There are multiple types of functionals and basis sets that can be used for different calculations in Gaussian such as optimizations, scans, and excited state energy calculations. A basis set is a set of basis functions. Each basis set is a different size and generally, the bigger the basis set size, the more accurate the results will be. The names of the basis sets accessible through Gaussian are 6-31G (which can include +,++, and different orbitals), STO-3G, 3-21G, 6-311G, cc-pVDZ, cc-pVTZ, cc-pVQZ, LanL2DZ, LanL2MB, SDD, DGDZVP, DGDZVP2, DGTZVP, GEN, and GENECP. However, really there are many more options available which are discussed more thoroughly on the following website (http://www.gaussian.com/g_tech/g_ur/m_basis_sets.htm). It is also possible to create your own basis set using Gaussian, but this can be time-consuming and complicated. In relation to 6-31G, the increasing size of the basis set in terms of +, ++, aug- (which are augmented basis sets) and p,d,f orbitals or *,** (polarization functions), the more that are included, the more accurate results these should be as well. Each basis set contains a different number of Cartesian (etc) basis functions, which can be found in the output file (ctrl-f “basis function”). The larger the number of basis functions corresponds to a longer calculation time.
I performed an optimization calculation on a new conformation of tryptophan and then ran excited state calculations using 16 combinations of functionals (b3lyp, cam-b3lyp, pbepbe, and wb97xd) and basis sets (6-31G, 6-31+G, 6-31+G(d,p), and cc-pVDZ). Since 6-31G is the smallest basis set here, it took the shortest time to complete calculations in all of the functionals. Also, within functionals, cc-pVDZ is similar in time to 6-31G. Below is a table showing the times and number of basis functions for each basis set that was used in calculating excited state energies of an optimized configuration of tryptophan.
A) Basis set: 6-31G
Cartesian Basis functions: 159
| Functional | b3lyp | cam-b3lyp | PBEPBE | wB97XD |
| Job CPU Time / Minutes | 7.733 | 9.717 | 7.45 | 10.1 |
B) Basis set: 6-31+G
Cartesian Basis functions: 219
| Functional | b3lyp | cam-b3lyp | PBEPBE | wB97XD |
| Job CPU Time / Minutes | 25.88 | 34.63 | 20.5 | 34.65 |
C) Basis set: 6-31+G(d,p)
Cartesian Basis functions: 345
| Functional | b3lyp | cam-b3lyp | PBEPBE | wB97XD |
| Job CPU Time / Minutes | 59.53 | 76.0167 | 48.05 | 81.783 |
D) Basis set: cc-pvdz
Cartesian basis functions: 285
| Functional | b3lyp | cam-b3lyp | PBEPBE | wB97XD |
| Job CPU Time / Minutes | 29.0167 | 39.267 | 24.6 | 39.03 |
To be nitpick, those basis sets you listed are just the ones listed on the GaussView pulldown menu. There are hundreds of basis sets available in gaussian, and you can even define your own if you are dedicated. You could link to the gaussian documentation webpage that describes basis sets, as a reference.
The terminology is confusing, but to be clear:
Using +, ++ or aug- essentially mean the same thing — called ‘augmented’ basis sets.
Using *, ** or (p,d) (2p,d,f) etc mean the same thing — called ‘polarization functions’.
Your post would be really sharp with a brief summary of some of the data you have — for Trp and each of several example basis sets, how many basis functions are there and how long does a particular sample calculation take with that set? You should be able to make a html table in the blog post.
Thanks for the updates. The table looks sharp! However its not clear that you are testing four different functionals (which ones go with which column)?
Also its redundant to list the name of the basis set and the number of basis functions for each column — that only needs to be listed once for each subtable. (I’m being nitpick with data presentation here — the data you’ve presented is surely correct)
Also in your text you flipped your references: the functionals are the wb97xd, etc. and the basis sets are the 6-31G, etc. in the part when you say “using 16 combinations of basis sets “… etc.