106 articles tagged with #scientific-computing. AI-curated summaries with sentiment analysis and key takeaways from 50+ sources.
Researchers introduced Cosmo3DFlow, a generative AI framework that combines 3D wavelet transforms with flow matching to reconstruct the early universe from present-day observations. The method achieves 46x faster sampling than diffusion models, reducing computational time from minutes to seconds for cosmological simulations at 128³ resolution.
This technical guide presents twelve practical recommendations for designing AI-driven high-performance computing (HPC) workflows that balance the iterative, probabilistic nature of modern AI with traditional HPC infrastructure. The article addresses critical system-level challenges including containerization, resource management, and I/O optimization, providing researchers with a framework to transition from rigid computational pipelines to adaptive, intelligent environments.
Researchers present novel residual-centric compression methods (LBRC and NGLR) for scientific data that improve upon existing learned compression approaches by tailoring the encoding of reconstruction residuals to their structural properties. The techniques achieve 30-60% better compression ratios than Guaranteed Autoencoders and outperform the SZ compressor in high-fidelity regimes, addressing a critical bottleneck in compressing massive spatiotemporal datasets from scientific simulations.
Researchers introduce SciVisAgentSkills, a framework of reusable agent capabilities designed to enhance AI coding agents for scientific data visualization tasks across tools like ParaView and napari. Testing on 108 benchmark tasks demonstrates that these domain-specific skills improve agent performance and token efficiency, suggesting that structured procedural knowledge is essential for reliable long-horizon scientific workflows.
Researchers introduce a reformulated Neural Operators framework that models embedding evolution in d+1 dimensions, using Fourier-based operators to improve function space mappings. The approach demonstrates superior performance across multiple benchmarks while reducing computational overhead compared to traditional embedding-scaling methods.
Researchers propose a curvature-aware dynamic precision controller for physics-informed neural networks (PINNs) that automatically switches between single-precision (FP32) and double-precision (FP64) during training. The method matches full FP64 accuracy while reducing computational costs, addressing a critical trade-off in simulating complex physical systems.
Researchers present a category-theoretic framework for agentic AI systems that can revise their own representational structures during scientific discovery, rather than merely generating answers within fixed assumptions. The work demonstrates how self-revising discovery systems can be engineered for materials science through two instantiated systems: Builder/Breaker and CategoryScienceClaw.
Researchers introduce History-Bootstrapped Flow Matching (HB-ARFM), a machine learning method for reconstructing complete spatiotemporal fields from partial observations, demonstrating particular success in recovering velocity and temperature fields from limited boiling dynamics data. The approach addresses a fundamental challenge in scientific inference where incomplete observations create ill-posed inverse problems that traditional single-timestep models cannot solve.
Researchers successfully deployed a physics foundation model trained on simulations to predict laboratory turbulence behavior, achieving zero-shot generalization to experimental data without direct exposure to lab conditions. The model resolved a decades-old discrepancy between simulated and experimental Rayleigh-Taylor instability measurements, suggesting initial conditions—not fundamental physics—explain the sim-experiment gap.
Researchers present LLMFI, a fault-injection framework that systematically studies how hardware errors propagate through large language model inference across multiple domains. The study identifies critical vulnerability patterns and proposes four software-only reliability improvements, providing practical guidance for deploying LLMs in high-performance computing environments.
Researchers introduce naPINN (Noise-Adaptive Physics-Informed Neural Networks), a novel machine learning approach that recovers accurate physical equations from corrupted or noisy measurement data without requiring prior knowledge of noise characteristics. The method uses energy-based models to identify and filter outliers while maintaining data integrity, substantially outperforming existing robust PINN methods across benchmark tests with non-Gaussian noise and varying outlier rates.
Researchers introduce SimpliPy, a rule-based simplification engine that accelerates symbolic regression by 100x compared to SymPy, enabling the amortized neural symbolic regression method Flash-ANSR to match state-of-the-art genetic programming approaches while producing more concise expressions.
