Multi-scale Machine Learning (D2.3)

The goal of Project D2.3 is to overcome the transferability bottleneck in multiscale simulations by using machine-learning models to translate information consistently across time and length scales. By fitting high-dimensional effective parameters to experimental data and fine-grained simulations, we will build ML surrogates that reproduce key observables on coarser levels, enabling predictive digital blueprints for laser 3D nanoprinting. This approach will allow us to capture the impact of process parameters such as quencher diffusion and surface roughness, going far beyond current simulation capabilities and providing an inverse-design framework for optimizing complex printing processes.