Efficient Implementation of A New and Accurate DFT Method

描述（由申请人提供）：密度泛函理论（DFT）可能是分子模拟中应用最广泛的量子化学方法，因为它能够准确有效地模拟各种分子系统。但是，在生物建模的两个主要方面，即化学反应的屏障（包括电荷转移和氢键）和疏水相互作用中突出的色散相互作用，还存在一些不足。最近，DFT的先驱之一Axel Becke和他的同事提出了两个概念新颖的功能模型来克服这些缺陷，他们的结果显示出很大的潜力。反应势垒的实空间相关（RSC）模型对现有的DFT数值算法提出了严峻的挑战。解决这一挑战也是有效利用色散模型的先决条件。RSC模型需要对每个网格点的Hartree-Fock （HF）交换能量密度进行评估，计算成本非常高。此外，它包含一个非光滑相关因子和一个辅助泛函的迭代解，从而防止了自洽域（SCF）解。在这个一期项目中，我们建议采用同一性分辨技术来降低RSC模型中hf交换能量密度的计算成本。我们将证明新算法的计算成本将降低到与传统DFT算法相同的水平。我们还将对辅助泛函进行解析插值，并平滑相关因子。所有这些发展将导致RSC模型的SCF解决方案，这在几乎所有DFT应用中都是需要的。这一新的发展将为新的精确DFT方法的推广应用提供条件，并使其进一步发展。

DESCRIPTION (provided by applicant): Density functional theory (DFT) is perhaps the most widely applied quantum chemistry method in molecular simulations due to its ability to accurately and efficiently model a wide range of molecular systems. Still, it has some deficiencies in two major aspects of biological modeling: the barriers of chemical reactions (including charge transfer and hydrogen bonding) and the dispersion interaction prominent in hydrophobicinteractions. Recently, Axel Becke, one of the pioneers of DFT, and his co-workers, have proposed two conceptually novel functional models to overcome the deficiencies and their results have shown a great deal of potential. The model for the reaction barrier, namely real-space correlation (RSC), imposes a serious challenge to the current DFT numerical algorithms. The solution to the challenge is also a prerequisite to the efficient exploitation of the dispersion model. The RSC model requires the evaluation of Hartree-Fock (HF) exchange energy density at each grid point, which incurs prohibitively high computational cost. Further more, it contains a non-smooth correlation factor and an iterative solution to an auxiliary functional, preventing a self-consistent-field (SCF) solution. For this Phase I project, we propose to apply the resolution-of-identity technique to reduce the cost of the calculation of the HF-exchange energy density in the RSC model. We will show that the computational cost of the new algorithm will be reduced to the same level as the conventional DFT algorithms. We will also develop an analytical interpolation to the auxiliary functional and smoothen the correlation factor. All those developments will lead to a SCF solution of the RSC model, which is needed in just about every application of DFT. The new development will afford the general application of the new and accurate DFT method and enable its further development.