Search

Since high data volume and complex data formats delivered in modern high-end production environments go beyond the scope of classical process control systems, more advanced tools involving machine learning are required to reliably recognize failure patterns. However, currently, such systems lack a general setup and are only available as application-specific solutions. We propose a process control framework entitled Health Factor for Process Control (HFPC) to bridge the gap between conventional statistical tools and novel machine learning (ML) algorithms. HFPC comprises two main concepts: (a) pattern type to account for qualitative characteristics (error patterns) and (b) intensity to quantify the level of a deviation. While the system retains large model generality, allowing a broad scope of potential application areas, we demonstrate its favorable mathematical properties in a theoretical analysis. In a case study from the semiconductor industry, we underline that (a) our framework is of practical relevance and goes beyond conventional process control, and (b) achieves high-quality experimental results. We conclude that our work contributes to the integration of ML in real-world process control and paves the way to automated decision support in manufacturing.

Microstructure reconstruction is a key enabler of process-structure-property linkages, a central topic in materials engineering. Revisiting classical optimization-based reconstruction techniques,they are recognized as a powerful framework to reconstruct random heterogeneous media, especially due to their generality and controllability. The stochasticity of the available approaches is, however, identified as a performance bottleneck. In this work, reconstruction is approached as a differentiable optimization problem, where the error of a generic prescribed descriptor is minimized under consideration of its derivative. As an exemplary descriptor, a suitable differentiable version of spatial correlations is formulated, along with a multigrid scheme to ensure scalability. The applicability of differentiable optimization realized through this descriptor is demonstrated using a wide variety of heterogeneous media, achieving exact statistical equivalence with errors as low as 0 in a short time. We conclude that, while still in an early stage of development, this approach has the potential to significantly alleviate the computational effort currently associated with reconstructing general random heterogeneous media., Comment: pre-print

The large amount of third-party packages available in fast-moving software ecosystems, such as Node.js/npm, enables attackers to compromise applications by pushing malicious updates to their package dependencies. Studying the npm repository, we observed that many packages in the npm repository that are used in Node.js applications perform only simple computations and do not need access to filesystem or network APIs. This offers the opportunity to enforce least-privilege design per package, protecting applications and package dependencies from malicious updates. We propose a lightweight permission system that protects Node.js applications by enforcing package permissions at runtime. We discuss the design space of solutions and show that our system makes a large number of packages much harder to be exploited, almost for free., Comment: 13 pages

The microvascular networks in the body of vertebrates consist of the smallest vessels such as arterioles, capillaries, and venules. The flow of RBCs through these networks ensures the gas exchange in as well as the transport of nutrients to the tissues. Any alterations in this blood flow may have severe implications on the health state. Since the vessels in these networks obey dimensions similar to the diameter of RBCs, dynamic effects on the cellular scale play a key role. The steady progression in the numerical modeling of RBCs, even in complex networks, has led to novel findings in the field of hemodynamics, especially concerning the impact and the dynamics of lingering events, when a cell meets a branch of the network. However, these results are yet to be matched by a detailed analysis of the lingering experiments in vivo. To quantify this lingering effect in in vivo experiments, this study analyzes branching vessels in the microvasculature of Syrian golden hamsters via intravital microscopy and the use of an implanted dorsal skinfold chamber. It also presents a detailed analysis of these lingering effects of cells at the apex of bifurcating vessels, affecting the temporal distribution of cell-free areas of blood flow in the branches, even causing a partial blockage in severe cases.

Previous research in Explainable Artificial Intelligence (XAI) suggests that a main aim of explainability approaches is to satisfy specific interests, goals, expectations, needs, and demands regarding artificial systems (we call these stakeholders' desiderata) in a variety of contexts. However, the literature on XAI is vast, spreads out across multiple largely disconnected disciplines, and it often remains unclear how explainability approaches are supposed to achieve the goal of satisfying stakeholders' desiderata. This paper discusses the main classes of stakeholders calling for explainability of artificial systems and reviews their desiderata. We provide a model that explicitly spells out the main concepts and relations necessary to consider and investigate when evaluating, adjusting, choosing, and developing explainability approaches that aim to satisfy stakeholders' desiderata. This model can serve researchers from the variety of different disciplines involved in XAI as a common ground. It emphasizes where there is interdisciplinary potential in the evaluation and the development of explainability approaches., Comment: 57 pages, 2 figures, 1 table, to be published in Artificial Intelligence, Markus Langer, Daniel Oster and Timo Speith share first-authorship of this paper

