Recent Implementations in Kylin 1.3: Improved Computational Efficiency of ab initio DMRG and a Spin-adapted Version of EC-MRCI^†

Accurate evaluation of electron correlations is essential for the reliable quantitative description of electronic structures in strongly correlated systems, including bond-dissociating molecules, polyradicals, large conjugated molecules, and transition metal complexes. To provide a user-friendly tool for studying such challenging systems, our team developed Kylin 1.0 J. Comput. Chem. 44, 1316 (2023), an ab initio quantum chemistry program designed for efficient density matrix renormalization group (DMRG) and post-DMRG methods, enabling high-accuracy calculations with large active spaces. We have now further advanced the software with the release of Kylin 1.3, featuring optimized DMRG algorithms and an improved tensor contraction scheme in the diagonalization step. Benchmark calculations on the Mn₄CaO₅ cluster demonstrate a remarkable speed-up of up to 16 fater than Kylin 1.0. Moreover, a more user-friendly and efficient algorithm J. Chem. Theory Comput. 17, 3414 (2021) for sampling configurations from DMRG wavefunction is implemented as well. Additionally, we have also implemented a spin-adapted version of the externally contracted multi-reference configuration interaction (EC-MRCI) method J. Phys. Chem. A 128, 958 (2024), further enhancing the program’s efficiency and accuracy for electron correlation calculations.