Seyong Lee

🚀 Revolutionizing Nano-Optics with Physics-Informed Neural Networks (PINNs) 🌐 Here some great insights from an incredible research paper titled: "Physics-informed neural networks for inverse problems in nano-optics and metamaterials" Authored by Yuyao Chen, Lu Lu, George Karniadakis, and Luca Dal Negro, This groundbreaking study demonstrates how Physics-Informed Neural Networks (PINNs) can tackle complex inverse scattering problems in photonic metamaterials and nano-optics. By integrating physical laws directly into deep learning frameworks, the authors have opened up new avenues for designing advanced optical materials. 💡 Highlights from the Paper: Effective Medium Reconstruction: PINNs successfully reconstructed material permittivity profiles with high accuracy, validated against Finite Element Method (FEM) simulations. Inverse Design: Enabled the retrieval of optical parameters for nano-optical systems, paving the way for novel photonic metamaterials. Cloaking Design: Achieved a remarkable 75% scattering reduction for dielectric-coated nanocylinders—a major step towards invisibility cloaking! Why This Matters: Traditional methods often struggle with the multi-scale complexity of metamaterials. This study shows that by leveraging deep learning infused with physics, we can: - Design more efficient and functional optical devices. - Explore new capabilities in remote sensing and material analysis. - Address challenges in ill-posed inverse problems. 📚 Takeaway PINNs are not just a tool but a transformative framework for advancing computational physics, nano-optics, and materials science. 📖 You can read the full paper here: https://lnkd.in/dHVqvhJX Kudos to the authors for this inspiring work! Let's discuss How do you see the role of PINNs evolving in the field of materials science and beyond? 🤔

36

5 Comments

RELATIVISTIC QUANTUM MECHANICS 2008; EULER - LAGRANGE EQUATION; HIGGS BOSON FOR GATE & CSIRNET #GATEMADEEASYKOTA, #CSIRNETMADEEASYKOTA, #PHYSICSMADEEASYKOTA

1

If anyone is curious as to what I've been up to post-academia - it's this (Accelerated DFT). This should be an excellent webinar on the latest and greatest in the products and services Azure Quantum Elements has to offer.

43

2 Comments

I'm pleased to share our latest open-access publication, "Physics-Based Reliability Modeling for Control Applications: Adaptive Control Allocation," in IEEE Access. https://lnkd.in/g-VjD-Tx This work presents a systematic approach to integrating physics-based reliability models within control systems. By addressing the limitations of conventional degradation models, we introduce a reliability framework that utilizes stress-derived variables, enabling real-time, physics-informed reliability assessments. Our approach is validated through simulations on a hexarotor UAV, demonstrating improved reliability predictions and adaptive workload distribution to extend system longevity. Special thanks to Joshua Desrosiers, Zac Heit, Ishimwe Uwantare, Andrew Goodchild Remoundos, and Anas Senouci for their dedicated work. #UAV #ReliabilityEngineering #PhysicsBasedModeling #AdaptiveControl #AerospaceEngineering #IEEE #Hexarotor #ControlSystems #ReliabilityModeling #UnmannedAerialVehicles #SafetyInnovation

65

Useful guidelines from the IEEE TSE board to support discussion about what constitutes software engineering for AI. Special thanks to Seb for championing and seeing this piece through.

8

Machine learning has changed the PHY-layer security landscape. See our survey paper published in IEEE Communications Surveys & Tutorials: https://lnkd.in/esrjQ3TR

12

This article discusses the extension of the Local Density Approximation (LDA), a key method in Density Functional Theory (DFT), to handle excited states of quantum systems. Traditionally, LDA and its variants have excelled in modeling ground states, relying on the homogeneous electron gas (HEG) as a reference system. This study introduces an ensemble-based approach, incorporating nonthermal HEG states to derive an "ensemble LDA" (eLDA). This method addresses limitations of previous approaches, such as Time-Dependent DFT (TDDFT), which struggled with certain excitations like charge-transfer and singlet-triplet inversions. The ensemble LDA uses the HEG framework to approximate exchange-correlation energy contributions and demonstrates improved accuracy in predicting low-lying excitations in atoms and molecules. The work outlines theoretical foundations, practical applications, and implications for expanding the applicability of DFT to excited states, marking a significant advancement in computational materials and chemical physics. For more details, please continue reading the full article under the following link: https://lnkd.in/eQdqkQAh -------------------------------------------------------- In general, if you enjoy reading this kind of scientific news articles, I would also be keen to connect with fellow researchers based on common research interests in materials science, including the possibility to discuss about any potential interest in the Materials Square cloud-based online platform ( www.matsq.com ), designed for streamlining the execution of materials and molecular atomistic simulations! Best regards, Dr. Gabriele Mogni Technical Consultant and EU Representative Virtual Lab Inc., the parent company of the Materials Square platform Website: https://lnkd.in/epxjgisN Email: gabriele@simulation.re.kr #materials #materialsscience #materialsengineering #computationalchemistry #modelling #chemistry #researchanddevelopment #research #MaterialsSquare #ComputationalChemistry #Tutorial #DFT #simulationsoftware #simulation

