PC GAMESS/Firefly-related discussion club

Alex Granovsky
gran@classic.chem.msu.su

The IRC code is designed so that it performs one or multiple
steps of the fixed step length.
Indeed, opttol is used by the GS2 code, however, the goal of the
IRC code is not to make gradient lower than opttol but rather to
perform next step along IRC. If gradient happens to fall below
opttol this is on of the possible criteria to stop IRC calculations.
However, this can only happen by chance. The most common criterium
to stop IRC calculation is thus the increase of the energy as
compared with the previosly found point on IRC.

I think the documentation is definitely not quite exact
concerning this point. However, if you set stride to some
very small value, you'll increase the probability to fulfill
the opttol-based criterium while walking along IRC; naturally
this would require much more compitations.

Regards,
Alex

On Fri Nov 14 '08 10:56pm, Maurizio Ciofalo wrote
-------------------------------------------------
>Dear all,
>I noticed that Firefly is not able to keep trace of the OPTTOL keyword ($IRC group), though it is explicitly reported in the official PC GAMESS manual of June, 12, 1999 (the last shared with Firefly, I suppose). In other words, Firefly seems to ignore the value of OPTTOL provided by the user, at least within the GS2 pace which I have tested, as I can see in the output file.
>Therefore it would be virtually impossibile at present to improve the accuracy of the convergence criteria for difficult cases, as Prof. Bernhard Schlegel suggests to do.
>Is there anybody else who faced that problem?
>Thank you very much for your attention.
>Maurizio
>