Ab initio calculations of molecular properties using the orbital optimized random-phase approximation method

在这个项目中，我将开发，实施和使用一种无参数的量子力学模拟方法，以负担得起的计算成本，用于分子性质的化学准确描述。该方法是基于随机相位近似（RPA）的相关能量。与以往的RPA方法相比，新的方法是独立于传统的半局域密度泛函理论（DFT）。新方法代表了Kohn-Sham/Hartree-Fock方案的推广，并且是精确描述弱化学键的有前途的工具，例如，远程色散相互作用，以及隔离器，半导体和金属类化合物的相关性。使用RPA，能量包含完全Fock交换和基于RPA的相关能量。与半定域密度泛函理论中的动能一样，交换和关联也用轨道表达式表示。轨道构成了分子基态的电子密度。在这个框架内的RPA相关能量计算使用有效的线性响应技术借用时间依赖DFT。这与单参考后Hartree-Fock方法相反，该方法利用属于电子激发到虚轨道的行列式。因此，RPA是精确波函数技术和基于DFT的合理近似的组合。到目前为止，轨道是由半局域DFT确定的。这些轨道不属于RPA能量超曲面的最小值。这使得在RPA水平上计算分子性质变得困难。因此，轨道应该从基于RPA的总能量的变化中推导出来。这将是第一次在这个拟议的项目中完成，并将导致一种称为轨道优化RPA的新方法。它将被用来预测的平衡结构和电子激发光谱的分散束缚的配合物，偶极束缚的阴离子，和金属簇有和没有吸附的分子。

In this project, I will develop, implement, and use a parameter-free quantum mechanical simulation method with affordable computational cost for chemically accurate description of molecular properties. The method is based on the random phase approximation (RPA) for correlation energy. In contrast to previous RPA methods, the novel approach is independent of conventional semi-local density functional theory (DFT). The new method represents a generalization of the Kohn-Sham/Hartree-Fock scheme and is a promising tool for accurate description of weak chemical bonding, e.g., long-ranged dispersion interactions, and for correlation in isolators, semiconductors, and metal-like compounds. Using RPA, the energy contains the full Fock exchange and the RPA-based correlation energy. Like the kinetic energy in semi-local DFT, exchange and correlation are represented by orbital expressions. The orbitals construct the electron density of the molecular ground state. Within this framework the RPA correlation energy is calculated using efficient linear response techniques borrowed from time-dependent DFT. This is in contrast to single reference post-Hartree-Fock methods which make use of determinants belonging to electron excitations into virtual orbitals. Hence, RPA is a combination of exact wavefunction techniques and reasonable approximations based on DFT. So far, the orbitals are determined by semi-local DFT. These orbitals do not belong to the minimum of the RPA energy hypersurface. This makes calculations of molecular properties difficult at the RPA level. Thus, the orbitals should be derived from variation of the RPA-based total energy. For the first time, this will be done in this proposed project and will lead to a new method called orbital optimized RPA. It will be used to predict equilibrium structures and electronic excitation spectra of dispersion-bound complexes, dipole-bound anions, and metal clusters with and without adsorbed molecules.