Presenting at ICML 2024

Karan Shah’s recent work on accelerating electron-dynamics simulations with neural networks was accepted as a poster at the ICML 2024 workshop “AI for Science: Scaling in AI for Scientific Discovery”, held on 26 July in Vienna, Austria. The method uses an autoregressive neural operator — trained with physics-based constraints and detailed data — to predict how electrons evolve in laser fields faster than conventional time-dependent density functional theory. Proof-of-concept tests on toy models were successful, suggesting the approach could ease real-time modelling of how molecules and materials respond to lasers; see the preprint.