{"slug": "evopinn-agentic-discovery-of-executable-algorithms-for-physics-informed-neural", "title": "EvoPINN: Agentic Discovery of Executable Algorithms for Physics-Informed Neural Networks", "summary": "Researchers introduced EvoPINN, an agentic framework that uses a large language model (LLM) to automatically discover executable algorithms for physics-informed neural networks (PINNs), reducing relative L2 error across diverse partial differential equation (PDE) regimes. The framework autonomously invented a novel architecture, SLRC-PINN, whose performance gains persist under parameter-matched comparisons, demonstrating the viability of execution-grounded agents for scientific computing.", "body_md": "arXiv:2607.26490v1 Announce Type: new\nAbstract: Physics-informed neural networks (PINNs) have emerged as a powerful paradigm for solving partial differential equations (PDEs), yet their performance heavily relies on the manual, trial-and-error engineering of neural representations, loss formulations, and optimization dynamics. While Large Language Models (LLMs) offer a promising avenue for automated design, unconstrained code generation often yields mathematically invalid or numerically unstable solutions under strict scientific computing constraints. To bridge this gap, we propose \\textbf{EvoPINN}, an agentic framework that reformulates PINN development from labor-intensive manual design into a rigorous, execution-grounded algorithm discovery problem. EvoPINN navigates a modular search space by decoupling neural representations from training programs, utilizing an LLM agent to iteratively propose memory-conditioned programmatic modifications. To ensure scientific validity, all candidates undergo strict structural verification and budget-matched PDE evaluation. Extensive experiments across diverse PDE regimes (oscillatory, elliptic, dissipative, and nonlinear transport) demonstrate that EvoPINN discovers PDE-specialized learning algorithms that significantly reduce relative $L_{2}$ error compared to baselines. Crucially, EvoPINN autonomously invented SLRC-PINN, a novel architecture whose performance gains persist under rigorous parameter-matched comparisons, establishing the viability of execution-grounded agents for discovering genuinely new scientific computing mechanisms.", "url": "https://wpnews.pro/news/evopinn-agentic-discovery-of-executable-algorithms-for-physics-informed-neural", "canonical_source": "https://arxiv.org/abs/2607.26490", "published_at": "2026-07-31 04:00:00+00:00", "updated_at": "2026-07-31 04:29:07.482946+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-agents", "ai-research"], "entities": ["EvoPINN", "SLRC-PINN", "Large Language Models (LLMs)", "Physics-informed neural networks (PINNs)"], "alternates": {"html": "https://wpnews.pro/news/evopinn-agentic-discovery-of-executable-algorithms-for-physics-informed-neural", "markdown": "https://wpnews.pro/news/evopinn-agentic-discovery-of-executable-algorithms-for-physics-informed-neural.md", "text": "https://wpnews.pro/news/evopinn-agentic-discovery-of-executable-algorithms-for-physics-informed-neural.txt", "jsonld": "https://wpnews.pro/news/evopinn-agentic-discovery-of-executable-algorithms-for-physics-informed-neural.jsonld"}}