1

The paper "MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials" presents a new approach to materials discovery using a Transformer-based model. The authors introduce MatterGPT, which generates solid-state materials with targeted properties by employing a notation system to encode crystal structures. This model allows for inverse design targeting both single and multiple properties simultaneously. The study showcases the potential for generating novel materials with specific properties, aiming to enhance computational material design in energy, electronics, and other fields. Please continue reading the full article under the following link: https://lnkd.in/eriedpBp #materials #materialsscience #materialsengineering #computationalchemistry #modelling #chemistry #researchanddevelopment #research #MaterialsSquare #ComputationalChemistry #Tutorial #DFT #simulationsoftware #simulation

2

I am very excited that our newly accepted IEEE TIM paper, entitled Enhancing Learning in Fine-Tuned Transfer Learning for Rotating Machinery via Negative Transfer Mitigation, can be accessed early on IEEE Xplore. In this work, we proposed a Dynamic-Feature Adaptive Threshold Normalization (D-FATN) to enhance the Transfer Learning's learning efficiency. So now, our team has a series of works to cover each step of a PdM flow: 1) sensor selection, 2) feature extraction, 3) model selection, and 4) model adaptation. Welcome to find more details below. 1) sensor selection and 3) model selection: https://lnkd.in/gFZs59we 2) feature extraction: https://lnkd.in/g5dsZxkS 4) model adaptation: https://lnkd.in/gi2H6nHy https://lnkd.in/g5dsZxkS

22

Google's new quantum chip hits error correction target https://lnkd.in/g5yUSa5U Nature paper: https://lnkd.in/gjQ8JHHb A quantum error correction break-through from Google. Quantum noise has been the biggest challenge in constructing useful quantum computers. Errors tend to compound exponentially as every next physical qubit is added. Quantum Error Correction has been under research for maybe 30 years or so. Now Google reports that they could get errors to go DOWN exponentially (the intent of error correction) with added qubits! If true that is indeed a remarkable breakthrough, and worth a Nobel!

30

3 Comments

Congratulations, Zhengliang Fang, on your paper titled: "Exploiting nonlinearity for sensitivity enhancement of novel tunable and low-power thin-film Piezoelectric-on-Silicon (TPoS) micromachined gas sensor," published in Measurement: https://lnkd.in/ebmwDwq3. This innovative research introduces a new type of AlN piezoelectric micromachined gas sensor. The sensor leverages both linear and nonlinear behaviours to achieve high sensitivity in detecting helium concentrations. Additionally, its unique design enables adjustable sensitivity thresholds, making it a versatile tool for various gas detection applications. This work was a key component of Zhengliang's PhD research, co-supervised by Stephanos Theodossiades and in collaboration with Nizar Jaber Jawabreh, PhD, CEM®. #MEMS #gassensor #sensor #MEMSsensor #AlNpiezoelectricMEMS #nonlinear #buckling #veering #powerefficientsensor

17

1 Comment

Want to leverage several quantum processing units for executing a single quantum circuit? Not possible you say? Enter "Circuit Cutting"... Near-term or NISQ quantum computers are quantum devices with limited capabilities: Only a limited number of qubits are available per quantum processing unit (QPU), and qubits / gates are prone to errors. One possible way to scale NISQ-based quantum computing coming from the software side is circuit cutting: A collection of methods to cut a large quantum circuit into a set of smaller circuits with reduced requirements regarding qubit number and circuit depth. In some cases, classical communication between the resulting sub-circuits is necessary. In other cases circuits can be cut without the need for any classical communication. Consequently, circuits can be distributed independently to all available QPUs. This calls for optimized resource utilization and a corresponding scheduling solution... Our quantum compilation research group at #FraunhoferIIS teamed up with the research group of Prof. Christian Mendl at Technische Universität München to look into the scheduling problem for QPUs in the context of quantum circuit cutting. Philipp Seitz presented our joint paper at "The Third International Workshop on Integrating High-Performance and Quantum Computing" (WIHPQC 2024) as part of the #IEEEQuantumWeek. Check out the preprint: https://lnkd.in/eMH-bmzX. Thanks for the great collaboration! 😊 #qce24 #quantum #quantumsoftware #hpc #hpcqc #fraunhofer

40

2 Comments

Pooya Rostami Mazrae from @umons presenting at MSR 2024 an open source tool, gawd, for computing syntactic differences between pairs of GitHub Actions workflow files. Download details can be found in attached photo. https://lnkd.in/eCqX8Pje

19

Researchers have discovered that applying pressure to the natural bulk heterostructure 6R-TaS₂ reveals unusual superconducting behavior and its interaction with charge-density-wave (CDW) order. The study shows that under pressure, the superconducting transition temperature (Tc) follows a nonmonotonic pattern, initially increasing, then plateauing, and eventually resurging at higher pressures. This pressure-tuned Josephson coupling reveals insights into the evolution of materials from insulator to superconductor, contributing to quantum material design. The findings were published in *Physical Review Letters*. Please continue reading the full article under the following link: https://lnkd.in/eQX3k_PP #materials #materialsscience #materialsengineering #computationalchemistry #modelling #chemistry #researchanddevelopment #research #MaterialsSquare #ComputationalChemistry #Tutorial #DFT #simulationsoftware #simulation

1