PC GAMESS-related discussion club

Yangsoo Kim
hit8run@gmail.com

It looks to me that the frequency calculation at B3LYP/6-31G* runs with OPTIMIZE & hessend = true can only perform with numerical Hessian, not with analytical Hessian under PCGAMESS. I wonder if this is true. When I use the following keywords,

$CONTRL SCFTYP=RHF DFTTYP=B3LYP RUNTYP=OPTIMIZE COORD=CART $END     
$BASIS  NGAUSS=6 GBASIS=N31 NDFUNC=1 $END
$STATPT hssend=.t. NSTEP=50 $END
$FORCE  method=analytic $END

I got error at the starting point of Hessian calculations. I wonder if this error means the analytical Hessian canĄ¯t be performed or if IĄ¯m using a bad keyword combination.

The reason for considering this issue is that generally numerical Hessian calculation seems more time consuming than analytical Hessian in frequency calculations. Is this true for frequency calculations with PCGAMESS?

Finally, I wonder if there is any other possibility to speed up the frequency calculation with PCGAMESS. I tried to perform runtype=HESSIAN separately with initial Hessian (hess=read) obtained from lower level of calculation (DFT/STO-3G), but it seems that the standalone runtype=HESSIAN calculation does not read the initial Hessian. IĄ¯m not sure if this approach helps, and if it is possible to read initial Hessian to the job, run runtype=HESSIAN.

Any suggestion would be greatly appreciated.

Thanks a lot in advance.

Yang Soo