two landmarks per vertebra registration enabled paired image-to-image translation from MR to CT and outperformed all unpaired approaches. The 3D techniques provided anatomically correct segmentations, avoiding underprediction of small structures like the spinous process., Comment: 35 pages, 7 figures, Code and a model weights available https://doi.org/10.5281/zenodo.8221159 and https://doi.org/10.5281/zenodo.8198697

We propose a novel mixed-integer programming (MIP) formulation for generating precise sparse correspondences for highly non-rigid shapes. To this end, we introduce a projected Laplace-Beltrami operator (PLBO) which combines intrinsic and extrinsic geometric information to measure the deformation quality induced by predicted correspondences. We integrate the PLBO, together with an orientation-aware regulariser, into a novel MIP formulation that can be solved to global optimality for many practical problems. In contrast to previous methods, our approach is provably invariant to rigid transformations and global scaling, initialisation-free, has optimality guarantees, and scales to high resolution meshes with (empirically observed) linear time. We show state-of-the-art results for sparse non-rigid matching on several challenging 3D datasets, including data with inconsistent meshing, as well as applications in mesh-to-point-cloud matching., Comment: 14 pages

Zero-cost proxies are nowadays frequently studied and used to search for neural architectures. They show an impressive ability to predict the performance of architectures by making use of their untrained weights. These techniques allow for immense search speed-ups. So far the joint search for well-performing and robust architectures has received much less attention in the field of NAS. Therefore, the main focus of zero-cost proxies is the clean accuracy of architectures, whereas the model robustness should play an evenly important part. In this paper, we analyze the ability of common zero-cost proxies to serve as performance predictors for robustness in the popular NAS-Bench-201 search space. We are interested in the single prediction task for robustness and the joint multi-objective of clean and robust accuracy. We further analyze the feature importance of the proxies and show that predicting the robustness makes the prediction task from existing zero-cost proxies more challenging. As a result, the joint consideration of several proxies becomes necessary to predict a model's robustness while the clean accuracy can be regressed from a single such feature., Comment: Accepted at DAGM GCPR 2023

Mesh adaptivity is a technique to provide detail in numerical solutions without the need to refine the mesh over the whole domain. Mesh adaptivity in isogeometric analysis can be driven by Truncated Hierarchical B-splines (THB-splines) which add degrees of freedom locally based on finer B-spline bases. Labeling of elements for refinement is typically done using residual-based error estimators. In this paper, an adaptive meshing workflow for isogeometric Kirchhoff-Love shell analysis is developed. This framework includes THB-splines, mesh admissibility for combined refinement and coarsening and the Dual-Weighted Residual (DWR) method for computing element-wise error contributions. The DWR can be used in several structural analysis problems, allowing the user to specify a goal quantity of interest which is used to mark elements and refine the mesh. This goal functional can involve, for example, displacements, stresses, eigenfrequencies etc. The proposed framework is evaluated through a set of different benchmark problems, including modal analysis, buckling analysis and non-linear snap-through and bifurcation problems, showing high accuracy of the DWR estimator and efficient allocation of degrees of freedom for advanced shell computations.

Sagittarius B2 (Sgr B2) is a giant molecular cloud complex in the central molecular zone of our Galaxy hosting several sites of high-mass star formation. The two main centers of activity are Sgr B2(M) and Sgr B2(N), which contain 27 and 20 continuum sources, respectively. Our analysis aims to be a comprehensive modeling of each core spectrum, where we take the complex interaction between molecular lines, dust attenuation, and free-free emission arising from HII regions into account. In this work, we determine the dust and, if HII regions are contained, the parameters of the free-free thermal emission of the ionized gas for each core, and derive a self-consistent description of the continuum levels of each core. Using the high sensitivity of ALMA, we characterize the physical and chemical structure of these continuum sources and gain better insight into the star formation process within the cores. We used ALMA to perform an unbiased spectral line survey of all 47 sources in ALMA band 6 with a frequency coverage from 211 GHz to 275 GHz. In order to model the free-free continuum contribution of a specific core, we fit the contained recombination lines (RRLs) to obtain the electron temperatures and the emission measures, where we use an extended XCLASS program to describe RRLs and free-free continuum simultaneously. In contrast to previous analyses, we derived the corresponding parameters here not only for each core, but also for their local surrounding envelope, and determined their physical properties. The distribution of RRLs we found in the core spectra closely fits the distribution of HII regions described in previous analyses. For the cores we determine average dust temperatures of around 236 K (Sgr B2(M)) and 225 K (Sgr B2(N)), while the electronic temperatures are located in a range between 3800 K and 23800 K.

Silicon Carbide (SiC) is an outstanding material, not only for electronic applications, but also for projected functionalities in the realm of photonic quantum technologies, nano-mechanical resonators and photonics on-a-chip. For shaping 3D structures out of SiC wafers, predominantly dry-etching techniques are used. SiC is nearly inert with respect to wet-etching, occasionally photoelectrochemical etching strategies have been applied. Here, we propose an electrochemical etching strategy that solely relies on defining etchable volumina by implantation of p-dopands. Together with the inertness of the n-doped regions, very sharp etching contrasts can be achieved. We present devices as different as monolithic cantilevers, disk-shaped optical resonators and membranes etched out of a single crystal wafer. The high quality of the resulting surfaces can even be enhanced by thermal treatment, with shape-stable devices up to and even beyond 1550{\deg}C. The versatility of our approach paves the way for new functionalities on SiC as high-performance multi-functional wafer platform., Comment: 13 pages

Hydrogen-rich Type II supernovae (SNe II) are the most frequently observed class of core-collapse SNe (CCSNe). However, most studies that analyse large samples of SNe II lack events with absolute peak magnitudes brighter than -18.5 mag at rest-frame optical wavelengths. Thanks to modern surveys, the detected number of such luminous SNe II (LSNe II) is growing. There exist several mechanisms that could produce luminous SNe II. The most popular propose either the presence of a central engine (a magnetar gradually spinning down or a black hole accreting fallback material) or the interaction of supernova ejecta with circumstellar material (CSM) that turns kinetic energy into radiation energy. In this work, we study the light curves and spectral series of a small sample of six LSNe II that show peculiarities in their H$\alpha$ profile, to attempt to understand the underlying powering mechanism. We favour an interaction scenario with CSM that is not dense enough to be optically thick to electron scattering on large scales -- thus, no narrow emission lines are observed. This conclusion is based on the observed light curve (higher luminosity, fast decline, blue colours) and spectral features (lack of persistent narrow lines, broad H$\alpha$ emission, lack of H$\alpha$ absorption, weak or nonexistent metal lines) together with comparison to other luminous events available in the literature. We add to the growing evidence that transients powered by ejecta-CSM interaction do not necessarily display persistent narrow emission lines., Comment: 27 pages, 16 figures, accepted for publication in MNRAS

The Pauli-Poisson equation is a semi-relativistic model for charged spin-1/2-particles in a strong external magnetic field and a self-consistent electric potential computed from the Poisson equation in 3 space dimensions. It is a system of two magnetic Schr\"odinger equations for the two components of the Pauli 2-spinor, representing the two spin states of a fermion, coupled by the additional Stern-Gerlach term representing the interaction of magnetic field and spin. We study the global wellposedness in the energy space and the semiclassical limit of the Pauli-Poisson to the magnetic Vlasov-Poisson equation with Lorentz force and the semiclassical limit of the linear Pauli equation to the magnetic Vlasov equation with Lorentz force. We use Wigner transforms and a density matrix formulation for mixed states, extending the work of P. L. Lions & T. Paul as well as P. Markowich & N.J. Mauser on the semiclassical limit of the non-relativistic Schr\"odinger-Poisson equation.

The optical intersite spin transfer (OISTR) effect was recently verified in Fe$_{50}$Ni$_{50}$ using magneto-optical Kerr measurements in the extreme ultraviolet range. However, one of the main experimental signatures analyzed in this work, namely a magnetic moment increase at a specific energy in Ni, was subsequently found also in pure Ni, where no transfer from one element to another is possible. Hence, it is a much-discussed issue whether OISTR in FeNi alloys is real and whether it can be verified experimentally or not. Here, we present a comparative study of spin transfer in Fe$_{50}$Ni$_{50}$, Fe$_{19}$Ni$_{81}$ and pure Ni. We conclusively show that an increase in the magneto-optical signal is indeed insufficient to verify OISTR. However, we also show how an extended data analysis overcomes this problem and allows to unambiguously identify spin transfer effects. Concomitantly, our work solves the long-standing riddle about the origin of delayed demagnetization behavior of Ni in FeNi alloys.

