Algorithms, Data Structures and Computing Platforms in Numerical Electromagnetics

November 9, 2017
Versione stampabile

Hour: 12.00
Venue: Room Garda, Polo scientifico e tecnologico "Fabio Ferrari", Building Povo 1, via Sommarive 5, Povo (Trento)

Speaker

Prof. Amedeo Capozzoli - Università di Napoli Federico II, Dipartimento di Ingegneria Elettrica e delle Tecnologie dell'Informazione

Abstract

The solution of many electromagnetic problems requires the use of advanced numerical tools able to analyze and synthesize complex systems with unprecedented accuracies and processing times. Indeed, when the working scenario becomes cumbersome, approaching as much as possible what happens in reality, the massive exploitation of numerical computing turns to be mandatory. Indeed, the use of simplified models that unburden the numerical analysis sacrificing accuracy becomes unacceptable when high performance is pursued, since, even if they provide excellent results at the design stage, in practice they lead to poor practical results as arising after the experimental testing. Obviously, when keeping complexity, computing times grow, until making unaffordable the analysis and, particularly, the synthesis of the system. This occurs because, typically, the synthesis requires the repeated calculation of the electromagnetic response of the system to make the optimization process behind the design possible. 

To face the problem, the following policies can be pursued:
1) Using smart algorithms able to reduce the computational complexity of the numerical process;
2) Using data structures contributing to the mitigation of the computational complexity;
3) Using high performance, massively parallel computing platforms.

The three above points cannot be faced independently, since, to be successful, the development, the set up and the implementation of a numerical code requires a global point of view, based on a solution scheme wherein the algorithms and the data structure conform to the underlying chosen hardware to fully profit of its potentialities. When the numerical code is designed according to this philosophy, dramatically superior performance is achieved.

The Lecture will show the use of algorithms, data structures and computing platforms for some hot problems in numerical electromagnetics as antenna synthesis and RCS computation of complex objects. The results will show the achievable performance boost.

About the Speaker

Amedeo Capozzoli received the “Laurea” degree (summa cum laude) in Electronic Engineering and the Ph.D. degree in Electronic Engineering and Computer Science from Università di Napoli Federico II, Naples, Italy, respectively. Since January 2005, he has been Associate Professor of Electromagnetic Fields at Università di Napoli Federico II. Since the academic year 2012/2013 he is professor of Electromagnetic Fields and Circuits at the Italian Air Force Academy.
Honours and distinctions he has been awarded include the Telecom Italia Prize for the best degree thesis in Electronic Engineering defended at Università di Napoli Federico II, the Barzilai Prize for young scientists in September 2002 from the Italian Society of Electromagnetism, the Best Technical Paper Award from the Antenna Measurement Technique Association (AMTA) for two consecutive years (2009 and 2010), the Honourable Mention at the 5th European Conference on Antennas and Propagation (EUCAP 2011), the Nomination for the Best Paper Award at the 8th European Conference on Antennas and Propagation (EUCAP 2014). On October 2015, he has received the Senior Membership form the Antenna Measurement Technique Association. On December 2016, he has received the 2016 Best Italian EMC Poster Price at the IEEE EMC Young Professional Paolo Corona Day.

He is one of the founders, and Chair, of the Italian AMTA node, the first European node of the Antenna Measurement Technique Association. He is Associate Editor of the Express Journal of the Applied Computational Electromagnetic Society.

He is member of the AMTA-IEEE STD 149 Working Group. Since July 2016, he is Chair of the Bachelor and Master Degree Studies in Telecommunication Engineering at Università di Napoli Federico II. He is Chair of the Microwave and Millimeter Wave LAB, and of the Numerical Electromagnetics LAB at Università di Napoli Federico II. Since 2013, he is responsible of the Course on Antenna Synthesis in the framework of the European School of Antennas (ESoA), and member of the ESoA Board. He his member of the Institute of Electrical and Electronic Engineering (IEEE), of the Institute of Electronics, Information and Communication Engineers (IEICE, Japan), of the Antenna Measurement Technique Association (AMTA) and of the Applied Computational Electromagnetic Society (ACES). He has been member of Optical Society of America (OSA). On March 2013 he has given, for ACES, the Short Course “Fast Reflectarrays Antenna Synthesis on GPUs” at the 29th Annual Rev. of Progr. in Appl. Comput. Electromagn., Monterrey, CA. Since 2008 he and the research team he coordinates at Università di Napoli Federico II are working on the use of High Performance Computing Platforms, GPUs, for Computational Electromagnetics applications. In particular, innovative results have been obtained in the Analysis and Synthesis of complex antenna systems, as arrays and reflectarrays, in Tomography and Remote Sensing, in Nuclear Resonance Imaging, and in ultra-fast Ray Tracing. Such a research activity has produced the fastest Ray Tracing Tool (including diffraction terms) available up to now, and an effective and efficient synthesis procedure to design Aperiodic Reflectarray Antennas, covered by a World Patent and a US Patent, sold to the European Space Agency (ESA). The 2014 ESoA Course on Antenna Synthesis held in October 13-17 2014, he has coordinated, for the first time has introduced the Ph.D. Students to elements of GPU Computing applied to Antenna Synthesis, and has been officially sponsored by NVIDIA.

His research interests include, among others, methods to extract synthetic information on systems of sources or scatterers from field data, Adaptive Optics in Optical Astronomy, Antenna Synthesis and Diagnosis, fast Numerical Methods in Electromagnetics, GPU Computing in Electromagnetics, advanced measurement approaches in Electromagnetics, Inverse Problems and Remote Sensing.

Contact Person Regarding this Talk: andrea.massa [at] unitn.it (Andrea Massa) 

Additional Notes
Additional information can be found at www.eledia.org