geometry optimization problems with the GAUSSIAN 90

[De windt <DEWINDT%VMSAC { *at * } OHSTVMA.ACS.OHIO-STATE.EDU>]

 I am a "gaussian 90" user, and I have encountered several problems
 when trying to run geometry optimizations on molecular structures
 involving transition metals with other lighter atoms. These difficul-
 ties remained even after beginning with a numerical computation of
 second derivatives in place of the classical unit hessian. Typically
 the optimization goes pretty well at the beginning, until a fatal
 point where the displacements are zero and remain so, though forces
 are far from being negligible (around a factor of 10 above the threshold
 ) restarting the whole thing at the stopping point often provides with a
 new "promenade" on the potential surface, and the same disappointing
 result at the end, sometimes at a different point.
 ...Mayday...
 Thank you very much in advance, and happy new year...
 address: KRYS at FRSAC11.BITNET