Rationale for mixing exact exchange with density functional approximations

Rationale for mixing exact exchange with density functional approximations
复制标题

DOI:
10.1063/1.472933
复制
发表时间:
1996-12-08
影响因子:
4.4
通讯作者:
Burke, K
Burke, K
中科院分区:
化学2区
文献类型:
--
作者:
Perdew, JP;Ernzerhof, M;Burke, K

文献摘要

被引文献

相似文献

电子系统的交换相关能量 E(XC)(DFA) 的密度函数近似通常通过混合一些精确的交换 E(X) 来改进:E(XC) 近似于 E(XC)(DFA) + (1/n)(E(X) - E(X)(DFA))。当 E(XC)(DFA) 中的误差源自 lambda = 0 或耦合常数积分积分(0)(1) d lambda E(XC lambda)(DFA) 的交换端时,此过程是合理的。我们认为最佳整数 n 大约是 Gorling-Levy 微扰理论的最低阶,该理论提供了 0 小于或等于 lambda 1 范围内耦合常数依赖性 E(XC,lambda) 的真实描述,因此典型分子的原子化能 n 近似于 4。我们还提出将 n 连续概括为相关强度的指数,以及二阶微扰理论与广义梯度近似的可能混合。 (C) 1996 年美国物理研究所。
Density functional approximations for the exchange-correlation energy E(XC)(DFA) of an electronic system are often improved by admixing some exact exchange E(X) : E(XC) approximate to E(XC)(DFA) + (1/n)(E(X) - E(X)(DFA)). This procedure is justified when the error in E(XC)(DFA) arises from the lambda = 0 or exchange end of the coupling-constant integral integral(0)(1) d lambda E(XC lambda)(DFA). We argue that the optimum integer n is approximately the lowest order of Gorling-Levy perturbation theory which provides a realistic description of the coupling-constant dependence E(XC,lambda) in the range 0 less than or equal to lambda 1, whence n approximate to 4 for atomization energies of typical molecules. We also propose a continuous generalization of n as an index of correlation strength, and a possible mixing of second-order pertubation theory with the generalized gradient approximation. (C) 1996 American Institute of Physics.