Speaker
Description
The effective design of new materials for sustainable energy conversion can be facilitated by the accurate prediction of electronic properties with moderate computational complexity and cost. The self-interaction error (SIE) of Kohn-Sham density functional theory (KS-DFT) is a non-physical, non-linear dependence of an orbital's energy on its own fractional occupation [Dabo et al., Phys. Rev. B, 82:115121, 2010]. The generalized Koopmans condition (GKC) ensures an atomic orbital's eigenenergy is invariant with its own fractional occupation, and is free of self-interaction. The constrained search of the Levy spin-density-functional theory yields orbital densities and delivers the ground-state energy in accordance with the Hohenberg-Kohn theorems. In this work, the electron density of the Kohn-Sham equations is thereby constrained to be orbital-density dependent, with a total energy functional linear with respect to variation of its orbital densities. That is, the KS multiplicative effective potential $v_{\text{s}}[{n}]$ for an orbital $\varphi_i[{n}]$ is constrained to a functional $v_{\text{eff}}[{n - n_i}]$. The result complies with the generalized Koopmans condition (GKC). Preliminary calculations show fundamental band gaps with an accuracy comparable to the $G_0W_0$ approximation of many-body perturbation theory (MBPT), with a level of complexity comparable to KS-DFT. In practice, software codes can be combined to remove SIE and to model composite material properties. With the subtraction of non-physical electron self-interaction from the Kohn-Sham equations, GKC-DFT can possibly improve computational efficiency, reduce the complexity of highly accurate DFT simulations, and facilitate the development of new applications.
