sorry for stupid question; however, may it be the case that
optimization and hessian runs used different numbering scheme
of atoms than the one used in the IRC run? This is quite a typical
problem when one is using MOs generated for symmetric
configuration as guess for runs without symmetry. The solution
is to reorder atoms in the input file for run without symmetry
so that the order is exactly the same as the one generated by
the symmetry replication code (the exact numbering can be found
in the output file).
One more comment - recent versions of ChemCraft are buggy
and can produce input files with symmetry unique atoms
selected differently as compared with input file they were
taken from. Take care!
On Fri Jun 10 '11 7:09am, Solntsev Pasha wrote
>I don't want start a new post. I steel talk about MCSCF+hessian.
>I did geometry optimization and hessian computation (hssend=.t.). I found i have one image frequency. I checked literature and indeed, such system at HF/DFT/MP2 also produces TS. Then i decided run IRC to understand nature of my transition state. I created new file and changed only runtyp=irc but also added $irc pace=gs2 npoint=3 nextpt=3 saddle=.t. $end . I tool optimized MCSCF orbitals from previous run (last $VEC block) and $HESS block. Also, i used coordinates for equilibrium geometry. But my MCSCF wavefunction has not been converged at even first iteration. It was really very strange, because and MO orbitals and geometry were taken from successfully optimized file. I also tried reoder initial orbitals but that doesn't work for me.