Nuts and Bolts of DFT Calculations

Throughout Chapter 2, we deliberately ignored many of the details that are necessary in performing DFT calculations in order to illustrate some of the physical quantities that these calculations can treat. This state of affairs is a useful starting point when you are learning about these methods, but it is, of course, not a useful strategy in the longer term. In this chapter, we dig into some of the details that make the difference between DFT calculations that provide reliable physical information and calculations that simply occupy time on your computer. 

Density Functional Theory A Practical Introduction. By David S. Sholl and Janice A. Steckel Copyright 2009 John Wiley Sons, Inc. 

The concept of numerical convergence is quite separate from the question of whether DFT accurately describes physical reality. The mathematical problem defined by DFT is not identical to the full Schrodinger equation (because we do not know the precise form of the exchange-correlation functional). This means that the exact solution of a DFT problem is not identical to the exact solution of the Schrodinger equation, and it is the latter that we are presumably most interested in. This issue, the physical accuracy of DFT, is of the utmost important, but it is more complicated to fully address than the topic of numerical convergence. The issue of the physical accuracy of DFT calculations is addressed in Chapter 10. 

In this chapter, we first concentrate on what is required to perform well-converged DFT calculations. After all, we need to be able to confidently find precise solutions to the numerical problems defined by DFT before we can reasonably discuss the agreement (or lack thereof) between DFT results and physical reality. 

---