| 10:33:20 AM PST - Mon, Jan 21st 2013 |
|
Linear dependencies
|
Hi Volker,
What you are seeing is the effect of removing linear dependencies in the basis set. Only two are found for the aug-cc-pv5z, while 33 are found for the aug-cc-pv6z. We have a pretty aggressive threshold for removing linear dependencies from the basis set of 1.0D-5 in the overlap matrix.
Keywords to adjust the linear dependencies are:
set lindep:tol 1.0d-5 (sets linear dependency cutoff at 1.0d-5)
set lindep:n_dep 0 (number specifies how many linear dependent functions should be removed. By setting this to zero all functions are included)
I would recommend using "set lindep:tol 1.0d-6". This will get you the behavior you want. You could even use 1.0d-7.
While "set lindep:n_dep 0" is an option it will not work here. There are such strong linear dependencies in the basis set that it the DFT energy will simply not converge.
Bert
|