This question is inspired by this post. In the Kohn-Sham framework of density functional theory, the total energy is expressed as: $$E=E_{kin}^{non}+E_{ext}+E_{H}+E_{xc}$$ in which

  • The first term is the kinetic energy of the non-interacting system: $$E^{non}_{kin}=-\dfrac{1}{2}\sum_i^{n}\phi_i^*(\vec{r})\nabla^2\phi_i(\vec{r})$$

  • The second term is the external energy: $$E_{ext}=\int\phi^*(\vec{r})U_{ext}(\vec{r})\phi(\vec{r}) d\vec{r}=\int U_{ext}(\vec{r})\rho(\vec{r})d\vec{r}$$

  • The third term is the Hartree energy: $$E_H=\dfrac{1}{2}\int \int \dfrac{\rho(\vec{r})\rho(\vec{r}')}{|\vec{r}-\vec{r}'|}d\vec{r}d\vec{r}'$$

  • The last term is the exchange-correlation term needs to be approximated, such as LDA, which includes:

    • Exchange energy;
    • Correlation energy;
    • The kinetic energy difference between the interacting system and the non-interacting system;
    • The self-interaction error in the Hartree energy.

In this post, Phil Hasnip has claimed that including a Hubbard U potential within a Kohn-Sham density functional theory (DFT) calculation is to study a material for which you expect a significant self-interaction error.

So is there any other interesting physics contained in the exchange-correlation functional in the framework of KS-DFT?


1 Answer 1


You have already detailed what the physics contained in the exchange-correlation functional is.

Self-interaction error is not physical; it is an artifact in density functional approximations that arises from imperfect cancellation of the Coulomb and exchange interactions. Although several schemes have been suggested to remove self-interaction error, the one by Perdew and Zunger being the most well-known, these schemes tend to be problematic in that they are

  • hideously expensive computationally
  • hard to minimize properly (e.g. the Perdew-Zunger functional requires complex orbitals even for a gas-phase atom without external fields)
  • may break molecular symmetries

Another failure of density functional approximations is in modeling systems with strong correlations, such as many transition metal complexes and multiple bond breaking.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .