In-context modeling as a retrain-free paradigm for foundation models in computational science

2026-04-24

AI Summary

This work addresses the challenge in computational science that models often fail to generalize across diverse physical systems and require repeated retraining. To overcome this, the authors propose In-Context Modeling (ICM), which reframes physical modeling as a context-based inference paradigm that eliminates the need for retraining. ICM encodes observed fields into a physical context and infers system-specific physical relationships in a single forward pass. It seamlessly integrates into finite element frameworks by combining unlabeled physics-informed training—guided by governing equations—with a context embedding mechanism. Experiments on hyperelasticity demonstrate ICM’s ability to generalize across varying materials, geometries, and loading conditions. Moreover, its performance scales significantly with increased data diversity and computational resources, exhibiting favorable scaling properties akin to those of foundation models.

Technology Category

• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Diffusion Models for Vision

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Building models that generalize across physical systems without retraining remains a central challenge in computational science. Here we introduce In-Context Modeling (ICM), a retrain-free paradigm that infers physical relationships directly from observational fields. Rather than encoding system-specific behavior in fixed parameters, ICM assimilates measurements as physical context and performs inference through a single forward pass. Trained in a physics-informed, label-free manner using governing equations, a single model generalizes across unseen materials, geometries, and loading conditions. Demonstrated on hyperelasticity, ICM integrates with finite-element simulations and is validated using experimental full-field measurements. Moreover, performance improves with increasing data diversity and computational budget, exhibiting favorable scaling behavior analogous to foundation models. By recasting physical modeling as in-context inference, this work establishes a transferable paradigm for retrain-free scientific learning and a foundation for scalable modeling across computational science.

Problem

• foundation models
• retrain-free
• physical systems
• generalization
• computational science

Innovation

• In-Context Modeling
• retrain-free
• physics-informed
• foundation models
• computational science