Multiscale Dynamic Material Modeling, Sandia National Laboratories, Albuquerque, New Mexico, USA.
J Comput Chem. 2010 Jul 15;31(9):1860-3. doi: 10.1002/jcc.21472.
We present results that compare the accuracy of the AM05 density functional (Armiento and Mattsson, Phys Rev B 2005, 72, 085108; Mattsson et al., J Chem Phys 2008, 128, 084714) to a set of chemical reaction energies. The reactions were generated from the singlet species in the well-known G2 test suite (Curtiss et al., J Chem Phys 1991; Curtiss et al., J Chem Phys 1997; 106, 1063). Our results show that, in general, the AM05 functional performs nearly as well as the other "pure" density functionals, but none of these perform as well as the hybrid B3LYP functional. These results are nonetheless encouraging because the AM05 functional arises from very simple assumptions, and does not require the calculation of the Hartree-Fock exchange integrals.
我们呈现了比较 AM05 密度泛函（Armiento 和 Mattsson，Phys Rev B 2005，72，085108；Mattsson 等人，J Chem Phys 2008，128，084714）与一组化学反应能准确性的结果。这些反应是从著名的 G2 测试套件中的单重态物种中生成的（ Curtiss 等人，J Chem Phys 1991； Curtiss 等人，J Chem Phys 1997；106，1063）。我们的结果表明，一般来说，AM05 泛函的性能几乎与其他“纯”密度泛函一样好，但没有一个泛函的性能比得上混合 B3LYP 泛函。然而，这些结果令人鼓舞，因为 AM05 泛函源于非常简单的假设，并且不需要计算 Hartree-Fock 交换积分。
