Model Predictive Control (MPC) is an d advanced control strategy widely used in the process industries and beyond. Therefore, industry is interested in the developments of MPC formulations that can enhance safety, reliability, and economic profitability of chemical processes. Motivated by these considerations, the first part of this talk focuses on the development of methods for integrating process operational safety and process economics within model predictive control system designs.

Ever wonder why, when you fly to Jeddah you don't end up in Riyadh?  The tracking devices use a nonlinear state space model to make sure your plane is on course. While inference for the linear Gaussian model is fairly simple, inference for nonlinear models can be difficult and often relies on derivative free numerical optimization techniques.  A promising method that I will discuss is based on particle approximations of the conditional distribution of the hidden process given the data. This distribution is needed for both classical inference (e.g., Monte Carlo EM type algorithms) and Bayesian inference (e.g., Gibbs sampler). 

Generalized Dynamic principal components are presented and it is shown how to define one side inear combinations of the present and past values of the series that minimize the reconstruction mean squared error (ODPC). It is shown that the ODPC introduced in this paper can be successfully used for forecasting high-dimensional multiple time series.

Magnetic sensors are deployed in many applications such as automotive, consumer electronics, navigation and data storage devices. Their market’s growth is driven by demands of higher performance and more integration; primarily to assist in the advancement of Internet of Things (IoT) and smart systems.

Gallium nitride (GaN)-based power device, e.g. Schottky barrier diodes (SBDs) and high electron mobility transistors (HEMTs), have attracted considerable research interest and well recognized as the next generation high power and high temperature devices, owing to their ultralow conduction loss and fast switching under high voltage and high frequency operations.

Since the pioneer works of Telatar, random matrix theory has found a variety of applications in engineering disciplines that, to name a few, include wireless communication and signal processing. Its scope is now going far beyond the field of mathematics, being recognized as an indispensable tool for advanced research in engineering disciplines as can be evidenced by the dramatic increase in related publications. Recently, random matrix theory has found its way into the field of big data processing, allowing accurate characterization of the performance of many algorithms met in the field of machine learning.

The ML Hub offers a 2-day short course on deep learning and the latest algorithms in artificial intelligence. The course will be given by Professor Xavier Bresson from the Nanyang Technological University (NTU) in Singapore, who is a leading researcher in the field of deep learning. The course will include the theory of deep learning techniques as well as practical exercises. Prerequisite knowledge: Basic knowledge of linear algebra (e.g. matrix multiplication) and script programming (e.g. Python, Matlab, R) are needed. The coding will be done in Python. Note, that this course has limited seating and filling registration form does not guarantee acceptance. If you are selected, you will receive a confirmation e-mail.

Mean-field game (MFG) systems of partial differential equations (PDE) arise in modeling huge populations of indistinguishable rational agents that play non-cooperative differential games. Mathematically, an MFG system comprises of a Hamilton-Jacobi-Bellman PDE coupled with a Kolmogorov-Fokker-Planck PDE in a highly nonlinear fashion. Hence, theoretical and numerical treatments of MFG systems are highly challenging problems.

This talk begins with an introduction to quantum chemistry for HPC researchers.  We discuss the main computational kernels, how their performance is limited by various bottlenecks, and algorithms and implementations that improve performance.  One particular kernel is the calculation of the Coulomb matrix.  Whereas the fast multipole method (FMM) can be used to rapidly compute the Coulomb potential for sets of point charges, continuous variants of FMM were developed in the 1990s for sets of charge distributions that arise in quantum chemistry.

The ML Hub offers a 2-day short course on deep learning and the latest algorithms in artificial intelligence. The course will be given by Professor Xavier Bresson from the Nanyang Technological University (NTU) in Singapore, who is a leading researcher in the field of deep learning. The course will include the theory of deep learning techniques as well as practical exercises. Prerequisite knowledge: Basic knowledge of linear algebra (e.g. matrix multiplication) and script programming (e.g. Python, Matlab, R) are needed. The coding will be done in Python. Note, that this course has limited seating and filling registration form does not guarantee acceptance. If you are selected, you will receive a confirmation e-mail.

Non-stationary spatial models are widely applicable in diverse disciplines, ranging from bio-medical sciences to geophysical studies. In many of theses applications, testing for structural changes in the trend and testing the specific form of the trend are highly relevant. A novel statistics based on a discrepancy measure over small regions is proposed in this paper. Such a measure can be used to construct tests for structural trends and to identify change boundaries of the trends. By virtue of the m-dependence approximation of a stationary random eld, asymptotic properties and limit distributions of these tests are established. The method is illustrated by simulations and data analysis.

This dissertation describes detailed performance engineering and optimization of an unstructured computational aerodynamics software system with irregular memory accesses on a wide variety of multi- and many-core emerging high-performance computing scalable architectures, which are expected to be the building blocks of energy-austere exascale systems, and on which algorithmic- and architecture-oriented optimizations are essential for achieving worthy performance.

In the past years, deep learning methods have achieved unprecedented performance on a broad range of problems in various fields from computer vision to speech recognition. So far research has mainly focused on developing deep learning methods for grid-structured data, while many important applications have to deal with graph-structured data.

