Park, Pennsylvania Park – “A revolution in physics, science and engineering is being promoted,” says Robin Truey, founder and chairman of physics, Tuluie, who was founded in May 30th in a reunion of Pennsylvania State Laboratory and Cosmos (IGC) alumni and friends, Tuluie, who was appointed in 1995, who was awarded in 1995. of engineering.
Deep learning is a type of artificial intelligence machine learning inspired by neural networks in the human brain, using several artificial neurons to process data in a variety of ways. Just like large-scale language models (LLMS), a type of Deep Learning model – Openai's ChatGpt and Microsoft Copilot “train” existing texts and languages, and logical sentences, large-scale physics models, large-scale geometric models train data from physics, geometry, spatial relationships, and to generate predictions of variables such as fluctuations in the geometry of air and fluids. From aircraft wings and hydropowered blades to Formula 1 vehicle tires and heat exkengers used to keep batteries cool in electric vehicles, PhysicsX addressed the engineering challenges of creating large-scale physics and geometric models to optimize structures.
According to Tuluie, traditional analysis of structural optimization could involve performing many simulations, such as how adjustments to the geometry sides of the plane's wings reduce resistance. But even the fastest supercomputers can take hours, days, or even months to run for the hundreds of simulations you need. However, after a day or two training, the deep geometry model can achieve the same results within a second.
“These deep learning models can do this within seconds and take advantage of that speed,” he said. “We have a real-time model with the fidelity of very high-performance numerical simulations.”
He said that the deep geometric model uses geometry to inform physics. In the example of airplane optimization, instead of describing wings with parameters such as the length and angle of the wings that are input to the model, the model uses a geometric “mesh.” This could result in 10 million or 100 million points, or even hundreds of millions of points in more complex shapes like cars. PhysicsX researchers trained plane optimization models with meshes from a variety of aircraft, from drones to planes and even birds and insects.
“When you start learning these geometry, you start to see these clusters. We go from what we know – birds, drones, or planes – we cover things in between,” Tuluie said. “When we are in the space between birds and planes and we are in the blanks, we can explore the space without a prescription. [We can] Discover geometry that transcends human dimensions, completely new geometry. And now we are exploring not only aerodynamic performance, but all sorts of geometry that makes planes safer. ”
Before founding PhysicsX, Tuluie was previously head of research and development at Renault (Alpine) F1. His innovation helped Formula 1 teams win consecutive double world championships, serving as Bentley Motors' vehicle technology director. His expertise in numerical modeling dates back to his postdoctoral studies in Pennsylvania, where he used numerical simulations to explore the background of cosmic microwaves – radiation filling all the spaces of the observable universe.
The IGC reunion event welcomed alumni and institute friends and current members for a series of presentations at IGC, networking opportunities, panel discussions, and showcases of new developments and initiatives. Other speakers include current members Tetiana Pitic, N3AS postdoctoral researcher at the University of California, Berkeley, and visiting scholars from Pennsylvania. Bingy Wang, an assistant professor of astronomy and astrophysics in Pennsylvania. Tristan McClary, a former Doctoral Scholar, professor of pure and applied mathematics at Trinity College Dublin. Goldana Tessick, data scientist in the Ottawa meta. Jonathan Trump, associate professor of physics at the University of Connecticut. Surabhi Sachdev, assistant professor of physics at Georgia Institute of Technology.
IGC is a research community that collectively seeks to push boundaries of understanding of the fundamental forces of nature and how to shape the evolution of the universe. To meet this challenge, it requires a joint effort between many areas of convergence of expertise and a passion for new ideas and new perspectives. IGC strives to provide a fertile environment for collaboration between domains of particle physics, gravity physics, mathematics, cosmology, astrophysics, statistics and computation.
