Tag
论文提出 L-SplineFNO，一种将 FNO 主干与稀疏可学习张量积 B 样条传输耦合的内存高效神经算子，用于在非结构化网格上近似 PDE 解算子，在六个三维工程基准上实现高精度并可扩展至 4600 万节点，还支持零样本超分辨率和基于 Bernstein 凸包与 Taylor 模型的严格预测界。
该论文研究了在稀疏噪声传感器观测下，神经代理模型在空间窗口与历史时间之间的输入分配权衡，并通过贝叶斯最优误差分析与 Burgers、Kuramoto-Sivashinsky 方程实验，给出了窗口按场的相关长度设定、输入预算随预测时距调整的设计规则。
提出DAIST：一种可组合的AI加速迭代求解器，通过域分解将3D-IC全局热仿真拆分为块级子问题，用神经算子替代子求解器并经界面温度/热流通量迭代耦合，实现跨拓扑复用并在多芯片系统上相对FEM取得最高178倍加速，温度误差低至0.068%。
The paper investigates whether text conditioning actually steers neural PDE surrogates using OperatorCLIP, a CLIP-style contrastively aligned FNO with FiLM control, finding that constant-prompt controls often match or beat task-text conditioning and that single-description training makes text alignment unidentifiable.
This paper introduces the Function-Space Transformer (FST), a framework that learns from functions over continuous domains by storing features at input-predicted anchor locations with recursive function-space refinement. FST outperforms Perceiver IO on PDEBench prediction, is competitive with Fourier Neural Operators, and beats ViT on ImageNet-1K with fewer parameters.
The paper proposes two neural network surrogate models that learn the spectrum of the Laplace operator directly from domain geometry, enabling fast differentiable shape optimization of eigenvalue functionals. A Fourier-coefficient MLP reaches 0.2% precision on star-shaped domains, while a landscape-function model achieves 1% mean relative error on the first ten eigenvalues, outperforming FNO baselines and recovering classical spectral optima.
KernelOnet proposes an interpretable neural operator framework that explicitly incorporates kernel functions in place of trunk networks, offering data-driven, physics-informed, and hybrid kernel strategies for PDE solving with higher accuracy and fewer parameters than DeepONet.
This paper proposes a minimal transverse-subspace representation that encodes a D-component divergence-free vector field as a D−1-component field, enabling end-to-end divergence-free operator learning without projection or auxiliary potentials, achieving lower error and greater robustness.
Animashree Anandkumar of Caltech and Accelerated Understanding will deliver the SC26 keynote on 'The Next Scaling Law after LLMs: AI that Understands the Physical World,' highlighting advances from FourCastNet to AI-driven scientific discovery in weather, fusion, and biology.
This paper investigates how shared spatial computation in neural operators can reduce energy consumption for virtual sensing applications, particularly in nuclear energy systems, while maintaining prediction accuracy.
These lecture notes introduce Physics Informed Neural Networks (PINNs) and Neural Operators, covering their implementation in PyTorch and open-source libraries like NVIDIA PhysicsNeMo, with applications in engineering and physics.
Researchers at UCLA, in collaboration with Caltech and NVIDIA, are using neural operators to advance AI-for-science applications in controlling complex quantum systems.
The paper introduces Physics-Informed Conformal Prediction (PI-CP), a framework that embeds PDE residuals into conformal prediction to provide distribution-free, spatially adaptive uncertainty quantification for neural operators like the Fourier Neural Operator (FNO).
This paper introduces topology as a distinct generalization axis for neural operators in PDE solving, using Hodge heat flow and the TopoBox-3D framework to show its influence across the Hodge spectrum beyond geometric compatibility.
This paper introduces LatentDDM, a method that pretrains neural operators on small subdomains and composes them to improve accuracy and reduce adaptation cost for physical simulations in varying domains.
The paper introduces equation recast, a method that reformulates parametric operator learning as a single canonical operator, enabling zero-shot prediction across new parameter regimes by analytically deriving parameter-induced variations.
The paper proposes a data-driven constitutive modeling framework using neural operators with causal attention to predict stress response in path-dependent materials, focusing on plasticity and damage, with accurate and parallelizable predictions.
This paper introduces a physics-integrated operator learning framework using a feed-forward Gaussian splatting representation to directly incorporate PDE operators, reducing errors in long-horizon autoregressive predictions for spatiotemporal systems.
This paper proposes a Sparse-Activation-ReLU (SAR) layer for low-latency, energy-efficient virtual sensing, achieving significant improvements in latency-error-energy metrics and reducing errors through synthetic knowledge distillation.
Accelerated Understanding Inc has launched a new AI model that replaces transformer architecture with neural operators, marking a shift in AI model design.