I³MS - Firaha Seminar
Dr. Dzmitry Firaha - Reconstruction of the Microcanonical Rate Constant from Experimental Thermal Data
The modeling of the kinetics of gas-phase chemical processes with complex reactions is a challenging task. Before modeling, the mechanism of the process needs to be proposed in the form of simple steps with known or estimated rate constants. Some chemical processes like combustion and thermal decomposition are highly sensitive to temperature and external pressure. In the straightforward approach, the modeling of such processes is a nontrivial task. In a direct approach, a lot of experimental data is needed to provide a smooth function of the rate constant depending on temperature and pressure, k(T, p). In the alternative approach (a master equation approach) pressure and temperature are used to specify the energy distribution among particles and the number of collisions between them. In such the approach, the knowledge of the energy dependent rate constant, k(E), is required. The so-called microcanonical rate constant, k(E), is hard to obtain directly from the experiment. However, the measurements are usually performed at fixed temperature providing the energy distribution for the reacting molecules allowing for recovering k(E) from k(T, p) employing the real Laplace inversion together with regularization procedure. In the talk, a summary of the possible methods for the Laplace inversion of the input data will be presented. Also, the issue on the regularization of the input data will be discussed to overcome the ill-posedness of the inversion problem.
I³MS - Sayadi Seminar
Dr. Taraneh Sayadi - Optimization and Control of Complex Flows using Adjoint-Based Methods
Institut für technische Verbrennung, RWTH Aachen University
Numerical simulations of multiphysics and multiscale phenomena in fluid mechanics have advanced remarkably over the past decades. Complex physical processes, including among others multiphase flows, combustion and acoustics problems, and turbulent and thermal flows, can now be simulated with an astonishing degree of fidelity and accuracy. In spite of these advances, specifically with regards to more complex flows, modeling and simulation technologies remain at the stage of observation, reproduction, and prediction. However, optimization and control of such flows, by enhanced designs or active control strategies, is crucial for improvements in performance and robustness, and is necessary for venturing beyond standard operating conditions. The transition from model-based numerical solvers to model-based design and optimal control requires additional technology that enables relatively easy access to “inverse" information or backward solution. To date, this information has only been extracted from simulations of simplified configurations with additional unrealistic assumptions. In related fields (aero-dynamics, aero-acoustics), inverse optimization and control have improved airfoil shapes and reduced noise levels. Complex flows, including combustion or interfaces, however, constitute a far larger step in complexity, due to the presence of unsteadiness and nonlinearities, and therefore, require advanced techniques such as adjoint-based optimization. Throughout this talk we will first introduce an adjoint-based algorithm suitable for complex flow configurations, and then provide examples of reactive and interfacial flows where this algorithm has been implemented.
I³MS - Gatto Seminar
Dr. Paolo Gatto - Efficient Preconditioning of hp-FEM Matrices by Hierarchical Low-Rank Approximations
I³MS - Kalidindi Seminar
Prof. Dr. Surya Kalidindi - Materials Data Science and Informatics: A Key Enabler for Accelerated Materials Design, Development, and Deployment
Woodruff School of Mechanical Engineering, Georgia Institute of Technology, USA
The slow pace of new/improved materials development and deployment has been identified as the main bottleneck in the innovation cycles of most emerging technologies. The recent advances in data science can be leveraged suitably to address this impediment by effectively mediating between the seemingly disparate, inherently uncertain, multiscale and multimodal measurements and computations involved in the current materials development efforts. Proper utilization of modern data science in the materials development efforts can lead to a new generation of data-driven decision support tools for guiding effort investment (for both measurements and computations) at various stages of the materials development. It should also be recognized that the success of such ecosystems is predicated on the creation and utilization of integration platforms for promoting intimate, synchronous, collaborations between cross-disciplinary and distributed team members. This presentation provides a summary of recent advances made in our research group, and outlines specific directions of research that offer the most promising avenues.
EU Regional School - Ney Seminar
Prof. Dr. Hermann Ney - Human Language Technology and Machine Learning: From Bayes Decison Theory to Deep Learning
RWTH Aachen University
Spoken and written language and the processing of language are considered to be inherently human capabilities. With the advent of computing machinery, automatic language processing systems became one of the corner-stone goals in artificial intelligence. Typical tasks involve the recognition and understanding of speech, the recognition of text images and the translation between languages. The most successful approaches to building automatic systems to date are based on the idea that a computer learns from examples (possibly very large amounts) and uses plausibility scores rather than externally provided categorical rules. Such approaches are based on statistical decision theory and machine learning. The last 40 years have seen a dramatic progress in machine learning for human language technology. This lecture will present a unifying view of the underlying statistical methods including the recent developments in deep learning and artificial neural networks.