预告|ElectromagneticScienceForum2023线上论坛即将召开

关注我们并设置为星标账号，第一时间获得本刊最新资讯！

The Electromagnetic Science Forum is an academic forum combining both online and offline formats. This forum invites renowned experts from the field of electromagnetism to deliver themed academic presentations. Each session in this expert-sharing series features no fewer than three experts, with an audience of several hundred attendees. The primary objective of this forum is to facilitate the concentrated sharing of cutting-edge academic perspectives from experts across various domains of electromagnetics, ultimately fostering academic influence.

Nanophotonic methods provide intriguing options for manipulating scintillation phenomena. We will outline recent developments in this domain, along with our theoretical framework for modeling these occurrences, supported by our experimental findings. Additionally, Smith-Purcell radiation, characterized by fast electrons interacting with nano-structured materials to produce light, offers a broad spectrum of possibilities for creation of novel light sources. We will discuss our new theoretical framework designed to comprehend and tailor such phenomena, as well as our techniques for boosting Smith-Purcell radiation.

Modern electromagnetic concepts and novel antenna designs have created new pathways for ingenious applications that could not be thought of or realized a decade ago. Incorporation of concepts such as nature-inspired optimizations integrated with artificial intelligence (AI), 3-dimensional (3D) printing, meta-materials and electromagnetic band gaps (EBGs), orbital angular momentum (OAM), flat lenses, deployable mesh reflectors and reflectarrays, modularized and adaptive multi-beam configurations, and others have resulted in out-of-the-box designs for diverse applications such as planetary, Internet of Things (IOT), SmallSats/CubeSats, remote sensing, and others.  The impact and inspiring opportunities of these novel concepts and designs specifically tailored for future communication systems will be highlighted. Topics presented in this plenary talk are primarily based on the author’s papers listed below (www.antlab.ee.ucla.edu):

Y. Rahmat-Samii, et al, "Development of Highly Constrained 1 m Ka-Band Mesh Deployable Offset Reflector Antenna for Next Generation CubeSat Radars," IEEE Transactions on Antennas and Propagation, vol. 67, no. 10, pp. 6254-6266, October 2019.

Y. Rahmat-Samii, et al,, "For Satellites, Think Small, Dream Big: A review of recent antenna developments for CubeSats," IEEE Antennas and Propagation Magazine, vol. 59, no. 2, pp. 22-30, February 2017.

Y. Rahmat-Samii and R. Haupt, "Reflector Antenna Developments: A Perspective on the Past, Present and Future," IEEE Antennas and Propagation Magazine, vol. 57, no. 2, pp. 85-95, April 2015.

Y. Rahmat-Samii and A.C. Densmore, "Technology Trends and Challenges of Antennas for Satellite Communication Systems," IEEE Transactions on Antennas and Propagation, vol. 63, no. 4, pp. 1191 - 1204, November 2014.

Y. Rahmat-Samii, "GTD, UTD, UAT, and STD: A Historical Revisit and Personal Observations," IEEE Antennas and Propagation Magazine, vol. 55, no. 3, pp. 29-40, May 2013.

Y. Rahmat-Samii, et al, "Nature-Inspired Optimization Techniques in Communication Antenna Designs," Proceedings of the IEEE, vol. 100, no. 7, pp. 2132-2144, July 2012.

A recent report by the US Department of Energy defines the area of scientific machine learning as “a core component of artificial intelligence (AI) and a computational technology that can be trained, with scientific data, to augment or automate human skills”, which has “the potential to transform science and energy research”. We explore the potential of scientific machine learning methods for problems in computational electromagnetics starting from standard microwave structure design and Multiphysics modeling, employing an unsupervised learning  strategy based on Physics-Informed Neural Networks (PINN). PINNs directly integrate physical laws into their loss function, so that the training process does not rely on the generation of ground truth data from a large number of simulations (as in typical neural networks).

Moreover, we demonstrate the impact of machine learning on the computational modeling of radiowave propagation scenarios. We build convolutional neural network models that can process the geometry of indoor environments, along with physics-inspired parameters, to rapidly estimate received signal strength (RSS) maps. We show the *generalizability* of these models, which is their ability to "learn" the physics of radiowave propagation and produce accurate modeling predictions in new geometries well beyond those included in their training set. These models can be used to rapidly optimize the position of transmitters in wireless area networks, to maximize coverage, channel capacity and other relevant system-level metrics.

Heterogeneous integration (HI) is bringing a revolutionary change to the packaging industry. Electronic design automation (EDA) will need to undergo transformational changes to manage these new challenges and simple incremental changes to their existing infrastructure may not be sufficient. The emergence of new materials and devices with applications in computing, communication, health care is expanding the stage for packaging and creating new opportunities for HI in which multi-physics, multi-scale and multi-level analyses will provide the needed solutions. Brain-inspired architecture are promising order of magnitudes in computing speed and efficiency. Chiplet-based system-in-package (SiP) integrated circuits are being proposed for these new applications where the diversity of computing environments and applications invariably means that a single processor design cannot be used in all the different environments, ranging from data centers to IoT devices.  Disagreggation from SoC to chiplet is currently performed in an ad-hoc heuristic manner in which there is no established methodology to optimize or determine how many separate chiplets should be used and what functional blocks should be integrated together within one chiplet. In addition, methods for placement and routing through the power distribution networks (PDN) for each chiplet are not well established. Artificial intelligence (AI) and machine learning (ML) are being proposed as solutions for both disaggregation and layout.

This talk focuses on current state-of-the-art, challenges and potential solutions for CoDesign. The vision is to create an environment where design closure is achieved with a minimum number of iterations meeting all requirements for performance and cost. Such environment must leverage from currently available technologies, namely computing power, algorithms and artificial intelligence. Novel techniques in AI/ML and new breakthroughs in materials that could potentially benefit HI will be presented. Discussions will assess these potentials as well as the role of the EDA community.

Non-Hermitian Hamiltonians, which can be realized in classical wave systems such as photonic lattices containing gain and loss, exhibit numerous unconventional behaviors. For instance, it has long been known that parity-time reversal symmetric Hamiltonians can host modes with real energy eigenvalues, despite not conserving energy in detail. In this talk, I discuss two newly discovered phenomena in non-Hermitian lattices: (i) a class of non-Hermitian Hamiltonians that have Dirac point degeneracies, giving rise to anomalous Klein tunneling, and (ii) lattices that host a continuum of bound states, in violation of the usual distinction between bound states and free states in Hermitian wave physics. These phenomena can be realized using appropriately designed photonic metamaterials.