New PolyNets emulate physics without training, rivaling numerical solvers

Researchers have developed a novel approach called PolyNets that can emulate physical systems without requiring traditional training. These networks integrate a continuous-time neural representation of polynomial dynamics with a recurrent network for discrete-time integration. When the governing coefficients are known, PolyNets bypass the need for learning, enabling efficient hardware execution and parallel simulations. In tests using the Korteweg-de Vries equation, PolyNets demonstrated accuracy comparable to established numerical solvers and superior performance to trained neural models, especially at amplitudes outside their training conditions. AI

IMPACT This approach could accelerate scientific simulations by enabling faster, training-free emulation of physical systems on hardware.

RANK_REASON Academic paper detailing a new method for neural emulation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• central processing unit
• graphics processing unit
• Hugging Face
• PolyNets
• Runge–Kutta method

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →