Alex Granovsky
gran@classic.chem.msu.su
The related section of Firefly carefully checks input options for
compatibility and adjust them to be compatible if needed. It is very
unlikely that the run will abort due to erroneous input in $NUMGRD.
I'd suggest you to try
$NUMGRD ORDER=4 CONFIT=1 $END
first.
The spline option is seldom used. Basically, spline differentiation
improves stability of differentiation process but decreases
the formal order of obtained finite differences.
Best regards,
Alex
On Sat Oct 29 '16 8:48pm, GrEv wrote
------------------------------------
>Hello,
>I wonder what would be good/optimal settings in $NUMGRD if ORDER=4 is used for searching for conical intersections at the XMCQDPT2 level. I mean:
>$NUMGRD ORDER=4 SPLINE=? CONFIT=?
>Would this be reasonable:
>$NUMGRD ORDER=4 SPLINE=4 CONFIT=1
>which I took from an example in the manual where however ORDER=6 was used. I can't afford ORDER=6 and ORDER=2 doesn't seem to work good.
>Best regards,
>Evgeniy