Mean-field game (MFG) systems of partial differential equations (PDE) arise in modeling huge populations of indistinguishable rational agents that play non-cooperative differential games. Mathematically, an MFG system comprises of a Hamilton-Jacobi-Bellman PDE coupled with a Kolmogorov-Fokker-Planck PDE in a highly nonlinear fashion. Hence, theoretical and numerical treatments of MFG systems are highly challenging problems. Day 2: I will present new Fourier approximation techniques for nonlocal MFG systems and discuss how these techniques fit the variational framework presented on Day 1. Then I will introduce the Chambolle-Pock primal-dual hybrid gradient optimization method and apply a variant of this method to approximate nonlocal MFG models via Fourier approximation techniques above.

During the past two decades the availability of big scholarly data repositories such as Google Scholar, Elsevier Scopus offered tremendous opportunities to analyze scientific production and help develop models for it. The development of computerized analysis methods for these voluminous scholarly data allows to understand, quantify and predict research activities and the corresponding outcomes. The focus of this talk is on techniques to forecast future impact of a scientist; this is a very interesting problem because it allows for making effective hiring/promotion decisions and research fund allocation, among others.

Although optical sensors incorporating grating inscribed and etched fibres are now sufficiently mature and well established in the market, however, designs based on more exotic nanowires and photonic crystal fibres are becoming increasingly important and showing much improved sensitivity by accessing a larger evanescent field. Similarly, novel planar design concepts, such as the silicon slot guide-based design is showing even greater promise, allowing the exploitation of well-developed CMOS fabrication technologies for potentially low-cost sensor elements.

As big data, articial intelligence, cloud services, cellular infrastructure, content delivery; all of which entail interconnected and  sophisticated computing and storage resources. Recent studies on traditional data center networks (DCNs) revealed two key challenges: a biased distribution of inter-rack trac, and unidentied ow multi-classes best known as delay sensitive mice ow (MF) and throughput-hungry elephant ow (EF).

In this seminar I will introduce the basics of the electrical discharges (plasma) in non-polar liquids and liquid/solid interfaces and their importance in specific applications such as high-powered transformers, capacitors, cables etc. The precursor mechanism of breakdown in these liquid and liquid/solid insulating systems are usually call streamers and their characteristics vary significantly with the nature of the liquid, voltage, pressure, electrode configuration, etc.

Many central problems in geometry, mathematical physics and biology reduce to questions regarding the behavior of solutions of nonlinear evolution equations. The global dynamical behavior of bounded solutions for large times is of significant interest. However, in many real situations, solutions develop singularities in finite time. The singularities have to be analyzed in detail before attempting to extend solutions beyond their singularities or to understand their geometry in conjunction with globally bounded solutions. In this context we have been particularly interested in qualitative descriptions of blowup.

Extreme value theory (EVT) deals with the asymptotic distributions of the extremes (maximum or minimum) of a set of N random variables, as N grows large. EVT is a powerful tool to analyze the performance of generalized user selection in modern wireless Communication systems. In this talk, we will provide an overview of EVT and its application to the performance of generalized user selection for multiuser traditional wireless and cognitive radio networks.

We will present some new methods for source and parameters estimation for partial and fractional differential equations and illustrate the results with some simulations and real applications.

Mean-field game (MFG) systems of partial differential equations (PDE) arise in modeling huge populations of indistinguishable rational agents that play non-cooperative differential games. Mathematically, an MFG system comprises of a Hamilton-Jacobi-Bellman PDE coupled with a Kolmogorov-Fokker-Planck PDE in a highly nonlinear fashion. Hence, theoretical and numerical treatments of MFG systems are highly challenging problems. Day 1: I will show how to transform suitable mean-field game (MFG) systems into infinite-dimensional convex optimization problems. Furthermore, I will present Uzawa’s algorithm and augmented Lagrangian approach for solving convex optimization problems. Finally, I will demonstrate how to apply these methods to approximate solutions of corresponding MFG systems.

Runge-Kutta methods are classical and widespread techniques in the numerical solution of ordinary differential equations (ODEs). Considering partial differential equations, spatial semidiscretisations can be used to obtain systems of ODEs that are solved subsequently, resulting in fully discrete schemes. However, certain stability investigations of high-order methods for hyperbolic conservation laws are often conducted only for the semidiscrete versions.

In this talk, I will introduce a few recent works on tree visualization. First I will present a  visualization technique for comparing topological structures and node attribute values of multiple trees. I will further introduce GoTree, a declarative grammar supporting the creation of a wide range of tree visualizations. In the application side, visualization and visual analytics on social media  will be introduced. The data from social media can be considered as graphs or trees with complex attributes. A few approaches using map metaphor for social media data visualization will be discussed.

This seminar highlights a new high-power density Interconnected Modular Multilevel Converter (IMMC) with sinusoidal output voltage for multiple applications. The proposed converter utilizes wide band gap devices at a high switching frequency to achieve compact size/weight/volume. The proposed converter is modular in construction, employs high frequency L-C components and can be stacked for voltage sharing.