| 10:05:59 PM PDT - Fri, Sep 11th 2015 |
|
the bug is located
|
It appears that the bug goes from Gaussian's routine for basis orthogonalization. NoBasisTransform option solves the problem, but significantly increases the job time:
http://www.primefan.ru/stuff/chem/ni_nobasistransform.txt
Surprisingly, for gold all seems to be OK. Orthogonalization routine leaves only 168 of 177 basis functions as it does for nickel, but the total energy do not change.
So I think some kind of linear dependency problems do occur in the ANO-RCC basis set.
|