Imaging of liquids and cryogenic biological materials by electron microscopy has been recently enabled by innovative approaches for specimen preparation and the fast development of optimised instruments for cryo-enabled electron microscopy (cryo-EM). Yet, Cryo-EM typically lacks advanced analytical capabilities, in particular for light elements. With the development of protocols for frozen wet specimen preparation, atom probe tomography (APT) could advantageously complement insights gained by cryo-EM. Here, we report on different approaches that have been recently proposed to enable the analysis of relatively large volumes of frozen liquids from either a flat substrate or the fractured surface of a wire. Both allowed for analysing water ice layers which are several microns thick consisting of pure water, pure heavy-water and aqueous solutions. We discuss the merits of both approaches, and prospects for further developments in this area. Preliminary results raise numerous questions, in part concerning the physics underpinning field evaporation. We discuss these aspects and lay out some of the challenges regarding the APT analysis of frozen liquids., Comment: submitted for publication assocaited to the APT&M 2020 conference

Recent advances in machine-learning interatomic potentials have enabled the efficient modeling of complex atomistic systems with an accuracy that is comparable to that of conventional quantum mechanics based methods. At the same time, the construction of new machine-learning potentials can seem a daunting task, as it involves data-science techniques that are not yet common in chemistry and materials science. Here, we provide a tutorial-style overview of strategies and best practices for the construction of artificial neural network (ANN) potentials. We illustrate the most important aspects of (i) data collection, (ii) model selection, (iii) training and validation, and (iv) testing and refinement of ANN potentials on the basis of practical examples. Current research in the areas of active learning and delta learning are also discussed in the context of ANN potentials. This tutorial review aims at equipping computational chemists and materials scientists with the required background knowledge for ANN potential construction and application, with the intention to accelerate the adoption of the method, so that it can facilitate exciting research that would otherwise be challenging with conventional strategies., Comment: 19 pages, 11 figures

Performance-influence models can help stakeholders understand how and where configuration options and their interactions influence the performance of a system. With this understanding, stakeholders can debug performance behavior and make deliberate configuration decisions. Current black-box techniques to build such models combine various sampling and learning strategies, resulting in tradeoffs between measurement effort, accuracy, and interpretability. We present Comprex, a white-box approach to build performance-influence models for configurable systems, combining insights of local measurements, dynamic taint analysis to track options in the implementation, compositionality, and compression of the configuration space, without relying on machine learning to extrapolate incomplete samples. Our evaluation on 4 widely-used, open-source projects demonstrates that Comprex builds similarly accurate performance-influence models to the most accurate and expensive black-box approach, but at a reduced cost and with additional benefits from interpretable and local models., Comment: Accepted for publication at ICSE'21

Block-sparse regularization is already well-known in active thermal imaging and is used for multiple measurement based inverse problems. The main bottleneck of this method is the choice of regularization parameters which differs for each experiment. To avoid time-consuming manually selected regularization parameter, we propose a learned block-sparse optimization approach using an iterative algorithm unfolded into a deep neural network. More precisely, we show the benefits of using a learned block iterative shrinkage thresholding algorithm that is able to learn the choice of regularization parameters. In addition, this algorithm enables the determination of a suitable weight matrix to solve the underlying inverse problem. Therefore, in this paper we present the algorithm and compare it with state of the art block iterative shrinkage thresholding using synthetically generated test data and experimental test data from active thermography for defect reconstruction. Our results show that the use of the learned block-sparse optimization approach provides smaller normalized mean square errors for a small fixed number of iterations than without learning. Thus, this new approach allows to improve the convergence speed and only needs a few iterations to generate accurate defect reconstruction in photothermal super resolution imaging., Comment: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. 11 pages, 10 figures

For the fatigue life of thin-walled components, not only fatigue crack initiation, but also crack growth is decisive. The phase-field method for fracture is a powerful tool to simulate arbitrary crack phenomena. Recently, it has been applied to fatigue fracture. Those models pose an alternative to classical fracture-mechanical approaches for fatigue life estimation. In the first part of this paper, the parameters of a phase-field fatigue model are calibrated and its predictions are compared to results of fatigue crack growth experiments of aluminium sheet material. In the second part, compressive residual stresses are introduced into the components with the help of laser shock peening. It is shown that those residual stresses influence the crack growth rate by retarding and accelerating the crack. In order to study these fatigue mechanisms numerically, a simple strategy to incorporate residual stresses in the phase-field fatigue model is presented and tested with experiments. The study shows that the approach can reproduce the effects of the residual stresses on the crack growth rate.

In a recent article in the American Economic Review, Tatyana Deryugina and David Molitor (DM) analyzed the effect of Hurricane Katrina on the mortality of elderly and disabled residents of New Orleans. The authors concluded that Hurricane Katrina improved the eight-year survival rate of elderly and disabled residents of New Orleans by 3% and that most of this decline in mortality was due to declines in mortality among those who moved to places with lower mortality. In this article, I provide a critical assessment of the evidence provided by DM to support their conclusions. There are three main problems. First, DM generally fail to account for the fact that people of different ages, races or sex will have different probabilities of dying as time goes by, and when they do allow for this, results change markedly. Second, DM do not account for the fact that residents in New Orleans are likely to be selected non-randomly on the basis of health because of the relatively high mortality rate in New Orleans compared to the rest of the country. Third, there is considerable evidence that among those who moved from New Orleans, the destination chosen was non-random. Finally, DM never directly assessed changes in mortality of those who moved, or stayed, in New Orleans before and after Hurricane Katrina. These problems lead me to conclude that the evidence presented by DM does not support their inferences.

