Development & Applications of Density Functional Methods

杜克大学的杨伟涛在化学系化学理论、模型和计算方法项目的支持下，开发新的电子结构方法。电子结构计算提供了有关分子结构和分子相互作用的基本信息，在科学和工程中有着广泛的应用。密度泛函理论(DFT)因其计算效率高、精度高等优点，成为大系统计算的首选方法。然而，在许多重要的应用中，当前的DFT方法并不能给出准确的结果。这项建议的目的是建立在最近的洞察力和进展的基础上，以发展下一代密度泛函近似，以准确和有效地确定在生物、化学、物理、工程和纳米科学技术中发现的大系统的电子结构。通过不同的理论策略，研究工作将集中于探索用分数电荷和自旋近似精确条件所需的泛函形式，遵循两个互补的方向：(1)从第一原理多体理论构造近似，特别是通过配对矩阵涨落公式；(2)将已知的精确条件以分数变量的形式施加到常用泛函上。所开发的泛函在理论上是合理的，并且具有最小的离域和静态相关误差。对于广泛的应用，它们将是准确的，并将DFT的前沿推向对热化学、反应动力学、电荷转移、有机和过渡金属分子到块状材料的正确描述。研究方法将建立在理论和物理推导的基础上，使用最少的参数。拟议的工作应有助于理论和计算化学的进步，使其成为与其他更传统的化学领域平等的合作伙伴。

Weitao Yang of Duke University is supported by the Chemical Theory, Models and Computational Methods program in the Chemistry Division to develop new electronic structure methods. Electronic structure calculations provide fundamental information about molecular structure and molecular interactions, with broad applications in science and engineering. Because of its efficiency and its good accuracy for many applications, density functional theory (DFT) is the method of choice for such calculations for large systems. There are, however, many important applications where current DFT methods do not give accurate results. This proposal aims to build on the insight and progress made recently in order to develop the next-generation density functional approximations for accurate and efficient determination of electronic structure of large systems that are found in biology, chemistry, physics, engineering and nano-science and technology.With various theoretical strategies, the research effort will focus on exploring the functional forms needed to approximate exact conditions with fractional charges and spins, following two complementary directions: (1) Constructing approximations from first-principles many-body theory, particularly via the pairing matrix fluctuation formulation and (2) Imposing known exact conditions in terms of fractional variables on commonly-used functionals. The functionals developed will be theoretically sound, and have minimal delocalization and static correlation errors. They will be accurate for a broad range of applications and advance the frontiers of DFT towards the correct descriptions of thermochemistry, reaction kinetics, charge transfer, organic and transition metal molecules to bulk materials. The research approach will be based on derivation from theory and physics, using the minimal number of parameters . The proposed work should contribute to the progress of theoretical and computational chemistry in becoming an equal partner with other more traditional fields of chemistry.