| Abstract |
The accurate modelling of quantum excited-state dynamics is imperative for the development of new materials, light-based medicines (e.g., photodynamic therapy), and energy technologies (e.g., solar panels). However, the traditional ways that these are modelled, which are quantum calculation-based methods, are computationally expensive and prohibit their usage for large-scale applications. This project proposes an AI-based, data-driven multi-fidelity framework that addresses current issues by combining physics-informed surrogate modelling with geometry-aware machine learning to efficiently predict vertical excitation energies, revolutionising scalable light–matter modelling for renewable energy technologies, sustainable materials, photodynamic cancer therapies, and next-generation optoelectronic devices. |
