[molpro-user] The linear dependence error and benzene with aug-cc-pv5z

[Hans-Joachim Werner]

I have not tried this yet, but if it converges it should be ok. Perhaps Kirk Peterson knows more about it.

---
Prof. Dr. Hans-Joachim Werner
Institut für Theoretische Chemie
Universität Stuttgart
Pfaffenwaldring 55
70569 Stuttgart, Germany
e-mail: werner at theochem.uni-stuttgart.de




> Am 16.05.2018 um 16:18 schrieb Bradley Welch <bkwx97 at mst.edu>:
> 
> For the vxz-f12 family would vqz-f12 for the hydrogens and v5z-f12 for the carbons be an acceptable fix? 
> 
> On Wed, May 16, 2018 at 12:37 AM, Hans-Joachim Werner <werner at theochem.uni-stuttgart.de> wrote:
> This problem for benzene is well known and also seen in other programs. Just use the non-augmented basis for the H-atoms and it should work.
> Joachim
> ---
> Prof. Dr. Hans-Joachim Werner
> Institut für Theoretische Chemie
> Universität Stuttgart
> Pfaffenwaldring 55
> 70569 Stuttgart, Germany
> e-mail: werner at theochem.uni-stuttgart.de
> 
> 
> 
> 
> > Am 16.05.2018 um 00:35 schrieb Bradley Welch <bkwx97 at mst.edu>:
> > 
> > Dear all, 
> > 
> > I was and am attempting a CCSD(T)/aug-cc-pv5z computation on benzene and got the linear dependence error  BASIS LINEARLY DEPENDENT OR WRONG S. I immediately lowered the threshold to 1.d-9 and the hartree fock portion completed successfully and in a sane number of iterations. Reading some emails in the mailing list I found that probably isn't the best idea. When the computation got to the CCSD portion it failed with the same error. I saw in Towards highly accurate ab initio thermochemistry of larger systems: Benzene, Harding et.al had success with using only the augmented 5z set on the hydrogens and using the regular 5z basis set on the carbons. Is an approach like this, or using a slightly stripped version (calendar dunning sets) of the aug-cc-pv5z my only options, or is there some way to pull off a full aug-cc-pv5z on benzene? 
> > 
> > 
> > Bradley Welch 
> > _______________________________________________
> > Molpro-user mailing list
> > Molpro-user at molpro.net
> > http://www.molpro.net/mailman/listinfo/molpro-user
> 
>