Researchers introduce EvoMD-LLM, a framework that adapts large language models to predict molecular dynamics by treating chemical reactions as temporal sequences with duration-aware tokens. The model achieves 66.14% accuracy on prediction tasks and demonstrates the ability to generate explanations for its predictions without explicit supervision, suggesting LLMs can effectively ground themselves in physical simulations through symbolic temporal modeling.
Researchers compared Claude Code and Codex on autonomously executing a gravitational wave analysis pipeline, revealing significant differences in speed, error handling transparency, and instruction interpretation despite converging scientific results. The study highlights critical considerations for deploying agentic AI in scientific workflows, including auditability trade-offs and the importance of precise data representation standards.
Researchers introduce Stochastic Lifting, a machine learning technique that generates diverse trajectories of stochastic physical systems by attaching random labels to state transitions during training. The method enables single-network inference to produce multiple plausible outcomes without collapsing to average predictions, advancing physics-informed AI applications.
Researchers propose Autoregression-Free Neural Operators (AFNO), a new approach for solving time-dependent partial differential equations that models continuous-time evolution in latent space rather than performing recursive predictions. By avoiding autoregressive rollout and using flow matching, AFNO reduces error accumulation over long-horizon predictions and demonstrates improved stability across six PDE benchmarks.
Researchers propose LNN-PINN, an enhanced physics-informed neural network framework that integrates liquid residual gating architecture to improve predictive accuracy for complex scientific problems. The method maintains existing physics modeling pipelines while refining the hidden-layer architecture, demonstrating consistent error reductions across benchmark tests without requiring hyperparameter adjustments.
Researchers propose SincKANs, a neural network architecture combining Sinc interpolation with Kolmogorov-Arnold Networks to improve function approximation and solve partial differential equations. The approach demonstrates superior performance compared to existing methods, particularly for functions with singularities, offering potential advances in physics-informed machine learning.
Researchers introduce SILO, a self-improvement imitation framework for protein design that optimizes protein sequences under limited evaluation budgets. The method combines hierarchical editing, stochastic beam search, and active learning to outperform existing reinforcement learning and generative approaches across multiple protein fitness landscapes.
Researchers have developed an AI agent framework that automates the translation of legacy finite-difference code into Devito, a modern computational framework. The system combines retrieval-augmented generation (RAG) with large language models and implements reinforcement learning feedback mechanisms to enable dynamic code transformation with validation across correctness, structure, and API compliance.
Researchers propose Cascaded Sensing, a machine learning framework combining autoencoders and diffusion models to reconstruct physical fields from extremely sparse sensor measurements. The approach addresses the ill-posed problem of inferring complete spatial data from limited observations by first establishing global structural anchors through coarse-scale estimation, then refining details through conditional diffusion sampling.
Researchers introduce Iterative Refinement Neural Operators (IRNO), a method that enhances neural operators by applying learned refinement modules iteratively to correct high-frequency prediction errors. The approach achieves up to 56% error reduction on turbulent flow simulations and demonstrates mathematical convergence guarantees through fixed-point iteration theory.
Researchers introduce Multi-Scale Attention Transformer (MSAT), a deep learning architecture that outperforms Fourier-based neural operators for solving PDEs on irregular domains. The model achieves 3.7x better accuracy than FNO on complex geometry problems while running 3,500x faster than competing approaches, with theoretical bounds explaining when attention mechanisms beat frequency-domain methods.
Researchers present a novel method for reconstructing continuous-time physical dynamics from discrete observations by enforcing the semi-group property of autonomous flows, using a metric called Symmetry Rupture to regularize training and guide adaptive step selection. The approach significantly outperforms Neural ODE baselines on diffusion-reaction and PDE benchmarks, reducing errors by 87% while requiring 5x fewer function evaluations.
Researchers introduce SuperMeshNet, a semi-supervised neural network framework that dramatically reduces the amount of expensive high-resolution training data needed for mesh-based simulations. By combining small paired datasets with abundant unpaired data through complementary learning, the system achieves superior accuracy while requiring 90% less supervised training data than fully supervised approaches.
