Double-zeta plus polarization applications

Although there is no strict relationship between the basis sets developed for, and used in, conventional ah initio calculations and those applicable in DFT, the basis sets employed in molecular DFT calculations are usually the same or highly similar to those. For most practical purposes, a standard valence double-zeta plus polarization basis set (e.g. the Pople basis set 6-31G(d,p) [29] and similar) provides sufficiently accurate geometries and energetics when employed in combination with one of the more accurate functionals (B3LYP, PBEO, PW91). A somewhat sweeping statement is that the accuracy usually lies mid-way between that of M P2 and that of the CCSD(T) or G2 conventional wave-function methods. 

The factorial growth of the FCI treatment with basis set and number of electrons clearly limits its applicability. However, FCI calculations can be performed in moderate-sized (double-zeta plus polarization (DZP) or better) basis sets to calibrate approximate methods of including electron correlation. Thus there is considerable incentive to utilize these FCI methods and the capabilities of modern supercomputers to perform large-scale benchmark calculations. In this section we consider several examples of the insight obtained from FCI benchmark calculations. In the sections dealing with applications, we consider additional FCI benchmark calculations in conjunction with specific applications. 

The full Cl is impractical except as a reference because the number of determinants is asymptotically N . That means its application is possible only for small molecules and small basis sets. In praaice, calculations using 25 x 10 determinants for H O in a double zeta plus polarization (DZP) basis have been made and —10 determinants for CH3 in a triple zeta plus polarization (TZP) basis. ... 

Full configuration interaction (FCI) by definition gives the exact n-particle energy within the given basis set. (Since it is the exact solution, this happens irrespective of the quality of the zero-order wave function.) Because its computational requirements ascend factorially with the size of the system, application to practical systems using one-particle basis sets of useful size will be essentially impossible for the foreseeable future. Even using the fastest available computational hardware and parallelized codes, an FCI calculation on H2O in a double-zeta plus polarization basis set is about the state of the art at present. ... 

---