The erratic SCF convergence behavior is indeed caused by quasi-linear
dependence and by the use of default (i.e. not very precise) inttyp=pople
two-electron integral package for conventional SCF, exactly as
indicated in the output file:
THE OVERLAP MATRIX HAS 5 EIGENVALUES BELOW 1.0E-05.
THE SMALLEST OF THESE IS 4.70910E-07.
THIS INDICATES A PARTIAL LINEAR DEPENDENCE IN YOUR ATOMIC BASIS.
TO OBTAIN SCF CONVERGENCE MAY REQUIRE
MORE ACCURATE INTEGRAL EVALUATION (INTTYP=HONDO, ICUT=11, ITOL=30 IN $CONTRL),
MORE ACCURATE DIRECT SCF FOCK MATRIX FORMATION (FDIFF=.FALSE. IN $SCF), OR
CHANGING CONVERGERS (DIIS=.T. SOSCF=.F. IN $SCF).
On Thu Dec 16 '10 7:30pm, Danas wrote
>Thank you for your answers.
>I attached bellow the log file of the molecule that I mentioned in my first message.
>As you will see in the attachment there is a message concerning linear dependence.
>As for a direct calculation, I think that the calculation time will increase dramatically. I shall try another process without HESS=CALC. It will take, probably, several days on my computer (2.4 GHz Intel dual processor T8300. The firefly implementation on a linux cluster is still pending)
>Please have a look at the log file and give me a hint.