New S2-PINN method advances uncertainty quantification for random PDEs

Researchers have developed S$^{2}$-PINN, a novel approach to uncertainty quantification for random partial differential equations. This method utilizes a learnable Gaussian spatial dictionary, Fourier temporal features, and a generalized polynomial chaos (gPC) stochastic basis, all integrated via a low-rank Canonical Polyadic (CP) tensor decomposition. S$^{2}$-PINN is trained using a hybrid loss function and has demonstrated superior accuracy and calibration compared to nine baseline methods across various benchmarks, including Poisson, Darcy, and Navier-Stokes problems. AI

IMPACT This research offers a more accurate and parameter-efficient method for uncertainty quantification in scientific computing, potentially improving simulations in fields relying on complex physical models.

RANK_REASON The cluster contains a research paper detailing a new method for solving random partial differential equations. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Canonical Polyadic Decomposition of Third-Order Tensors: Reduction to Generalized Eigenvalue Decomposition
• Darcy
• Fourier
• Gaussian function
• NAVIER STOKES ANALYSIS OF THE AERODYNAMIC PROPERTIES OF COAXIAL ROTORS
• Poisson
• S$^{2}$-PINN

• paper
• infra

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