Spot welding is a crucial process step in various industries. However, classification of spot welding quality is still a tedious process due to the complexity and sensitivity of the test material, which drain conventional approaches to its limits. In this paper, we propose an approach for quality inspection of spot weldings using images from laser thermography data.We propose data preparation approaches based on the underlying physics of spot welded joints, heated with pulsed laser thermography by analyzing the intensity over time and derive dedicated data filters to generate training datasets. Subsequently, we utilize convolutional neural networks to classify weld quality and compare the performance of different models against each other. We achieve competitive results in terms of classifying the different welding quality classes compared to traditional approaches, reaching an accuracy of more than 95 percent. Finally, we explore the effect of different augmentation methods., Comment: 9 pages,11 figures

Dynamics of adsorption and desorption of (4S)-N on amorphous solid water are analyzed using molecular dynamics simulations. The underlying potential energy surface was provided by machine-learned interatomic potentials. Binding energies confirm the latest available theoretical and experimental results. The nitrogen sticking coefficient is close to unity at dust temperatures of 10 K but decreases at higher temperatures. We estimate a desorption time scale of 1 {\mu}s at 28 K. The estimated time scale allows chemical processes mediated by diffusion to happen before desorption, even at higher temperatures. We found that the energy dissipation process after a sticking event happens on the picosecond timescale at dust temperatures of 10 K, even for high energies of the incoming adsorbate. Our approach allows the simulation of large systems for reasonable time scales at an affordable computational cost and ab-initio accuracy. Moreover, it is generally applicable for the study of adsorption dynamics of interstellar radicals on dust surfaces.

We implemented a geometry optimizer based on Gaussian process regression (GPR) to find minimum structures on potential energy surfaces. We tested both a two times differentiable form of the Mat\'{e}rn kernel and the squared exponential kernel. The Mat\'{e}rn kernel performs much better. We give a detailed description of the optimization procedures. These include overshooting the step resulting from GPR in order to obtain a higher degree of interpolation vs. extrapolation. In a benchmark against the L-BFGS optimizer of the DL-FIND library on 26 test systems, we found the new optimizer to generally reduce the number of required optimization steps.

We implemented a gradient-based algorithm for transition state search which uses Gaussian process regression. Besides a description of the algorithm, we provide a method to find the starting point for the optimization if only the reactant and product minima are known. We perform benchmarks on 27 test systems against the dimer method and partitioned rational function optimization as implemented in the DL-FIND library. We found the new optimizer to significantly decrease the number of required energy and gradient evaluations.

A trivial flaw in the utilization of artificial neural networks in interpolating chemical potential energy surfaces (PES) whose descriptors are Cartesian coordinates is their dependence on simple translations and rotations of the molecule under consideration. A different set of descriptors can be chosen to circumvent this problem, internuclear distances, inverse internuclear distances or z-matrix coordinates are three such descriptors. The objective is to use an interpolated PES in instanton rate constant calculations, hence information on the energy, gradient and Hessian is required at coordinates in the vicinity of the tunneling path. Instanton theory relies on smoothly fitted Hessians, therefore we use energy, gradients and Hessians in the training procedure. A major challenge is presented in the proper back-transformation of the output gradients and Hessians from internal coordinates to Cartesian coordinates. We perform comparisons between our method, a previous approach and on-the-fly rate constant calcuations on the hydrogen abstraction from methanol and on the hydrogen addition to isocyanic acid.

Canonical instanton theory is a widespread approach to describe the dynamics of chemical reactions in low temperature environments when tunneling effects become dominant. It is a semiclassical theory which requires locating classical periodic orbits on the upside-down potential energy surface, so-called instantons, and the computation of second order quantum corrections. The calculation of these corrections usually involves a matrix diagonalization. In this paper we present an alternative approach, which requires to solve only linear systems of equations involving sparse matrices. Furthermore, the proposed method provides a reliable and numerically stable way to obtain stability parameters in multidimensional systems, which are of particular interest in the context of microcanonical instanton theory.

