A Data-Efficient Machine-Learning Approach for Modeling the Photodynamics of All-trans Hexatriene based on Multireference Configuration Interaction Calculations

Key Points

• The research aims to improve simulations of the excited-state photodynamics of polyenes using a machine-learning approach.
• Utilized machine-learning algorithms to approximate multireference configuration interaction calculations
• Focused on modeling the coexistence of ionic and covalent states
• Compared performance against traditional electronic-structure methods.
• Achieved significant reduction in computational costs while maintaining accuracy of photodynamic simulations
• Success in modeling excited-state behavior of all-trans hexatriene
• Performance improvements demonstrate potential for broader applications in molecular photodynamics.