Advancing large-scale excited-state materials simulations for quantum technologies

Understanding and predicting electronic excited-state properties is crucial for the design of novel materials for quantum technologies. We have developed a powerful and versatile suite of first-principles excited-state simulation codes implementing (spin-flip) time-dependent density functional theory (TDDFT) [1] with spin-orbit coupling, many-body perturbation theory (GW-BSE) [2], and quantum defect embedding theory [3] for multi-configurational systems. Analytical excited-state forces are derived within hybrid TDDFT and GW-BSE to obtain accurate excited-state geometries that are essential for modeling emission processes. Our formulations circumvent the computational bottleneck associated with the slowly converging summation over empty states, and employ low-rank approximations, localization techniques, and machine learning interatomic potentials to achieve further acceleration without compromising accuracy. The unique integration of multiple state-of-the-art excited-state methods and the excellent scalability of the codes to thousands of graphics processing units have enabled a comprehensive computational investigation of the formation energies, electronic structures, optical cycles, and coherence times of a variety of spin defects, including nitrogen-vacancy centers interacting with each other or with nearby dislocations in supercells containing over one thousand atoms [4]. Our work opens new avenues for engineering solid-state spin qubits and demonstrates how advanced theoretical frameworks and high-performance computing can drive innovative scientific discoveries.
[1] Jin, Yu, et al. JCTC 19. 8689 (2023)
[2] Yu, Jin, et al. JCTC 20, 10899 (2024)
[3] Chen, Yu, et al. JCTC 21, 7797 (2025)
[4] Zhang, Yu, et al. arXiv:2507.12387 (2025)