4.7 Article

Performance and Self-Consistency of the Generalized Dielectric Dependent Hybrid Functional

期刊

JOURNAL OF CHEMICAL THEORY AND COMPUTATION
卷 13, 期 7, 页码 3318-3325

出版社

AMER CHEMICAL SOC
DOI: 10.1021/acs.jctc.7b00368

关键词

-

资金

  1. MICCoM, as part of the Computational Materials Sciences Program - U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division
  2. laboratory Directed Research and Development (LDRD) - Argonne National Laboratory
  3. U.S. DOE [DE-AC02-06CH11357]
  4. Office of Science of the U.S. Department of Energy [DE-AC02-05CH11231]
  5. Argonne Leadership Computing Facility
  6. DOE Office of Science User Facility [DE AC02 06CH11357]

向作者/读者索取更多资源

We analyze the performance of the recently proposed screened exchange constant functional (SX) (Brawand et al. Phys. Rev. X 2016, 6, 041002) on the GW100 test set, and we discuss results obtained at different levels of self-consistency. The SX functional is a generalization of dielectric dependent hybrid functionals to finite systems; it is nonempirical and depends on the average screening of the exchange interaction. We compare results for ionization potentials obtained with SX to those of CCSD(T) calculations and experiments, and we find excellent agreement, on par with recent state of the art methods based on many body perturbation theory. Applying SX perturbatively to correct PBE eigenvalues yields improved results in most cases, except for ionic molecules, for which wave function self-consistency is instead crucial. Calculations where wave functions and the screened exchange constant (alpha(sx)) are determined self consistently, and those where alpha(sx) is fixed to the value determined within PBE, yield results of comparable accuracy. Perturbative G(0)W(0) corrections of eigenvalues obtained with self-consistent alpha(sx) are small on average, for all molecules in the GW100 test set.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.7
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据