PC GAMESS/Firefly-related discussion club

Learn how to ask questions correctly

Alex Granovsky
gran@classic.chem.msu.su

There are several possible ways how to implement spherical basis
sets in QC package. Firefly uses a kind of shift/projection technique.
This means that program does not explicitly work with spherical basis
functions - it still works with cartesian components but limits the
entire cartesian AO space to its pure spherical subspace.
More specificaly, all six cartesian d functions are required to form
five spherical d functions.

> And one more question: why energies (with and without "d5=1") are equal for K+?

Because ndfunc/nffunc do not necessary add extra basis functions...
as they are not defined for some basis sets and for some elements.
It is always a good idea to look into output file and check whether
they were actually added.


>P.S. It is a pity that on page related to this option (http://classic.chem.msu.su/gran/gamess/spherical.html) there is no warning of this stunning fact…

It is supposed that everybody who uses this option knows its meaning
and the difference between two types of basis sets.

>P.P.S. Theoretical group analysis is always based on existence of five d-AO (in books for chemists). The spectral data too specify existence of five d-AO. Where to extract energy of these five AO?

They are the energies of the d-type MOs...


>P.P.S. Please, excuse me for verbosity...  Just I've looked two my calculations: optimization of a complex 6Na*Nb(5)Cl6 with and without "d5=1" option. In the first case calculation converges, and in the second - 6th cation goes away, i.e. the complex keeps 6 cations in the first case, and isn't in the second case. This is fundamental difference. What calculation is preferable? Basic sets are identical in both calculations.

Some basis sets were designed as spherical while others
as cartesian - typically one needs to use the settings
that are suitable for each particular basis set.

regards,
Alex Granovsky


>-----------------------------------------------

>On Sun Nov 1 '09 10:23pm, Alex Granovsky wrote
>----------------------------------------------
>>Hi,

>>The reason is very simple - cartesian basis contains
>>more functions than the pure spherical one.

>>E.g. any cartesian d-shell has six basis functions.
>>There are five linear combinations of them that
>>indeed transforms according to D irrep, while the last one is of S type
>>(namely, the dx2+dy2+dz2 one). The latter is exactly what is called
>>spherical contaminant and is removed or project out if one deals
>>with pure spherical basis sets.

>>regards,
>>Alex Granovsky

>