| Excitation of an atom or molecule can occur when it is exposed to light or heat,causing one or more electrons to transition to a higher energy level than the ground state.These excited states have significant importance in various fields including chemistry,biology,and material science.Applications such as photosynthesis,optoelectronic materials,solar cells,quantum computing,and photocatalytic reactions are just a few examples of areas where excited states play a critical role.However,many important processes involving excited states cannot be observed directly.Even when the spectra of the target states are experimentally measurable,quantum chemical calculations are still required to interpret the experimental data.Thus,the development of accurate theoretical methods to predict spectroscopic properties of excitedstate species is crucial in enhancing our understanding of reactions involving excited states.By elucidating the mechanism of these reactions,we can design experiments and promote advancements in related fields.Electron correlation is a central topic of quantum chemistry that has posed challenges in accurately describing strongly correlated systems,including excited states.Conventional quantum chemistry methods have struggled to achieve a balanced treatment of static and dynamic correlations or have required extensive computational resources,such as diagonalizing large matrices.To address these challenges,Liu and his colleagues have developed a series of quantum chemistry methods within the restricted Static-Dynamic-Static(SDS)framework,such as iCI,iCIPT2,XSDSCI,SDSCI,XSDSPT2,SDSPT2,iCISCF(2),iCAS,iVI,and more.Among these methods,the variational SDSCI method provides a balanced treatment of static and dynamic correlations while only requiring the construction and diagonalization of a small 3Np× 3Np matrix to obtain Np electronic states.The SDSPT2 method,which is an approximation to SDSCI,offers more accurate results than conventional multireference perturbation theory(MRPT)methods when dealing with systems with quasi-degenerate states.Additionally,the iterative CI(iCI),selected CI,and second-order Epstein-Nesbet perturbation theory have been combined to develop the near-exact FCI method iCIPT2.Although these methods have been developed and benchmarked,their performances have not been thoroughly investigated.Therefore,in this thesis,we present a systematic study of several methods in the SDS family for studying excited states.We compare their performance with various state-of-the-art methods,such as MS-CASPT,MS-NEVPT2,EOM-CCSD,CC3,etc.,in determining vertical excitation energies in both closed-shell and open-shell systems.The main research objectives of this thesis are summarized as follows.(1)We firstly evaluated the performance of various quantum chemistry methods,namely MS-NEVPT2,SDSPT2,SDSCI,and ic-MRCISD,in predicting vertical excitation energies for 28 medium-sized organic molecules from Thiel’s test set.Our results indicate that the SDSCI offers high accuracy,closely matching the excitation energies generated by ic-MRCISD for 236 valence excited states.Among the perturbative methods,SDSPT2 outperforms MS-NEVPT2 and MS-CASPT2.Differences between the results by ic-MRCISD and CC3 methods are observed for certain excited states in the test set.We have employed iCIPT2 method,which can accurately reproduce experimental results after taking into account basis set effects,to obtain approximate FCI results for the five molecules showing significant deviations.Using the iCIPT2 results as a benchmark,we determined that CC3 has higher accuracy than ic-MRCISD for valence-excited states of closed-shell molecules.(2)To further validate the performance of multireference methods,namely MS-NEVPT2,SDSPT2,SDSCI,and ic-MRCISD,in open-shell systems,we constructed a test set comprising of 24 medium-sized organic cations based on Thiel’s test set.Our results show that the theoretical vertical ionization potentials obtained using the above methods aligne closedly with experimental values,and these methods maintain high accuracy in calculating 222 excited states of organic cations as well.These findings indicate that the SDSCI and SDSPT2 methods are suitable for studying excited states of open-shell systems.(3)Finally,we conducted a comprehensive examination of the iCIPT2 approach by generating 150 FCI-level vertical excitation energies for 24 small-sized radicals in the QUEST#4 test set.The excitation energies predicted by iCIPT2 are highly consistent with those obtained using the CIPSI method of the same category,and closely aligns with experimental data.We further investigated the accuracy of X-TDA,MS-NEVPT2,SDSPT2,SDSCI,and ic-MRCISD methods for both valence and Rydberg excitation energies of 24 open-shell systems.Our findings suggest that multireference methods such as ic-MRCISD,SDSCI,SDSPT2,and MS-NEVPT2 exhibit superior performance when compared to EOM/LR-UCC3 and are comparable to that of EOM/LR-UCCSDT. |