The magneto-optical Kerr effect (MOKE) in the extreme ultraviolet (EUV) regime has helped to elucidate some of the key processes that lead to the manipulation of magnetism on ultrafast timescales. However, as we show in this paper, the recently introduced spectrally-resolved analysis of such data can lead to surprising experimental observations, which might cause misinterpretations. Therefore, an extended analysis of the EUV magneto-optics is necessary. Via experimental determination of the dielectric tensor, we find here that the non-equilibrium excitation in an ultrafast magnetization experiment can cause a rotation of the off-diagonal element of the dielectric tensor in the complex plane. In direct consequence, the commonly analyzed magneto-optic asymmetry may show time-dependent behaviour that is not directly connected to the magnetic properties of the sample. We showcase such critical observations for the case of ultrafast magnetization dynamics in Ni, and give guidelines for the future analysis of spectrally-resolved magneto-optical data and its comparison with theory.

In quality control, microstructures are investigated rigorously to ensure structural integrity, exclude the presence of critical volume defects, and validate the formation of the target microstructure. For quenched, hierarchically-structured steels, the morphology of the bainitic and martensitic microstructures are of major concern to guarantee the reliability of the material under service conditions. Therefore, industries conduct small sample-size inspections of materials cross-sections through metallographers to validate the needle morphology of such microstructures. We demonstrate round-robin test results revealing that this visual grading is afflicted by pronounced subjectivity despite the thorough training of personnel. Instead, we propose a deep learning image classification approach that distinguishes steels based on their microstructure type and classifies their needle length alluding to the ISO 643 grain size assessment standard. This classification approach facilitates the reliable, objective, and automated classification of hierarchically structured steels. Specifically, an accuracy of 96% and roughly 91% is attained for the distinction of martensite/bainite subtypes and needle length, respectively. This is achieved on an image dataset that contains significant variance and labeling noise as it is acquired over more than ten years from multiple plants, alloys, etchant applications, and light optical microscopes by many metallographers (raters). Interpretability analysis gives insights into the decision-making of these models and allows for estimating their generalization capability.

The focused ion beam (FIB) is a powerful tool for the fabrication, modification and characterization of materials down to the nanoscale. Starting with the gallium FIB, which was originally intended for photomask repair in the semiconductor industry, there are now many different types of FIB that are commercially available. These instruments use a range of ion species and are applied broadly in materials science, physics, chemistry, biology, medicine, and even archaeology. The goal of this roadmap is to provide an overview of FIB instrumentation, theory, techniques and applications. By viewing FIB developments through the lens of the various research communities, we aim to identify future pathways for ion source and instrumentation development as well as emerging applications, and the scope for improved understanding of the complex interplay of ion-solid interactions. We intend to provide a guide for all scientists in the field that identifies common research interests and will support future fruitful interactions connecting tool development, experiment and theory. While a comprehensive overview of the field is sought, it is not possible to cover all research related to FIB technologies in detail. We give examples of specific projects within the broader context, referencing original works and previous review articles throughout., Comment: This publication is based upon work from the COST Action FIT4NANO CA19140, supported by COST (European Cooperation in Science and Technology) https://www.cost.eu/. Financial support from COST Action CA19140 is acknowledged http://www.fit4nano.eu/ Version 3 has many text and language edits as well as layout tuning but no substantial new content

Ion irradiation can cause burrowing of nanoparticles in substrates, strongly depending on the material properties and irradiation parameters. In this study, we demonstrate that the sinking process can be accomplished with ion irradiation of cube-shaped Ag nanoparticles on top of silicon; how ion channeling affects the sinking rate; and underline the importance of the amorphous state of the substrate upon ion irradiation. Based on our experimental findings, the sinking process is described as being driven by capillary forces enabled by ion-induced plastic flow of the substrate., Comment: the manuscript has 25 pages and 6 figures

The continuous advancement of photorealism in rendering is accompanied by a growth in texture data and, consequently, increasing storage and memory demands. To address this issue, we propose a novel neural compression technique specifically designed for material textures. We unlock two more levels of detail, i.e., 16x more texels, using low bitrate compression, with image quality that is better than advanced image compression techniques, such as AVIF and JPEG XL. At the same time, our method allows on-demand, real-time decompression with random access similar to block texture compression on GPUs, enabling compression on disk and memory. The key idea behind our approach is compressing multiple material textures and their mipmap chains together, and using a small neural network, that is optimized for each material, to decompress them. Finally, we use a custom training implementation to achieve practical compression speeds, whose performance surpasses that of general frameworks, like PyTorch, by an order of magnitude., Comment: 22 pages, accepted to ACM SIGGRAPH 2023 Transactions on Graphics