Canonical instanton theory is known to overestimate the rate constant close to a system-dependent crossover temperature and is inapplicable above that temperature. We compare the accuracy of the reaction rate constants calculated using recent semi-classical rate expressions to those from canonical instanton theory. We show that rate constants calculated purely from solving the stability matrix for the action in degrees of freedom orthogonal to the instanton path is not applicable at arbitrarily low temperatures and employ two methods to overcome this. Furthermore, as a by-product of the developed methods, we derive a simple correction to canonical instanton theory that can alleviate this known overestimation of rate constants close to the crossover temperature. The combined methods accurately reproduce the rate constants of the canonical theory along the whole temperature range without the spurious overestimation near the crossover temperature. We calculate and compare rate constants on three different reactions: H in the M\"uller--Brown potential, methylhydroxycarbene $\to$ acetaldehyde and H$_2$ + OH $\to$ H + H$_2$O.

Formaldehyde (H2CO) is one of the most abundant molecules observed in the icy mantle covering interstellar grains. Studying its evolution can contribute to our understanding of the formation of complex organic molecules in various interstellar environments. In this work, we investigated the hydrogenation reactions of H2CO yielding CH3O, CH2OH, and the hydrogen abstraction resulting in H2+HCO on an amorphous solid water (ASW) surface using a quantum mechanics/molecular mechanics (QM/MM) model. The binding energies of H2CO on the ASW surface vary broadly, from 1000 to 9370 K. No correlation was found between binding energies and activation energies of hydrogenation reactions. Combining instanton theory with QM/MM modeling, we calculated rate constants for the Langmuir Hinshelwood and the Eley Rideal mechanisms for the three product channels of H+H2CO surface reactions down to 59 K. We found that the channel producing CH2OH can be ignored, owing to its high activation barrier leading to significantly lower rates than the other two channels. The ASW surface influences the reactivity in favor of formation of CH3O (branching ratio 80%) and hinders the H2CO dissociation into H2+HCO. In addition, kinetic isotope effects are strong in all reaction channels and vary strongly between the channels. Finally, we provide fits of the rate constants to be used in astrochemical models.

The vibrational spectra of simple CH$_3^+$-Rg (Rg=He, Ne, Ar, Kr) complexes have been studied by vibrational configuration interaction (VCI) theory relying on multidimensional potential energy surfaces (PES) obtained from explicitly correlated coupled cluster calculations, CCSD(T)-F12a. In agreement with experimental results, the series of rare gas atoms leads to rather unsystematic results and indicates huge zero point vibrational energy effects for the helium complex. In order to study these sensitive complexes more consistently, we also introduce configuration averaged vibrational self-consistent field theory (CAVSCF), which is a generalization of standard VSCF theory to several configurations. The vibrational spectra of the complexes are compared to that of the methyl cation, for which corrections due to scalar-relativistic effects, high-level coupled-cluster terms, i.e. CCSDTQ, and core-valence correlation have explicitly been accounted for. The occurrence of tunneling splittings for the vibrational ground-state of CH$_3^+$-He has been investigated on the basis of semiclassical instanton theory. These calculations and a direct comparison of the energy profiles along the intrinsic reaction coordinates (IRC) with that of the hydronium cation, H$_3$O$^+$, suggest that tunneling effects for vibrationally excited states should be very small.

TauD dioxygenase is one of the most studied $\alpha$-ketoglutarate dependent dioxygenases ($\alpha$KGDs), involved in several biotechnological applications. We investigated the key step in the catalytic cycle of the $\alpha$KGDs, the hydrogen transfer process, by a QM/MM approach (B3LYP/CHARMM22). Analysis of the charge and spin densities during the reaction demonstrates that a concerted mechanism takes place, where the H atom transfer happens simultaneously with the electron transfer from taurine to the Fe=O cofactor. We found quantum tunneling of the hydrogen atom to increase the rate constant by a factor of 40 at 5\textcelsius{}. As a consequence a quite high KIE value of close to 60 is obtained, which is consistent with the experimental value.

Atom tunneling in the hydrogen atom transfer reaction of the 2,4,6-tri-tert-butylphenyl radical to 3,5-di-tert-butylneophyl, which has a short but strongly curved reaction path, was investigated using instanton theory. We found the tunneling path to deviate qualitatively from the classical intrinsic reaction coordinate, the steepest-descent path in mass-weighted Cartesian coordinates. To perform that comparison, we implemented a new variant of the predictor-corrector algorithm for the calculation of the intrinsic reaction coordinate. We used the reaction force analysis method as a means to decompose the reaction barrier into structural and electronic components. Due to the narrow energy barrier, atom tunneling is important in the abovementioned reaction, even above room temperature. Our calculated rate constants between 350 K and 100 K agree well with experimental values. We found a H/D kinetic isotope effect of almost 106 at 100 K. Tunneling dominates the protium transfer below 400 K and the deuterium transfer below 300 K. We compared the lengths of the tunneling path and the classical path for the hydrogen atom transfer in the reaction HCl + Cl and quantified the corner cutting in this reaction. At low temperature, the tunneling path is about 40% shorter than the classical path.