🌋 From volcanic processes to advanced glass manufacturing, viscosity is one of the key properties controlling the behavior of silicate melts — and predicting it across complex chemical systems is still an open challenge.
Last month in Murano, at Le mille vite del vetro – Bringing glass back home!, Michele presented our work on AI-driven prediction of silicate melt viscosity using Transformer architectures and multi-head cross-attention mechanisms.
Our approach models the entire temperature–viscosity evolution as a structured thermorheological sequence, conditioned on melt composition and query temperature, moving beyond traditional empirical formulations.
Almost contemporary, Eros presented a poster at the Artificial Intelligence for Advanced Materials (AI4AM) conference in Madrid (Spain), disclosing the work we are carrying on on the use of Physics-Informed Neural Networks (PINNs) for the understanding of nanoparticles crystallization kinetics. ⚛️
This will open up possibilities on the synthetic optimization and a priori prediction of nanoparticles dimensions, which is an important parameter for their properties.
A big thank to the organizers and everyone involved in these beautiful and inspiring events! 🔎