Large language models for partial differential equation workflows

2026/08/04 by Han Wan, Rui Zhang, Hao Sun
Computer Science · #cs.AI

paper · pdf

arxiv created 2026/08/04 · arxiv updated 2026/08/05

Abstract

Partial differential equations (PDEs) become actionable in science and engineering not as isolated formulae, but as executable workflows that connect modelling assumptions, governing equations, numerical solvers, diagnostics, and decisions. Large language models (LLMs) are beginning to support such workflows by linking natural language, symbolic mathematics, code, solver outputs, and feedback. Here we examine recent advances in LLM-assisted PDE research across three stages: the discovery and formulation of governing models, the generation and revision of executable numerical solvers, and the use of simulation feedback to support control, design, and optimization. Across these stages, current systems act primarily as workflow-level interfaces. Despite this progress, the field remains limited by the scarcity of high-quality datasets and benchmarks, especially for knowledge discovery and real-world applications, where expert annotation, executable problem construction, and task-level feedback require substantial domain effort. A further challenge is the persistent gap between simulation-based results and real-world scientific and engineering systems, which limits the direct transfer of numerical simulations, control policies, and optimized designs to practical settings. These challenges make LLM-assisted PDE workflows a critical testbed for developing scientific AI systems that can connect language, computation, physical constraints, and real-world decision-making.

Citations

• Agentic Symbolic Search: Characterizing PDEs Beyond Hand-crafted Expressions, Meshes, and Neural Networks
• PDE-SHARP: PDE Solver Hybrids through Analysis and Refinement Passes
• Lang-PINN: From Language to Physics-Informed Neural Networks via a Multi-Agent Framework
• Large language model-empowered next-generation computer-aided engineering
• Text-Trained LLMs Can Zero-Shot Extrapolate PDE Dynamics, Revealing a Three-Stage In-Context Learning Mechanism
• Automated Code Development for PDE Solvers Using Large Language Models
• DrSR: LLM based Scientific Equation Discovery with Dual Reasoning from Data and Experience
• Toward Knowledge-Guided AI for Inverse Design in Manufacturing: A Perspective on Domain, Physics, and Human-AI Synergy
• CodePDE: An Inference Framework for LLM-driven PDE Solver Generation
• Generative Discovery of Partial Differential Equations by Learning from Math Handbooks
• Foam-Agent: Towards Automated Intelligent CFD Workflows
• PeSANet: Physics-encoded Spectral Attention Network for Simulating PDE-Governed Complex Systems
• Large Language Model Driven Development of Turbulence Models
• OpenFOAMGPT 2.0: end-to-end, trustworthy automation for computational fluid dynamics
• Fine-tuning a Large Language Model for Automating Computational Fluid Dynamics Simulations
• From Equations to Insights: Unraveling Symbolic Structures in PDEs with LLMs
• OptMetaOpenFOAM: Large Language Model Driven Chain of Thought for Sensitivity Analysis and Parameter Optimization based on CFD
• Accelerating scientific discovery with Co-Scientist
• DeepSeek vs. ChatGPT vs. Claude: A Comparative Study for Scientific Computing and Scientific Machine Learning Tasks
• A Multimodal PDE Foundation Model for Prediction and Scientific Text Descriptions
• PDE-Controller: LLMs for Autoformalization and Reasoning of PDEs
• MetaOpenFOAM 2.0: Large Language Model Driven Chain of Thought for Automating CFD Simulation and Post-Processing
• PINNsAgent: Automated PDE Surrogation with Large Language Models
• OpenFOAMGPT: a RAG-Augmented LLM Agent for OpenFOAM-Based Computational Fluid Dynamics
• Evaluating LLMs' Divergent Thinking Capabilities for Scientific Idea Generation with Minimal Context
• Using Large Language Models for Parametric Shape Optimization
• Text2PDE: Latent Diffusion Models for Accessible Physics Simulation
• MetaOpenFOAM: an LLM-based multi-agent framework for CFD
• FLUID-LLM: Learning Computational Fluid Dynamics with Spatiotemporal-aware Large Language Models
• Unisolver: PDE-Conditional Transformers Towards Universal Neural PDE Solvers
• LLM and Simulation as Bilevel Optimizers: A New Paradigm to Advance Physical Scientific Discovery
• LLM4ED: Large Language Models for Automatic Equation Discovery
• LLM-SR: Scientific Equation Discovery via Programming with Large Language Models
• UPS: Efficiently Building Foundation Models for PDE Solving via Cross-Modal Adaptation
• Physics simulation capabilities of LLMs
• The Impact of Large Language Models on Scientific Discovery: a Preliminary Study using GPT-4
• MyCrunchGPT: A chatGPT assisted framework for scientific machine learning
• PINNacle: A Comprehensive Benchmark of Physics-Informed Neural Networks for Solving PDEs
• PDEBENCH: An Extensive Benchmark for Scientific Machine Learning
• MRKL Systems: A modular, neuro-symbolic architecture that combines large language models, external knowledge sources and discrete reasoning
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• Factorized Fourier Neural Operators
• Gradient-enhanced physics-informed neural networks for forward and inverse PDE problems
• Evaluating Large Language Models Trained on Code
• Encoding physics to learn reaction-diffusion processes
• Machine learning–accelerated computational fluid dynamics
• Physics-informed neural networks with hard constraints for inverse design
• Physics-Informed Neural Networks with Hard Constraints for Inverse Design
• Fourier Neural Operator for Parametric Partial Differential Equations
• Learning to Control PDEs with Differentiable Physics
• Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators
• Direct shape optimization through deep reinforcement learning
• Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control
• Data-driven discovery of partial differential equations
• Discovering governing equations from data: Sparse identification of nonlinear dynamical systems