Mosol patch

[James P. Schmidt <jims.,at,.duce.medicine.rochester.edu>]

Again, I have made a faux pas in patch to Mosol as described below by
 Dave Mullally:
 Your suggested change to the SLOW logical is a good idea.
 Unfortunately, it disables the effect of the NOANCI keyword.  If you look
 closly, NOANCI causes the finite differencing derivatives to be done
 from the starting geometry.  At this point, GNORM is 0.0, and GRLIM = 0.0001,
 or GRLIM=0.01.  (Your optimization is done when you reach these
 values of GNORM.)
 I suggest that the change should be to:
       SLOW=.FALSE.
       NOANCI=.FALSE.
       IF(HALFE) THEN
          NOANCI=(INDEX(KEYWRD,'NOANCI').NE.0 .OR. NOPEN.EQ.NORBS)
          SLOW=(NOANCI.AND.(GNORM .LT. GRLIM .OR. SCF1))
       ENDIF
       SLOW = ((INDEX(KEYWRD,'AQUO') +
      1         INDEX(KEYWRD,'ENVAQ') .NE. 0).OR.SLOW)
 This retains the NOANCI functionality.  Oh, one of the effects of the MOPAC
 5.0 keyword 'FULSCF' was to cause derivatives to be done by finite
 differencing.
 You must be getting tired of all of the replies that you have gotten on
 this subject.  I hope that this helps though.
 jims.,at,.duce.medicine.rochester.edu