partial differential equation
PulseAugur coverage of partial differential equation — every cluster mentioning partial differential equation across labs, papers, and developer communities, ranked by signal.
3 天有情绪数据
-
Paper analyzes SGD dynamics in high-dimensional linear networks
A new paper details the high-dimensional behavior of stochastic gradient descent (SGD) on diagonal linear networks. The research shows that in high dimensions, SGD dynamics can be accurately modeled by a stochastic diff…
-
New method boosts PDE pre-training with adaptive operator transformation
Researchers have developed AOT-POT, a novel method for pre-training neural operators on diverse partial differential equation (PDE) datasets. This approach transforms complex solution operators into simpler, aligned for…
-
NEST framework learns local physics for geometry-universal PDE solving
Researchers have developed a new framework called NEST (Neural-Schwarz Tiling) for solving partial differential equations (PDEs) across various geometries and scales. Unlike previous methods that trained global operator…
-
New hybrid neural integrator improves accuracy for nonlinear dispersive equations
Researchers have developed HIN-LRI, a novel hybrid framework that combines classical numerical solvers with neural operators to improve the accuracy of solving nonlinear dispersive partial differential equations (PDEs).…
-
PerFlow model speeds up spatiotemporal dynamics reconstruction with physics embedding
Researchers have introduced PerFlow, a novel method for reconstructing spatiotemporal dynamics governed by partial differential equations (PDEs) from sparse data. This physics-embedded rectified flow model decouples obs…
-
New VMLFN method accelerates multiphysics simulations with neural networks
Researchers have developed a new method called Variational Matrix-Learning Fourier Networks (VMLFN) to create efficient surrogate models for multiphysics simulations. This approach uses a sine neural representation and …
-
PILIR模型克服谱偏差，提高PDE求解精度
研究人员推出了一种新颖的物理信息神经网络方法PILIR，旨在克服谱偏差的限制。PILIR将物理域分离为离散的潜在特征空间和连续的解码器，并使用可学习的网格来编码空间局部性。这使得模型能够更有效地捕捉高频细节，从而提高求解偏微分方程的收敛性和准确性。
-
AI solves complex inverse partial differential equations, a major math challenge
Researchers have employed artificial intelligence to solve a challenging class of mathematical problems known as inverse partial differential equations (PDEs). This AI-driven approach offers a novel method for finding s…
-
New Dirac-Frenkel-Onsager principle enhances PDE solution parametrization
Researchers have introduced a new principle for minimizing residuals in nonlinear parametrizations of partial differential equation (PDE) solutions. This Dirac-Frenkel-Onsager principle addresses ill-conditioning by int…
-
New grey-box method integrates physics models into generative AI
Researchers have developed a novel grey-box method that integrates incomplete physics models into generative AI models, specifically flow matching and diffusion models. This approach learns dynamics from observational d…
-
LatentPDE framework reconstructs sparse scientific data using interpretable PDE representations
Researchers have developed LatentPDE, a new framework that uses latent diffusion models to improve scientific data reconstruction. This model addresses challenges like noise, incomplete data, and low resolution by creat…
-
New PDE models enhance image despeckling while preserving details
Researchers have developed two new partial differential equation (PDE) based frameworks to improve image despeckling, a process that reduces noise while preserving details. One framework uses a weighted combination of s…