• This part of the school will equip early-career researchers with both conceptual foundations and practical tools to integrate AI rigorously with computational mechanics (CM):

• Critically assess when and why AI adds genuine value to CM workflows, and when it does not.

• Trace the evolution of physics-informed and hybrid AI approaches from theoretical grounding to working implementation.

• Deliver end-to-end case studies spanning structural diagnostics, materials design, and hybrid solver applications.

• Build hands-on PyTorch skills covering PINNs, physics-constrained CNNs, and generative models.

• Convene a panel discussion on the path forward for interpretable, scalable, and generalisable AI-CM integration

Day 1 — Foundations and Physics-Informed Learning

AI, Digital Twins and Computational Mechanics

Overview of AI-driven computational mechanics, digital twins, and hybrid physics–data approaches.

Why AI for Computational Mechanics?

Critical discussion of where AI adds value, limitations of purely data-driven models, and trustworthy AI integration.

Physics-Informed Learning and Operator Methods

Introduction to PINNs, DeepONets and operator learning for mechanics applications, including challenges and failure modes.

Hands-on Practical

Implementing PINNs in PyTorch using Google Colab, including training, loss balancing and convergence analysis.

Day 2 — Applications, Generative AI and Future Directions

AI-Enhanced Mechanics Case Studies

Digital twins, structural health monitoring, hybrid AI–solver coupling, and materials modelling applications.

Generative AI for Engineering Design

Diffusion models, physics-constrained generative AI, and inverse design workflows for materials and mechanical systems.

Hands-on Practical

Physics-constrained CNNs and generative AI workflows for material property prediction and engineering design.

Panel Discussion

Future directions of machine learning in computational mechanics with invited speakers from academia and industry.

All practical sessions run on Google Colab (no local installation required). Course materials will be released openly under a CC-BY licence after the course.