Performance and Properties of Density Functional Methods

An evaluation of the performance of the plethora of different functionals for a variety of properties is a major undertaking. We will here just quote two sets of results  

While results with the above basis sets are not converged to the basis set limit, the residual basis set errors are presumably well below the inherent errors in the functionals, and the performance thus reflects the quality of the exchange-correlation functionals (Table 6.2). 

Note that the performance ordering of the functionals is not the same for the two sets of results. Only a minor part of this discrepancy can be attributed to the difference in basis sets, the remaining discrepancy is due to differences in the data sets. 

Schrodinger equation (exactly) with a wave mechanics approach. Indeed, unless one believes that the Schrodinger equation contains superfluous information, this is likely to be the case. Since exact solutions are generally not available in either approach, the important question is instead what the computational cost is for generating a solution of a given accuracy. In this respect, DFT methods have very favourable characteristics. 

Despite the many successes of DFT, there are some areas where the current functionals 

---