The utility questbas
This utility is available in $ADFBIN/questbas.exe. Running this executable results in the following questions:
Please provide the atomic number of the element
For Carbon, we type 6
As a next question a long list of default basis sets is presented. These correspond to predefined choices for the number of diffuse, tight, and polarization functions, and are used to automatically generate all kinds of ET basis sets. If any of these standard basis sets are chosen, the answers to all questions are filled in automatically.
Choose 0 for the general basis set procedure
SUGGESTED CHOICE: 0
Here, we follow the suggested choice which allows us to explain all options, and type 0.
Now the script tells us we have the choice between 4 different basis sets for the occupied shells. The basis sets which have been most thoroughly investigated at present are the ET basis sets in which the factor beta (β) [defining the spacing between two subsequent STOs] is fixed at the value 1.7 for all l-values in the basis. These basis sets form the starting point for the directories VI and VII. For carbon, the output tells us that we would start from a 5s4p basis with this choice. In another ET basis set, with a different design philosophy, β is variable (typically 1.55 for s, 1.7 for p, 2.0 for d), but α is fixed. These basis sets occur in a large (valence quadruple zeta, core double zeta) and medium (augmented double zeta) variety. In the case of carbon, they are of the size 6s4p and 4s3p. These variable β basis sets have been less intensively tested in molecular calculations than the fixed β variety. For that reason a final verdict has to be postponed as to which type is preferable. The fixed β seems the safest choice at present, because it has been used more in molecular calculations. The large variable β basis leads to names like VIB and VIIB. We type here 1 for the fixed β variety. As output we now get information on our intermediate basis set
As a starting point you have selected an even-tempered basis with following values for n, alpha, beta:
S: 6 0.6592211993160408 1.700000047683716
P: 4 0.5316027776326796 1.700000047683716
The script follows with the following question:
You can specify here how many additional fus you want for each l-value. Suggested: 0 0 0 0
This option allows the user to add functions to the final basis by keeping the endpoints, most diffuse and tighest functions, fixed and reducing the β value. This option is not often useful and we ignore it by typing 0 0 0 0
The next question concerns the ZORA option. As this has not yet been thoroughly tested we choose 0 for a nonrelativistic basis set.
The next question is if we want to add functions. This is definitely needed, because the basis without polarization functions is quite poor, so we type 1.
Then we are asked how many polarization functions should be added. As a default is suggested 0 2 1, meaning 2d and 1f polarization functions. This means that we get more polarization functions than in basis V if we use the default. The geometric mean of the exponent of the polarization functions is stored inside the program in a data statement. This gives the 'best' exponent for a single polarization function. If more than polarization function is used, this will be the middle of these polarization functions. The beta value for the polarization function is taken identical to the beta value of the highest occupied l-value.
After typing the suggested default of 0 2 1, we are asked if we want to add diffuse functions. Whether such functions are needed in your application is discussed elsewhere in this document. Let us add one diffuse p function by typing 0 1 0 0
We are then asked about additional tight functions to improving the description of the core region. In this example, we do not add such functions by typing 0 0 0 0
Then we are asked what size of fit set we require. We type 3 to get the largest fit set, the only thoroughly tested option.
Finally we are asked if we want to let the program add diffuse function to make the basis set suitable for (hyper)polarizability calculations. We follow the suggested choice by typing 0 0. Otherwise, the utility would add diffuse functions until the most diffuse function is more diffuse than a stored default diffuse value (provided by Prof. Del Chong, based on field-induced polarization functions).
The output of the questbas utility is given in the (local) file raf_in. This file gives the input for the rafbas utility described next.
