CCL:combination of b3lyp and cc-pVQZ

Frank Jensen:
 >  We have recently done some benchmarking of
 > various basis set for DFT. For the H,C,N,O,F molecules
 > in the G3 data set, the MAD and Maximum error of some
 > popular basis sets for total atomization energies
 > with the BLYP functional are (kJ/mol):
 >
 >               MAD	Max
 >
 >  6-31G*	16	64
 >  cc-pVDZ	41	82
 >  cc-pVTZ	3.6	14.5
 >  cc-pVQZ	2.9	10.9
 >
 >  These numbers are errors relative to a saturated basis
 > set, i.e. the Kohn-Sham limit.
 These data are interesting. Why does cc-pVDZ perform so much worse
 than cc-pVTZ? Is this an indication that the valence-triple-zeta
 extension is physically very significant, or does the cc-pVDZ basis
 set suffer from som particular 'unbalance'? Comparison with the
 relatively satisfactory 6-31G* results might suggest that something
 is 'wrong' with cc-pVDZ.
 Any similar benchmarks for Raman intensities?
 Jens >--<
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
 JENS SPANGET-LARSEN         Office:     +45 4674 2710
 Department of Chemistry     Fax:        +45 4674 3011
 Roskilde University (RUC)   Cell-Phone: +45 2320 6246
 P.O.Box 260                 E-Mail: JSL - at - virgil.ruc.dk
 DK-4000 Roskilde, Denmark   http://virgil.ruc.dk/~jsl
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-