TU Dortmund University (German: Technische Universität Dortmund) is a technical university in Dortmund, North Rhine-Westphalia, Germany with over 35,000 students, and over 6,000 staff including 300 professors, offering around 80 Bachelor's and master's degree programs. It is situated in the Ruhr area, the fourth largest urban area in Europe. The university is highly ranked in terms of its research performance in the areas of physics, electrical engineering, chemistry and economics. The university pioneered the Internet in Germany, and contributed to machine learning (in particular, to support-vector machines, and RapidMiner).
Due to their distinct ability to interconvert heat and electrical energy, thermoelectric materials may play an important role in advancing sustainable energy solutions. Contributing to the development of high-performance thermoelectric composites, we focus on a scale-bridging understanding of the underlying microstructure–property relation. To this end, we derive the governing set of multiscale equations by making use of asymptotic expansion analysis with particular focus on the temperature dependence of the material parameters and use classic scale-bridging arguments to recast the originally non-symmetric problem into its symmetric entropy-flux form. This reformulation provides the basis for the Lippmann–Schwinger solver developed in this work. In particular, we show that classic Moulinec–Suquet, forward–backward finite difference and rotated staggered grid discretisation schemes naturally extend to the thermo-electrically coupled cell problem and use state-of-the-art solution techniques to solve the coupled system of integral equations. We put particular emphasis on the derivation of the coupled Eshelby–Green operator, provide exact representations for the generalised macroscale conductivity tensors, validate the computational approach against analytical solutions, study its efficiency and accuracy for varying phase contrast, and exemplify its applicability by detailed studies of polycrystalline multiphase materials.
Aggregation schemes provide a means to reduce the computational complexity of power system operation by reducing the number of devices that are considered individually. This can be achieved with tools of computational geometry, where the feasible set is projected onto the decision variables of the point of interconnection. Set projection is computationally expensive, especially in the context of multi-period power system operation. This calls for efficiency improvements via structure exploitation of set representations. This paper proposes efficient flexibility aggregation via constrained zonotopes. We evaluate the performance of the proposed method on a 15-bus distribution grid with time-dependent elements for up to 96 timesteps. The results suggest that the presented method significantly improves computation times compared to classic polytope projection approaches.
Background Enhancing student learning with the use of cognitively and linguistically challenging questions is considered key for fostering student learning in classroom discourse. Theoretical models suggest that teachers’ use of questions is related to their achievement expectations for students. However, evidence for the interplay of teacher expectations, teacher questions, and student learning is scarce. Aims This study (1) describes the use of teacher questions in whole-classroom discourse, (2) examines relations between teacher expectations and teacher questions, (3) and investigates the effects of teacher questions on student achievement in reading, vocabulary, and mathematics. Sample The sample includes 329 first grade students and 17 teachers (nschools = 13) from 15 German language and 14 mathematics classrooms in Germany. Methods First, we coded the language-promoting quality of teacher questions based on video recordings of whole-classroom discourse. Second, we employed multilevel modeling to examine links between teachers’ (inaccurate) expectations, teacher questions, and later student achievement. Results Teacher questions were mostly of low language-promoting quality. Teachers asked significantly more questions in mathematics, including more open-ended questions. For mathematics, (inaccurately) low expectations were linked with higher frequencies of questions that promote elaboration and description. A positive effect on vocabulary emerged from questions that motivate students to hypothesize and conclude. Conclusions Overall, language-promoting questions occur comparatively rarely in content lessons. Evidence for class-level expectancy effects on teacher behavior was limited. Our findings suggest that the effectiveness of teacher questioning is domain-specific.
Data-driven mechanics aims at substituting traditional material models with discrete data sets of stresses and strains and has recently attracted increasing attention in computational mechanics. A crucial step towards the practical applicability of this method is the development of robust frameworks for history-dependent materials. We established such a framework for the one-dimensional case by introducing a neural network acting as propagator, constructing and updating an accompanying history surrogate purely from stress–strain data. When extending this framework to spatially higher-dimensional settings, additional algorithmic challenges emerge. In this contribution, we present an enhanced two-dimensional data-driven mechanics solver for evolutionary problems aimed at overcoming these challenges. This is addressed through three core enhancements: (i) we introduce a modification of the governing equations of the data-driven solver in which the Lagrange parameters, originally introduced to enforce equilibrium of forces, are reinterpreted as an incremental displacement update and thus contribute directly to the physically admissible strain solution. (ii) We adopt a different recurrent neural network architecture specifically designed to provide a compact and effective representation of the material history. (iii) We exploit the stress prediction of this network by means of a warm-start technique to reduce the inherent sensitivity of the data-driven solver with respect to its initial state. A comprehensive numerical study is carried out in order to assess the individual and combined influence of the proposed enhancements for two-dimensional plasticity problems and to compare them with our previous approach. The results demonstrate that the modified framework significantly improves both accuracy and stability compared to the earlier formulation, while preserving the fundamental characteristics of data-driven mechanics, namely the enforcement of equilibrium and kinematic compatibility, and the use of a fixed, unaltered data set. The neural network is employed exclusively for history compression and informed warm-starts, whereas the data-driven solution step is shown to effectively exploit the information contained in the discrete data, thereby achieving higher accuracy for the considered examples than a purely neural network-based finite element formulation.
Many adolescents with an immigration background report experiencing discrimination in their everyday lives, which can hinder their positive development. To better understand the impact of discrimination, this study investigates the associations between two types of perceived discrimination (personal and group) and the development of self-esteem and academic self-concept among students with an immigrant background in upper secondary schools in Germany. We examine a potential compensation effect of higher ethnic identity orientation with the country of origin. Four hundred twenty-two adolescents (52.5
