Madriaga Jose P, Kodrycka Monika, Crawford T Daniel
Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061, United States.
J Phys Chem A. 2025 May 22;129(20):4601-4610. doi: 10.1021/acs.jpca.5c01617. Epub 2025 May 9.
Reduced-scaling approaches have yielded significant improvements in the computational efficiency of coupled cluster methods, making them more feasible for studying large molecules. In this work, we extend the use of pair natural orbitals (PNOs) to frequency-dependent quadratic response properties. We evaluate the performance of PNOs alongside methods optimized for response properties that derive from an approximate field-perturbed density matrix known as perturbation-aware PNOs (PNO++). Additionally, we concatenate the PNO and PNO++ spaces to obtain the combined-PNO++ method, which is tailored to simultaneously maintain the accuracy of the CCSD correlation energies and response properties. We analyze the truncation errors associated with these methods using first electric dipole hyperpolarizability - specifically the average second-harmonic generation and optical refractivity, using canonical coupled cluster singles and doubles (CCSD) as a reference. The performance analysis of the PNO family provides valuable insights into the viability of implementing CCSD quadratic response properties at a full-production level, highlighting which techniques may yield the most successful results.
缩尺度方法在耦合簇方法的计算效率方面取得了显著提升,使其在研究大分子时更具可行性。在这项工作中,我们将对自然轨道对(PNO)的应用扩展到频率相关的二次响应性质。我们将PNO的性能与针对响应性质进行优化的方法一起评估,这些方法源自一种称为微扰感知PNO(PNO++)的近似场微扰密度矩阵。此外,我们将PNO和PNO++空间连接起来以获得组合PNO++方法,该方法旨在同时保持CCSD相关能和响应性质的准确性。我们使用第一电偶极超极化率,特别是平均二次谐波产生和光学折射率,以正则耦合簇单双激发(CCSD)作为参考,分析与这些方法相关的截断误差。PNO系列的性能分析为在全生产水平上实现CCSD二次响应性质的可行性提供了有价值的见解,突出了哪些技术可能产生最成功的结果。