Protochips的动态

?????? A spiking neuromorphic algorithm for solving large, sparse linear systems (Ax=b) like those arising from finite element methods for solving partial differential equations ?? Solving Sparse Finite Element Problems on Neuromorphic Hardware Bradley H. Theilman, James B. Aimone Abstract We demonstrate that scalable neuromorphic hardware can implement the finite element method, which is a critical numerical method for engineering and scientific discovery. Our approach maps the sparse interactions between neighboring finite elements to small populations of neurons that dynamically update according to the governing physics of a desired problem description. We show that for the Poisson equation, which describes many physical systems such as gravitational and electrostatic fields, this cortical-inspired neural circuit can achieve comparable levels of numerical accuracy and scaling while enabling the use of inherently parallel and energy-efficient neuromorphic hardware. We demonstrate that this approach can be used on the Intel Loihi 2 platform and illustrate how this approach can be extended to nontrivial mesh geometries and dynamics. ?? https://lnkd.in/dAf5nMiz #machinelearning

Given the large cost of computing and available state-of-the-art tools, the accuracy of an atomistic simulation is often sacrificed for the efficiency (modeling or muddling?). Recently, the rise of #machinelearning potential has nicely solved this problem in molecular dynamics simulation. However, for simulation in the Grand Canonical ensemble (such as #GCMC), the varying number of particles/molecules in simulation necessitates a large set of training data generated from DFT calculations for all possible distributions of particle/molecules, which is still challenging. Here, we developed a hybrid-potential method: the sorbate-sorbate interaction is described by classical potentials, while the sorbate-sorbent interaction is rendered by a machine-learning interatomic potential. ?For the first time, we show both accuracy and efficiency for GCMC of #chemisorption.

?? We're thrilled to unveil our latest research from the Clarendon Laboratory, the University of Oxford, and the Universit?t Hamburg. Our study presents a breakthrough in computational fluid dynamics (#CFD) using quantum-inspired tensor network algorithms. ?? Focusing on wall-bounded flows, this work extends beyond the conventional direct numerical simulations (DNS) by employing tensor networks to efficiently simulate fluid dynamics under complex boundary conditions. Our framework utilizes matrix product states to represent velocity components, significantly reducing the complexity and computational resources required, showcasing an order of magnitude improvement in runtime on standard hardware setups. ?? Our model addresses the incompressible Navier-Stokes equations within a lid-driven cavity scenario, achieving excellent validation against low Reynolds numbers benchmarks and exploring high Reynolds dynamics. Notably, this methodology is not only faster but also scalable and adaptable to a broader range of flow types. ?? This research not only paves the way for practical quantum computational fluid dynamics but also demonstrates the potential for significant computational speed improvements in existing classical systems. ?? Dive into our full paper for a deep dive into the mechanics, results, and implications of this innovative approach: https://lnkd.in/ej8bUsiZ ?? We thank our colleagues and the research community for their invaluable support and discussions that have greatly contributed to this project. #FluidDynamics #QuantumComputing #TensorNetworks #Innovation #ResearchImpact #UniversityOfOxford #ComputationalScience Martin Kiffner Dieter Jaksch

???????????????? ??????-??: ???????????????????????? ?????????????????????????????? ?????????????????? ???????????? ???????? ????/???? Electromagnetic devices are everywhere in our daily lives, ranging from consumer electronics, home appliances to electric vehicles and electric aircraft. Traditional simulations of electromagnetic components rely heavily on numerical methods. With the advancements of machine learning (ML), particularly the geometric deep learning and physics-informed ML, the governing equations of electromagnetic phenomena can now be solved more efficiently. This presentation will showcase the workflow of using Ansys Maxwell, our gold standard electromagnetic simulation tool, together with Ansys SimAI for electromagnetic field training and prediction. Interested to learn more ? Register and bookmark this session: https://ansys.me/3JQJ4by Peng (Ben) HAN #SimulationWorld

Researchers at Max Born Institute reveal fundamental spatial limits of all-optical magnetization switching in GdFe alloys, marking a significant step for future data storage technology. The minimum size for AOS is around 25 nm... https://lnkd.in/gUktJyde

?? Exciting announcement! Make sure to check out the highly viewed paper titled "Models for Simulation of Fractal-like Particle Clusters with Prescribed Fractal Dimension" ??Link: https://brnw.ch/21wOdYA #fractaldimension #particlecluster #simulation #generationalgorithm

I am thrilled to announce that our latest research article has been published in the journal Computational Materials Science. Our work, titled "3D Denoising Diffusion Probabilistic Models for 3D microstructure image generation of fuel cell electrodes", presents a novel approach to generating realistic synthetic 3D microstructures, extending its application to multi-phase materials such as fuel cell electrodes. This research combines Denoising Diffusion Probabilistic Models (DDPM) techniques with material science to address complex challenges in the field, offering a new perspective on understanding and optimizing material behavior. ?? You can access the article here: https://lnkd.in/eEqCZrZm #Research #ComputationalMaterialsScience #MachineLearning #MaterialScience #DDPM #3DMicrostructures #FuelCellElectrodes

Modelling the viscosity in liquids defied physicists for centuries, but in this paper I present a new and intriguing model, based on the Dual Model of Liquids (DML), an equally intriguing statistical model of liquids, published in several recent papers. Experimental findings of the last 30 years performed by means of X-ray scattering, revealed the hidden nature of liquids. On the mesoscopic scale, liquids exhibit a grainy nature: they are made by quasi-solid islands, swimming through the amorphous liquid phase, as if the liquid did not melt instantly at the melting temperature, but gradually. Energy and momentum perturbations propagate through the liquid by means of wave-packets. Their interaction with the quasi-solid icebergs allows propagation of energy and momentum through them by means of quasi-harmonic waves. The DML allows explaining the thermo-mechanical effects in liquids out of equilibrium, providing a physical interpretation of the heat propagation (Cattaneo) equation, providing the expression for the specific heat of liquids, up to give the correct order-of-magnitude for the relaxation times. In this last application, the DML reveals its capability in providing the physical mechanism by means of which momentum flux propagates through liquids, allowing to model the shear viscosity. The model even provides the first-ever theoretical interpretation of an unexpected mechano-thermal effect recently discovered in liquids. The expression of viscosity shows the typical Arrhenius trend, shows an explicit dependence upon the sound velocity and the collective vibratory degrees of freedom, and depends upon the Boltzmann and Planck constants. Finally, the physical model is coherent with the Onsager postulate of microscopic time reversibility as well as with time’s arrow for macroscopic dissipative mechanisms.

?? Fresh on the arXiv ?? Our latest efforts on implementing the Lattice Boltzmann Method (LBM) on quantum computers. We've developed a quantum algorithm to handle that pesky quadratic nonlinearity in the Lattice Boltzmann collision operator. Our approach decomposes the operator into manageable quantum gates, significantly reducing circuit width and depth. Verified with 1D and 2D flow test cases, this method promises a leap forward in quantum fluid dynamics simulations. #QuantumComputing #FluidDynamics #LatticeBoltzmann #ResearchInnovation

"From Classroom to Laboratory: How I'm Using Light to Solve Surface Chemistry Mysteries" Have you ever wondered how we can "see" electrical charges on surfaces at the molecular level? Laser technology measures these invisible forces that shape everything from battery efficiency to biological processes. This is something unique and novel: https://lnkd.in/gdNnpCkn