alex
yakovenko.alexander@gmail.com
Dear Pedro, and anybody else who have experience in firefly code, would you be so kind to answer a question: is it true to say that small value of MP2 correction energies means (in the most cases) small error in gradients and thus validate poor convergence criteria at the input? If so I can calculate say 1000 of dimmers with light parameters NCONV=3 CUTOFF=1e-7 and save a lot of CPU time then recalculate only ~5-10 (<1%!!!!) with more demanding ones.
Alex
On Sat Jan 2 '10 4:01pm, Pedro Silva wrote
------------------------------------------
>Your input has:
> $MP2 METHOD=1 DM2MET=-2 DIRECT=.T. CUTOFF=1.d-7 $END
>
> $SCF DIRSCF=.T. NCONV=3 DIIS=.T. SOSCF=.F. $END
>You should decrease you cutoff sharply (I suggest 1d-13), as well as the convergence criterion (I use NCONV=7). NCONV=3 gives you very poor gradients, and MP2 compuations need good gradient information.
>Pedro