Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Valentin Monev
vmonev@orgchm.bas.bg

the SCF convergence problem could be treated in the following way:

Use a spherical basis set. The (automatic) removal of spherical contaminants usually solves the quasi-linear dependence problem. You need to use d5=1 in $CONTRL.

State of the art spherical basis sets are Dunning's cc-pVTZ and aug-cc-pVTZ basis sets. They are triple zeta basis sets with polarization functions that were optimized to take into account electronic correlation.

In my experience, using these basis sets in Firefly, SCF and geometry optimizations converge beautifully, in HF, DFT and MP2. There is no need to deal with manual trimming of the basis set to avoid linear dependence problems in SCF convergence, which arise frequently when using diffuse functions (the notorious ++ in the basis set).

For example, to use cc-pvtz basis set, put $basis extfil=1 gbasis=cc-pvtz $end  in the input file. In Firefly commander, use C:\ff\lib\cc-pvtz.lib in Settings, External basis set library file (if the file cc-pvtz.lib is in c:\ff\lib).

Do NOT use aug- part on H atoms ... aug is very diffuse and thus it is appropriate e.g. for LiH but is not very important for H-bonds.

Possible modifications of these basis sets:

A "-" basis set could be obtained by removal of D-shell for hydrogen, plus removal of the F shell for all other elements, from cc-pVTZ (the latter called cc-pVTZ-F or cc-pVTZ(-F) in the literature). Please find attached the correspondingly modified lib file ccpvtz-.lib (which I put in c:\ff\lib). I use $basis extfil=1 gbasis=cc-pvtz $end  in the input file. In Firefly commander, I use C:\ff\lib\ccpvtz-.lib in Settings, External basis set library file.

A "+" basis set could be obtained by using cc-pVTZ on all atoms except N, O, F, Cl, etc...
where aug-cc-pVTZ is used. Please find attached the correspondingly modified lib file ccpvtz+.lib (which I put in c:\ff\lib). I use  $basis extfil=1 gbasis=cc-pvtz $end  in the input file. In Firefly commander, I use C:\ff\lib\ccpvtz+.lib in Settings, External basis set library file.

Probing for tendencies, I use the "-" basis set (about 4 times lighter in MP2 calculations). I use the "+" basis set for final results.

Off-topic: for geometry optimizations, use of delocalized coordinates ($zmat dlc=1 auto=1 $end, nzvar=1 in $contrl) and symmetry (for planar molecules, at least Cs) is recommended.

This message contains the 68 kb attachment [ lib_1.zip ] lib files ccpvtz- ccpvtz+ cc-pvtz acc-pvtz

[ This message was edited on Sun Dec 20 '09 at 2:04pm by the author ]