High-Performance Semiempirical Excited-State Molecular Dynamics: A Step Toward Data-Driven Photodynamics

We introduce excited-state molecular dynamics (ESMD) in PYSEQM, a GPU-accelerated semiempirical quantum chemistry engine implemented in PyTorch. The new module enables Born-Oppenheimer (single-surface) dynamics using configuration-interaction singles and random phase approximation for excited-states, allowing long trajectories and large statistical ensembles to be simulated efficiently on a single GPU. We also implement an extended-Lagrangian ESMD (XL-ESMD) scheme that propagates auxiliary electronic variables, enabling relaxed ground and excited-state convergence thresholds without compromising energy conservation. The ESMD implementation scales smoothly from small chromophores to a nearly 900-atom dendrimer (taking 6.5 s per ESMD step). PYSEQM also supports batched execution, allowing many geometries or trajectories to be evaluated in a single GPU launch, substantially increasing throughput and making ensemble-based protocols routine. As a demonstration, we compute absorption, emission, and infrared spectra from trajectories propagated on the ground and first excited states, illustrating practical utility of PYSEQM. The XL-ESMD scheme yields identical spectra at significantly lower computational cost, establishing the role of extended Lagrangian based dynamics for efficient ESMD simulations. Beyond raw performance, PYSEQM’s Pytorch foundation provides automatic differentiation for forces, efficient GPU batching, and seamless interfacing with machine learning models. These capabilities position PYSEQM as practical platform for machine learning-augmented excited-state dynamics and lay the foundation for future data-driven nonadiabatic ESMD modeling of ultrafast spectroscopic probes.