Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Re^3: SA-CASSCF: different behavior between Firefly 7.1.C and 7.1.G

Alex Granovsky


that is OK. These gradients are obtained using semi-analytical approach described here;
thus the errors in them are somewhat larger than for pure
analytical gradients, and they are more sensitive to round-off
errors and the overall precision of calculations.


On Sat Apr 17 '10 10:40pm, Pasquale Morvillo wrote
>Now it works, but the optimization steps are 18 instead of 14 and the frequencies are slightly different.
>I have attached the two output files obtained with FF 7.1.C and 7.1.G.
>On Sat Apr 17 '10 4:14pm, Alex Granovsky wrote

>>it seems there are some numerical instabilities issues here.

>>Check these settings:

>> $mcscf acurcy=1d-8 ENGTOL=1.0d-13 $end

>>and leave NOZERO=1 - this is a MUST!


>>On Mon Apr 12 '10 12:56pm, Pasquale Morvillo wrote
>>>Dear Alex,
>>>I am studing the job file from the archive found at:


>>>I got a different behavior running this job with Firefly 7.1.C (as the corresponding output found in that archive) and 7.1.G.

>>>1. Using Firefly 7.1.G, the first gradient of the geometry search is to high and the job stop (MAXIMUM GRADIENT 29.22...).
>>>2. Setting NOZERO=0 in $moorth, the job runs but the final energy (after geometry optimization) is higher than the one obtained in the corresponding output found in the archive and there are negative frequencies.
>>>3. Setting ENGTOL=1.0d-11, the final energy (after geometry optimization) is equal to the one obtained in the corresponding output found in the archive; no negative frequencies were found (but they are slightly different from the output of the archive).

>>>Can you comment the above findings?

>>>Many thanks.



[ Previous ] [ Next ] [ Index ]           Fri Apr 30 '10 6:42pm
[ Reply ] [ Edit ] [ Delete ]           This message read 912 times