PC GAMESS-related discussion club

Alex Granovsky
gran@classic.chem.msu.su

Yes the PC GAMESS does not support QMTTOL option at present.
There are two reasons why it is not implemented:

1. To avoid sudden changes of the number of basis functions
used in calculations like geometry optimization, etc...
The fully automated approach could skip different number
of linear combinations at different points of PES,
resulting in discontinuous PES.

2. To avoid minor errors in energy gradients and hessians
introduced by the use of QMTTOL.

However, PC GAMESS lists the most linearly dependent basis
functions and allows deleting them manually. We are planning
to automate this to some degree in future releases, i.e., to
introduce user-convenient way of how to delete them from
calculations.

Best regards,
Alex Granovsky  


On Wed Jul 2 '08, Roman Zubatyuk wrote
--------------------------------------
>Dear Andrea,

> As far as I know PC GAMESS doesn't have this functionality. It could not drop some functions in order to reduce linear dependence. So, in the case of diffuse basis sets and/or large molecules you need to increase ICUT in $CONTRL (look at http://classic.chem.msu.su/gran/gamess/inttyp.html). For of MP2 calculations you also need to increase TOL1 and TOL2 in $MP2GRD. You need to be careful while doing MP2 with diffuse basis set: if integral thresholds will not be tight enough you have a chance to get wrong result!

>On Wed Jul 2 '08, andrea wrote
>------------------------------
>>Dear PC-GAMESS users,
>>I would be pleased to known if there is a command to increase the linear dependence threshold to achieve SCF convergence, similar to QMTTOL in GAMESS-US version.
>>Thanks very much.

>> �