Project 6: Development of NN/MM algorithms for excited-state dynamics, incorporating solvent effects in applications such as multi-chromophoric systems and complex excited-state processes

2. March 2026

Machine learning models have proven highly successful in predicting ground-state potential energy surfaces (PES) of organic molecules and inorganic solids. In contrast, the description of excited-state PES remains a major challenge and has so far been limited to very small molecular systems. In particular, the combination of delta machine learning with accurate but computationally expensive ab-initio quantum chemistry and efficient semiempirical DFTB calculations has not yet been systematically explored for this purpose.

The goal of this project is to develop an efficient computational workflow for the simulation of excited-state PES, using retinal as a representative photo-biological system and azobenzene as a simpler model compound. The approach integrates three levels of theory: reliable ab-initio quantum chemistry, significantly faster DFTB calculations, and machine learning models. Active learning strategies will be employed to sample the relevant conformational space and to identify molecular geometries that require high-level ab-initio calculations. Based on these data, delta-ML models will be trained to correct DFTB predictions. The resulting bias-corrected DFTB surfaces will then be used to perform semi-classical surface-hopping simulations to explore excited-state dynamics.

This project aims to establish a transferable multiscale modeling framework that enables accurate and computationally efficient simulations of photochemically and photobiologically relevant molecules. By using semiempirical methods as an intermediate layer between ab-initio calculations and machine learning, the approach represents an important step toward the simulation of larger, more complex light-responsive biological systems.

Team

Prof. Dr. Marcus Elstner

Principal Investigator (Karlsruhe Institute of Technology)

Phone: +49 721 60845705

More Information

Team

Prof. Dr. Marcus Elstner

Principal Investigator (Karlsruhe Institute of Technology)

Phone: +49 721 60845705

More Information

Team

Prof. Dr. Marcus Elstner

Principal Investigator (Karlsruhe Institute of Technology)

Phone: +49 721 60845705

More Information

Switch to the German homepage or stay on this page